Progress you can follow.

Review scope first opens the existing source-selection workflow with your current question.

The review shows the exact file, sheet, metric, date boundaries, included/excluded counts and up to ten original input examples.

A local calculation preview and explicit confirmation precede AI submission; changing the question or selection invalidates the review.

The approved scope remains bound to source bytes and enforced during delivery; arbitrary question interpretation is not certified.

Explicit 502, 503 and 504 upload responses allow at most two retries with bounded waiting.

Lost responses and task creation are never blindly retried; persistent preparation failures explain that no task started.

Download again uses the delivered file in this page; the browser still controls saving it to your device.

Independent formula verification, missing saved results and incomplete checks are shown separately.

Explicit cell assignments, numeric zeros and requested live formulas are checked against the saved workbook.

Supported named-column product totals are verified from raw inputs; an AI ratio cannot replace a requested sum of products.

Repair retains the original input scope. Wrong ratios, hardcoded totals and omitted rows remain blocked.

These bounded checks cover supported request patterns, not every natural-language instruction or Excel feature.

Clearly unfinished first messages and creation requests without a purpose ask for clarification before file processing. Short contextual replies remain supported.

Submission locking and keyboard composition handling protect against repeated clicks and premature Enter. Failures restore instructions without overwriting a new draft.

Saved task receipts resume the same task. Expired or unconfirmed tasks offer instruction recovery instead of an automatic restart.

Connection and access errors have localized next steps. Reload recovery requires opt-in device storage; independent regression checks cover interrupted transport and saved sessions.

SUBTOTAL modes 1–11 and 101–111 support bounded A1 and structured table references, hidden rows, saved filters and nested subtotals.

Uploaded XLSX files receive a local edit preflight. Unsupported formulas and calculation errors show their sheet and cell before AI planning.

Uncertain hidden/filter distinctions remain blocked. Editing active filter criteria requires clearing the filter in Excel and uploading again.

Cell preview and approval remain required; version rollback stays available. Compatibility checks cover supported editing, not every Excel feature or business interpretation.

Bounded five-language routing keeps supported questions, how-to requests and negated commands out of controlled editing. Arbitrary natural-language intent is not certified.

Header candidates scan all imported rows within existing limits. Physical header and last-row selection require renewed scope approval before planning.

Excel table totals and recognized footer rows are shown outside suggested ranges. Gaps, repeated blocks and ambiguous total labels require an explicit last row.

Timeouts, interrupted connections, busy services and unsupported edits have localized next steps. Retry is manual and respects the returned waiting period.

Synthetic routing, layout and failure-path checks include late headers and independent saved-file recalculation. Desktop/mobile browser interaction remains unverified.

AI plans bounded edits from the instruction, selected headers and row boundaries; workbook records stay on the device in this workflow.

Cell-level before/after review includes dependent recalculations. Approval is bound to source, plan and output hashes; other parts and chart definitions are preserved.

Version comparisons follow the actual parent after a rollback or branch. Original downloads and the four retained versions remain available; device persistence is opt-in.

Up to 1,000 target cells and 5,000 formulas per edit. Unsupported formulas, complex structures, row/column shifts and table expansion are blocked or require clarification.

16 synthetic regression groups and 28 independent saved-file/native calculation assertions passed. Five-language controls; desktop/mobile browser interaction remains unverified.

Bounded Turkish/English request parsing checks literal cell values, types, coordinates and required live formulas against the saved file. Unrecognized prose is not certified.

Sums of products are verified from aligned raw input pairs, including zero and negative operands, with first/last row, blank and table-growth probes.

Added bounded CHOOSE, IFS, ROWS, COLUMNS and SIGN evaluation plus literal COUNTA arguments. Explicit empty operators over formula-produced empty text remain unverified.

Synthetic regression tests retain rejected originals, diagnostics and corrected outputs. Customer workbooks are not automatically archived for testing.

Round 1 totals distinguish unsold allocation from a verified burn. A successful receipt must contain matching finalization and ARXAI burn events; current supply is read separately.

Five-language round panel with confirmed-block timestamps, transaction links and explicit unavailable/stale states. Desktop/mobile browser interaction remains unverified.

Repeated invoice IDs and mixed currencies in recognized tables are flagged before AI analysis; unresolved business choices require clarification.

Approved file, sheet, period, column and record selections are checked by the server and browser; changed scopes with equal totals cannot authorize delivery.

COUNTBLANK and exact empty-string criteria recalculate supported formula dependencies. Explicit = and <> criteria on formula-produced empty text remain unverified.

Application regression checks run before the production build and stop it on failure. These checks do not create a new live AI accuracy score.

Header detection is bounded and does not certify arbitrary intent. Desktop and mobile browser interaction remains unverified for this release.

Review one original XLSX/CSV sheet, numeric metric, optional category/date/currency fields, period and exact category filter before running the package

Nine-sheet report includes all selected source rows, numeric summaries, groups, months, highest/lowest records, IQR review flags, source hashes and native Excel charts

Blank values stay distinct from zero; exact duplicate rows are retained and flagged. Invalid dates with active period filters, selected mixed currencies and unsupported source formulas block delivery

Original source bytes and reviewed settings are embedded for deterministic replay; every saved workbook part, including chart ranges and caches, must match before delivery

Optional AI explanation must return all 13 metrics exactly; one bounded correction is allowed, and persistent mismatches or changed report hashes withhold delivery

Reusable settings require identical sheet headers and renewed scope approval; controls support English, Turkish, German, Spanish and French

Four live application client/API scenarios passed: quality and outliers, recalculated formulas, empty numeric values and approved month/category filters; all 52 metric claims matched

22 synthetic package groups and 23,124 independent Python comparisons passed, including complete export/replay of 10,000 records; existing settlement, reviewed explanation and connection tests also passed

One original source up to 2 MiB, 10,000 selected rows, 50 columns and 100,000 source cells; output limits also apply. Floating-point statistics use 15 significant digits and are not exact-decimal accounting

The report embeds the entire original source. AI explanation sends both to the external AI service. Desktop/mobile browser interaction and native Office rendering remain unverified

Uploaded XLSX source formulas and filter cells now use the bounded dependency calculator, with stale-cache counts and dependency audits; unsupported formulas and errors withhold source answers

Additional IFNA, error/type predicates and conservative text functions; original COUNT/COUNTA/record semantics stay distinct

A reviewed primary analysis scope pins the original source hash, metric, ranges, exclusions, operation and up to six filters; a correct total from the wrong scope is withheld

Advanced settlement supports up to three exact typed key columns, explicit positive payment/refund/credit records and user-approved dated conversion rates

Per-record integer conversion with half-away-from-zero rounding; credits beyond invoices, refunds beyond selected receipts and unknown rates require review

Original amounts, currencies, rates, transaction types and source addresses remain in the replayable report; net invoices/receipts and gross adjustments are separate

Synthetic boundary checks cover 50,000 rows / 250,000 cells, 49,999 source formulas and 10,000 settlement records; exceeding limits fails without silent truncation

Five-language controls. These checks do not certify arbitrary natural-language intent, unsupported Excel formulas, live FX rates or accounting completeness. Browser/mobile interaction remains unverified

User-reviewed invoice and payment tables from XLSX/CSV sources or separate sheets; up to 10,000 selected records with exact typed, case-sensitive IDs

Multiple unique payment IDs may reference one invoice; duplicate invoices or payments, missing amounts, source formulas, negative amounts and currency differences stay under review with all affected records

Explicit single currency and independent input number formats; integer minor-unit arithmetic with 0–6 decimal places, no source rounding or offsetting open balances against overpayments

Ten-sheet Excel report separates closed, partial, unpaid, overpaid, unmatched and review records; the payment ledger, selected source snapshots, file hashes and original cell references remain available

Exports re-read original files and reopen the saved report; every row, amount, classification and source reference must agree before delivery

Optional AI explanation returns 11 required metrics checked against a replay of the exact report; one bounded repair, with unresolved or mismatched responses withheld

English, Turkish, German, Spanish and French controls; source or setting changes invalidate the previous result

27 new synthetic test groups and 142 independent Python Decimal, SQLite and saved-XLSX comparisons passed

The initial scope does not allocate refunds, convert currencies, evaluate source formulas, infer due dates or certify accounting completeness; desktop/mobile browser interaction remains unverified

Record counts and distinct-customer counts preserve repeated transactions, real zero, leading-zero IDs and exact text matching

Up to six conditions are reapplied to original source rows, including numeric comparisons, status/customer matches and calendar-day boundaries

ISO dates and XLSX 1900/1904 calendars are supported; ambiguous dates and formula-based conditions remain unverified

Source cards show included, filtered, excluded and ignored addresses; incomplete scope inspection is clearly marked

Linked saved output cells must match independently recomputed original-source results before delivery; unsupported or mismatched bindings withhold the file

One unresolved business choice is asked before calculation; bounded repairs cannot discard declared conditions or output bindings

31 new synthetic verification groups and 40 independent Python Decimal/date/scope comparisons passed; the retained browser fixture now passes all three source counts/totals and all three output bindings

Checks apply to declared scopes and supported operations, not arbitrary business rules, joins or untested mobile behavior

Generated workbook previews read the downloadable Blob directly, retaining the existing format and size checks

Local source exports preserve blanks, real zero, identifiers and mixed text instead of coercing missing or nonnumeric cells to zero

Uploaded workbook summaries use the correct overview label; contextual suggestions and quality labels are translated

Live desktop checks exercised upload, source inspection, AI pause/resume, verified file delivery and device-storage restoration using synthetic data

Post-fix desktop preview and download showed independently checked totals 500 / 225 / 4; source values and formulas were preserved

Mobile interaction remains unverified because this test browser does not expose device emulation

Separate numerator and denominator scopes support sums, numeric counts, record counts, equality filters and multiple worksheets

Weighted means include only numeric value/weight pairs, preserving real zero while excluding the weight of a missing value

Blank or explicitly requested zero fallback is checked when the denominator is empty or zero

Bounded table probes append and delete rows and reorder columns; fixed summaries that miss new records are blocked

Automatic repair must preserve the original metric definitions and their zero-denominator policy

Verification cards show checked ratio and weighted-average counts in five languages

22 new metric test groups and 80 independent Python Decimal checks passed; 42 existing table groups and 2,264 retained office-oracle comparisons passed

Live fixed-range output produced correct rate and weighted mean values; automatic metadata verified only the rate. Explicit raw-source replay checked both metrics across 20 probes. Table-based live acceptance remains incomplete

Negative weights, unsupported formula syntax and unchecked business assumptions remain unverified; browser and mobile interaction are not covered by these checks

Interrupted task replies and download streams can retry without starting a second AI job

Server-requested waiting periods are respected; long delays leave the task paused for later continuation

Incomplete task replies cannot overwrite the saved recovery token or display unchecked results

Pause and cancellation remain available during reconnect waits

Expired tasks and invalid workbooks remain blocked; calculation and source checks are unchanged

12 simulated connection scenarios, 13 delivery groups, 13 session groups and 24 workbook-agent checks passed

Provider responses are simulated in these regression checks; browser and mobile interaction remain unverified

Missing-file recovery continues the signed task in its existing workspace; clarification questions and provider refusals do not trigger creation

Recovered files pass the same source, structure and calculation checks; a second missing delivery stops with a clear message in five languages

Explicit workbook generation uses a structured delivery envelope to reduce malformed output metadata

From-scratch generation marks uploaded-source verification as not applicable while preserving all generated-file checks

An explicitly requested JSON block label is restored only for one unambiguous valid payload; all numbers and strings remain byte-for-byte unchanged

Record counts and numeric completeness checks use their actual input columns; unsupported compound filters remain disclosed

Average checks also clear all declared numeric inputs while retaining record IDs, catching empty-scope division failures missed by single-cell probes

Regression coverage adds seven recovery groups and updates historical service tests for the signed source catalogue

Live testing spans nine synthetic domains; initial delivery failures are retained for replay and are not counted as successful results

Browser and mobile interaction, full Excel compatibility and arbitrary business logic are not certified by these checks

Counts, comparison-entry percentages, individual differences and per-comparison sums are recalculated from the exact reviewed workbook before display

A strict model response selects bounded findings; the app supplies the displayed values and labels instead of treating free-form prose as verified evidence

“How was this calculated?” shows the calculation, included entry count, report cells and original source addresses; downloadable JSON retains every reference

Decimal calculations preserve blanks versus real zero, selected numeric formats and leading-zero record IDs; percentages disclose their denominator and six-decimal rounding

Sums stay within one status group and one source pair; duplicate or ambiguous review records are never silently aggregated

The original reviewed file hash is bound to the signed task; changed source bytes and unsupported or incorrect claims block delivery

One bounded repair must preserve all original claim IDs, including when waiting is paused and resumed

English, Turkish, German, Spanish and French finding cards; causal comments and suggestions are explicitly marked as unverified

18 new synthetic verification groups and 57 independent Python Decimal checks across inventory, invoice-payment and personnel-call fixtures passed

Real browser/mobile interaction, actual IndexedDB reloads and an anonymized customer pilot remain pending; these checks do not certify arbitrary AI prose, business causes or tax compliance

Queued, working, checking and reconnecting states show elapsed time and the last server update; delayed work is disclosed without a made-up progress percentage

Transient poll and download failures retry the same task twice; pause and resume retain the existing token within its 20-minute session

Report source addresses are checked against the original uploaded files and exact worksheet names; a repair cannot remove a rejected citation to bypass validation

Historical references copied from an earlier report are retained and labelled separately, without claiming the original sources were reverified

Reviewed comparisons produce a deterministic six-sheet workbook with summary, four record groups and complete source scope; every saved row and value is checked before download

AI explanations read the complete reviewed workbook; four summary totals must match independent checks, and delivery preserves the exact reviewed file bytes

13 new delivery checks, 28 workspace checks, 13 recovery checks, 16 evidence checks and 24 workbook-agent checks passed; an independent Python oracle verified all 55 synthetic source records

Desktop/mobile browser interaction and real IndexedDB reload tests remain unavailable; arbitrary AI-generated reports and qualitative recommendations are not certified by reviewed matching checks

Pending tasks block source replacement, rule changes and version restoration until they are resumed or discarded, including uploads from local mode

Temporary request failures retain the existing task checkpoint; recovery polls the same task without a new upload or task start

Late checkpoints and answers from discarded tasks cannot overwrite a newer workspace task

Device-saving transitions share one ordered write queue; blocked consent storage is reported without claiming saving is enabled

AI file limits are enforced before updating saved workspace sources, including files brought in from local analysis

13 session and recovery regression groups plus 28 existing workspace groups passed; three synthetic business scenarios account for all 55 source records and 63 pairwise source occurrences

Independent Python Decimal and XLSX checks compare every reviewed report row with original source records, including leading-zero IDs, duplicates, blanks, credit amounts and exact tolerance boundaries

Live staff and clarified invoice reports passed independent record-level checks; the original invoice source-label failure and a timed-out three-source report remain recorded

A separate three-source recovery replay used one task start and independently matched totals 66 / 63 / 59; an earlier source-evidence assertion failed, so consistent live reliability is not claimed

Desktop/mobile browser interactions and actual IndexedDB reloads remain unverified because the supported browser test tool was unavailable; synthetic fixtures are not customer pilot data

Up to three complete XLSX/CSV inputs per AI task, 6 MiB each and 12 MiB combined; source checks identify the exact original file as well as the worksheet and cells

Reviewed one-to-one comparisons separate matched, differing, unmatched, duplicate, blank-key and formula-dependent records; no fuzzy joins, silent aggregation or currency conversion

Up to 10,000 selected source records per reviewed comparison, with explicit sheet/header/end-row/key/value choices and downloadable Excel plus source evidence

User-approved business rules accompany file-agent requests; imported rules remain disabled until reviewed

Latest four output files can be selected as the next working copy; cell values, formulas and cell styles can be compared with disclosed visual limits

Optional browser-device storage retains sources, rules, the last answer and recent outputs for recovery; active jobs use their existing signed token within a 20-minute session

Portable import profiles reuse field mappings and settings for business-system Excel/CSV exports; they contain no source rows or credentials and require a new comparison approval

Independent formula checks add bounded rounding, weighted sums, exact lookups and numeric date/time functions, including 1900 and 1904 workbook date systems

Five-language workspace controls; unsupported calculations remain explicitly unverified

28 new workspace regression groups and 127 expanded calculation comparisons/boundary checks passed; existing source, formula and table suites remain green

Two synthetic live AI tasks passed: three-source rule totals with recovery of the same provider job, and a five-sheet reconciliation file independently checked against all 15 source rows; browser reload and visual behavior were not tested

Live ERP connections, tax-law certification, chart/pivot comparison and automatic fuzzy matching are outside this release

Supported Excel structured references now resolve saved table and column names, including current-row calculated columns and explicit data/header/total selectors

Declared raw input checks can follow named table columns instead of fixed row ranges

Temporary scenarios append one or three rows beyond the saved table, remove a seeded first row and swap input columns; first/last, zero, blank and category probes remain active

Incorrect supported calculations enter the existing single repair attempt; the original table and column check scope remains anchored across repair

Explicit automatic-growth requests check for saved Excel table metadata; existing fixed-layout requests retain their scope

Five-language result cards distinguish table formulas, passed growth scenarios and checks that could not be verified

2,264 table-formula results across 21 fixtures match independently constructed A1 references recalculated by LibreOffice; this does not certify native Excel editing behavior

Live sales, inventory, accounting and call-center outputs pass 12 key calculations and 100 temporary scenarios; 60 summary calculations also match raw inputs across 20 separately saved edited copies

Four blocked synthetic attempts remain recorded. Invalid control JSON led to provider-constrained output schemas; the final inventory replay passed without file repair; strict validation and independent calculation gates remain enforced

Checks are bounded to supported formulas and declared raw inputs. Table totals-row relocation, general Excel edit rewriting, chart/pivot expansion and validation-range expansion remain unverified

Supported saved formulas are evaluated without trusting AI claims or cached values; unsupported syntax and exhausted budgets remain visibly unverified

Declared key totals are independently reduced from literal input ranges; this checks arithmetic within the declared scope, not the interpretation of every business rule

Temporary first-row, last-row, zero, blank and category changes exercise saved formula dependencies without changing the downloaded file

Explicit supported worksheet counts, quoted sheet names, required header lists and requested dropdown presence are checked against the saved workbook

Calculation, scenario and structural failures enter the existing single repair attempt; unresolved failures block download

Repair cannot evade a failed check by omitting or changing its original declared target and source scope

Five-language result cards distinguish evaluated formulas, matched calculations, passed scenarios, unavailable checks and stale saved results

48 basic formula results and 139 formulas in a populated 23-sheet workbook match independent LibreOffice recalculation; 3 key reductions and 28 mutation scenarios pass on that workbook

Two real AI runs delivered a new workbook and a preserved 23-sheet revision without repair: 142 saved formulas evaluated, 6 key calculations matched and 41 temporary scenarios passed; downloaded copies received independent file and LibreOffice checks

Advanced formulas, named and structured references, volatile functions, some empty-string criteria and ambiguous request requirements remain outside the supported verification scope

A shared 100-sheet AI limit covers input, generated output and flexible preview; dashboards and helper sheets count toward the total

20 representative worksheets plus dashboard, management and lookup sheets pass deterministic file and preview checks

Native internal links and literal HYPERLINK targets are checked against existing worksheet names and valid cell addresses

External and computed link destinations remain blocked; byte, cell, row and archive limits still apply

Specific output failure reasons and opaque error references survive the single automatic file-repair attempt

Previous-file availability is mentioned only when this conversation actually contains an earlier generated workbook

15 new regression groups cover sheet limits, navigation, blocked destinations, error reporting and chart placement near navigation links

A live replay produced 23 worksheets after one status clarification; 111 independent checks passed, while a visual chart/link overlap informed a new bounded repair check

Structural and label-width heuristics do not certify every formula, business rule or spreadsheet rendering

For explicit full-record helper requests with one unambiguous ID table, original source cells and the complete ordered record-ID sequence are fingerprinted from the upload

Missing, duplicated or reordered helper records and changed source cells enter the bounded repair flow; unresolved faults block download

Supported explicit chart rectangles are checked alongside the print area, including both chart corners

Value-axis titles that incorrectly repeat the category source heading are detected in supported bar charts

Ten new regression groups cover historical failures, valid localized headers, unsupported formulas, source changes and signed server-side enforcement

Six unchanged synthetic sector requests delivered six workbooks, with 30 of 30 summary values independently recalculated correctly; three used automatic file repair

A separate accounting-chart replay passed all five summary calculations and visual review after the axis-title check was strengthened; retained failed outputs remain in the report

These bounded checks do not certify arbitrary Excel formulas, ambiguous layouts, all languages or visual rendering

Source mismatches receive one reinspection; corrected evidence is recalculated against the original uploaded file

Previously failing calculations cannot bypass checks by disappearing from the repaired answer; unresolved mismatches remain blocked

Generated workbooks are screened for direct blank-cell references counted as numeric data, charts outside a single print area and numeric limits on category axes

Detected output faults enter the existing bounded file-repair loop; originals are preserved

Incomplete runs and eligible provider server failures receive one bounded continuation in the same workspace; completed outputs still undergo source and file checks

Instructions now separate record-weighted means from group means, preserve blanks and real zero, and verify saved chart extents and value axes

16 regression groups cover historical failure files, repair paths, bounded continuation and blocked bypasses

Repeated live tests produced workbooks in six sectors; 29 of 30 latest summary values matched independent recalculation, with a retail row-coverage defect and chart-label/layout issues still open

These targeted checks do not certify all formulas, business rules or Excel rendering

Spreadsheet namespace prefixes are normalized without changing uploaded cell values

Worksheet, formula and style inspection supports these valid OOXML variants

Actual cell coordinates enforce preview bounds when dimensions are absent or stale

External relationships, unsafe formulas and size limits remain checked

Cross-sector synthetic evaluation fixtures and repeatable live evaluation harness added for independent inspection of generated files

Shared CSV parsing chooses stable column structure over separator frequency and preserves physical lines in multiline records, BOM and explicit separator files

Empty and cover worksheets remain available without blocking data sheets; the first sheet containing records opens automatically

Unsafe large-number arithmetic stays unverified; tighter comparison prevents decimal-sized errors being accepted in large totals

Analysis answers with missing or invalid evidence receive one bounded repair with a strict response schema, followed by independent source recalculation

Conflicting duplicate join keys require a user decision before grouping; original records must not be silently multiplied

16 added regression groups cover real source files, parser failures, blank versus zero, stale formula caches, evidence repair, ownership and cleanup

Synthetic live-model evaluation complements automated checks; source arithmetic checks do not certify business rules or full Excel recalculation

5% of the original 575M reserve and 2.875% of the initial one-billion supply

The dated reserve record remains 431.25M; 402.5M would remain after the planned transfer with no other movements

Funding wallet, transaction evidence, eligibility, per-person rewards and distribution dates are pending

The funded 28.75M contribution program remains separate; no additional tokens are minted

AI inspects title rows, merged and multi-level headings, independent table regions, repeated headers and subtotal candidates before calculating

One material clarification at a time, with selectable answers and follow-up context for scope, dates, numeric separators, units and missing-value rules

Independent source recalculation for declared sums, averages, minimums, maximums and numeric counts; mismatched answers are withheld

Review source sheets and ranges, exclusions, blank rules, numeric interpretation, recalculated values and a downloadable evidence record with the source file fingerprint

Formula caches, unsupported calculations, business rules and table-selection correctness remain explicitly outside the arithmetic guarantee

Local imports retain colliding column names and trailing unnamed fields, recognize headers after longer prefaces and correctly read grouped decimal-comma values

Source check and clarification controls support English, Turkish, German, Spanish and French

Full CSV source retained for AI processing after upload; local import limits remain visible

Follow-up edits can continue from workbooks created in local mode

AI answers display readable headings, lists and tables without executing HTML

Uncached formula cells show the formula in previews instead of appearing empty; AI file limits are shown separately

Actual AI outputs checked for inventory, leave tracking, student grades, project tracking and monthly budgets

Continued edits preserve original data and charts; a complete 1,005-record CSV produced the verified total of 505,515

Opt-in full-file XLSX and CSV processing with clear disclosure before sending data to an external AI service provider

Python tool execution can inspect uploaded workbooks, calculate from actual cells and create new XLSX copies

Follow-up requests use the latest generated file; earlier downloadable versions can be selected again within the browser session

Independent output checks block malformed packages, formula error cells, macros and external connections; one repair attempt is available for rejected output

Preview reads actual downloaded cells, shared strings, basic formatting and saved formula results

Missing formula caches are disclosed; file-structure checks do not certify business rules or fully recalculate Excel formulas

Background progress, cancellation, bounded task duration and signed file-access tokens

Temporary provider workspace cleanup is requested after each task; uploaded files also have a 24-hour expiry

File processing controls and privacy explanations support English, Turkish, German, Spanish and French

Local mode retains its existing browser-only workbook tools

Preview reads the same XLSX bytes as the download, including worksheet cells, cached formula results, column widths, colors and chart sources

All generated worksheets can be browsed with row pagination, including the usage sheet

Chart previews show their saved placement and up to 30 populated data points; the Excel file retains all cached points

Up to six supported edits can be staged together in AI or local mode, with one approval for the complete revision

Header colors and named column widths can be combined with themes, bounded text-column edits, monthly summaries, averages and chart moves

A missing detail, unsupported step, conflicting setting or invalid dependency leaves the workbook unchanged until resolved

Before-and-after review includes header-color and column-width changes

Preview and revision messages support English, Turkish, German, Spanish and French

Browser rendering is approximate; conditional formatting and table stripe styles are not rendered in the preview

Before-and-after changes are computed from workbook snapshots, including fields, formulas, cell values, row capacity and chart placement

Earlier plans can be restored within the same browser conversation; restoration requires fresh approval and does not replace earlier downloads

Bounded column removal checks dependent formulas, dropdowns and chart sources before changing the plan

Monthly totals and averages use a chosen year, date field and metric, with separate month keys to prevent mixing years

Charts can move to a dedicated or existing worksheet while retaining their source data

Ambiguous student-grade averages ask for the method; shared exam weights must total 100%

Strict row averages remain blank when a required input is missing

AI revisions are checked for unintended changes to existing data, fields and charts

Local structural edits are handled one at a time; unsupported or ambiguous requests remain explicit

Version history is session-based and is not durable cloud storage

Creation runs independently of an uploaded analysis file; new requests open a workbook plan

Explicit row counts, charts, monetary exclusions, named sheets and appearance are checked before download

Worksheet previews show sample values and calculated cells; four appearance choices preserve the plan

Uploaded Excel files can explicitly guide a new AI template through sheet names, column types and calculated-field flags

Reference planning sends structural metadata; source cell values and original formulas stay local. Exact data, formula and style preservation uses the existing workbook editor

Bounded text-column additions and appearance changes preserve existing blueprint data and formulas

Local student-grade planning with shared exam weights complements the existing starting plans

AI plans receive one repair attempt for structural or explicit-requirement mismatches

File output uses content-aware widths and coordinated chart and table themes

AI mode plans arbitrary purposes before considering any local starter

Up to eight planned sheets plus instructions, 24 columns per sheet and 4000 prepared rows overall

Reference dropdowns, conditional formatting, table formulas and column, bar or line charts

Shared arithmetic, ratios, date/time differences, lookups and filtered aggregates

Complete blueprint and up to 30 workbook instructions retained for revisions; changes shown before fresh approval

Local starting plans for inventory, shifts and leave, projects, call logs, budgets and sales

Open questions and unsupported requirements block file creation instead of being silently dropped

Four visual themes with distinct input and formula cells, frozen headers, filters and a usage sheet

English, Turkish, German, Spanish and French; fictional samples or empty input tables

Charts cover prepared source rows; expansion needs a chart-range update. Checks do not certify all business rules or every spreadsheet application

Design requests enter planning directly; currency and sample choices update the same request

Team, row and representative counts are distinct; unsupported revisions are surfaced

Sales, representative roster, targets, KPI dashboard and instructions in one workbook

Six KPI cards, team comparisons and representative metrics use explicit won-sales rules

Excel table references support new table rows; editable representative plans include a dynamic ID list

Revisions require fresh approval; older plans cannot be approved accidentally

Technical XLSX checks and supported-requirement checks are shown separately, without a blanket 100/100 quality claim

Five-language plans and clear empty-template versus fictional-example choices

Formula and package checks do not certify every business requirement or spreadsheet application

Eight reproducible seed pairs compare development and frozen deployment code

Claims and principal exits during normal, exhausted and emergency settlement

Independent balance, reward-cap and surplus-burn checks

Full test output, state digests and scope published in Reports

Local tests; no on-chain changes. Independent audit remains pending

First-use guidance connects upload, question entry, calculation review and saved evidence

Both periods recalculated locally from their imported worksheet rows

Explicit group and measure mapping, units, blank rules and fresh approval

Sum and record-count comparisons for up to 50 groups

New, removed and blank-label groups remain visible

Group differences reconcile with the total; shares may be negative or above 100%

Both-period source cells and review rules downloadable as evidence JSON

Arithmetic contributions do not establish business causes; imported scope and sensitive-field limits remain visible

English, Turkish, German, Spanish and French; original source files unchanged

Application confirmation includes a private tracking link with copy and manual-copy options

Received, under review, approved, rejected and paid states in five languages

Approved ARXAI amounts and verified BscScan payment links

Only a hash of the tracking key is stored; reference numbers alone cannot unlock a record

Application text, email, wallet and internal review notes remain hidden

Existing applications retain the support-email follow-up route; no automatic emails

Public ledger panel distinguishes paid rewards, reserved approvals and available budget

Visible pages refresh every 30 seconds with a last-checked time and an explicit failure state

After reporting an AI answer, write and review a contribution explanation before transfer

Consent required; files, rows, filenames, prompts and answers are not automatically copied

Application opens in a new tab so the current analysis stays open

Drafts expire after 15 minutes; evidence, email and receiving wallet must be entered before submission

English, Turkish, German, Spanish and French; no guaranteed award

Remaining reserve: 431,250,000 ARXAI (43.125% of initial supply), 15 tranches of 28.75M

Contribution rewards: 28,750,000 ARXAI (2.875%); unlocked team-approved payments

On-chain snapshot at BNB Chain block 127300237 confirms both reserve and contribution Safe balances

Contribution Safe uses the same three owners as staking, with a 2-of-3 threshold

Fixed initial supply remains 1 billion ARXAI; no additional minting

Dated balance and governance evidence: /reports/arxai-reserve-allocation-2026-10-13.json

Contribution applications with evidence, contact details and a BNB Chain receiving address

Protected review, reward approval and budget tracking

Team-executed Safe payments without a reward lock-up

Payment records verified against successful on-chain token transfers

Applications open until the budget is exhausted; no guaranteed award

Missing and duplicate column mappings shown before calculation

Sensitive fields and identifier/status numeric mappings blocked

Explicit source currency labels checked against metric units

Invalid mappings disable calculation and are rechecked by the calculation engine

Renamed fields require manual selection; no fuzzy semantic matching

Five-language review messages; source files remain unchanged

Matching names do not certify units or business meaning

Source cell inspector now includes the first 24 columns of its saved row

Original worksheet and cell coordinates retained

Cited cell highlighted among neighboring fields

Sensitive field values masked in the saved row

Answer-time snapshots remain unchanged after source replacement

Inspection controls available in English, Turkish, German, Spanish and French

Saved row examples are separate from the full calculation trace

Worksheet selector with imported row and column counts

Missing-cell and exact duplicate-row counts

Column types, missing values and distinct counts

Date ranges withheld when date formats require review

Status distributions and numeric averages suggested from available non-sensitive fields

Five-language overview controls; original files remain unchanged

Context labels remain tentative; overview covers imported rows only

Editable sum, average, count, minimum and maximum plans

Worksheet, metric, grouping, ordering and up to three reviewed filters

Simple explicit local requests and supported AI query plans routed to approval

Manual plan review available from the data menu

Fresh approval required after source, worksheet, scope or data-definition changes

Included and excluded rows retain original coordinates and reasons

Invalid numeric inputs, missing groups and unmatched filters distinguished

Complete browser-local trace downloadable as JSON; original workbook remains unchanged

All valid groups calculated before limiting displayed groups

English, Turkish, German, Spanish and French review controls

Trace covers imported rows only; other analysis workflows retain their existing behavior

Inspectable answer-time file, worksheet, column, row and cell values

Numeric citations retain explicit column coordinates even when their labels identify a record or group

Sensitive source values remain masked; examples stay bound to the original answer

Best, successful and risky rankings require a named numeric field and higher/lower ordering

Unresolved follow-up questions continue requesting criteria without calling the AI service

Five-language source controls and criterion guidance

Cell examples are not complete calculation inputs; unsupported citations remain disclosed

30-day reward catch-up and 1,024-position dense settlement

256-position proportional exhaustion and emergency settlement

Per-batch gas receipts and transaction counts published in Reports

Principal access, reward liability conservation and repeat withdrawal rejection checked

Local synthetic fixtures; no production transactions or deployment changes

Historical gas samples are not live BNB fee quotes; independent audit remains pending

Missing cells, exact duplicate rows and fully empty columns shown on upload

Ambiguous day/month dates, mixed date formats and unrecognized dates flagged for review

Non-numeric values in mostly numeric fields and duration fields without explicit units identified

Multiple plausible header contexts shown without assuming a single business domain

Empty columns remain categorical rather than inferred dates or measures

English, Turkish, German, Spanish and French review text

Signals apply to imported rows; no automatic cell corrections or raw-value upload

Real Safe 1.4.1 and SafeL2 1.5.0 runtime exercised with EIP-712 signatures

Single, duplicate and outsider signatures rejected; two-owner execution and replay protection checked

Active 90-day and flexible positions tested during permanent emergency closure

Immediate principal exits, earned cutoff rewards, bounded settlement and surplus-only burn verified

Delayed withdrawals do not extend accrual or pay principal twice

Local runtime fixtures do not prove current live Safe configuration or factory deployment

No production contract changes; independent audit remains pending

30/60/90-day withdrawals tested exactly one second before and at maturity

Unauthorized and repeated flexible claims rejected

Exhausted allocations compared with checkpoint batches of 1 and 64

New scenarios run against development and frozen mainnet compiler-input artifacts

No deployed contract, rate or reward budget changed

Internal tests are not an independent audit; audit remains pending

Per-answer source file, worksheet scope and imported row count captured when answering

Exact source-cell coordinates where the cited field and row can be resolved

Sensitive source fields masked in evidence examples

Calculation inputs and filters displayed when a calculation plan is available

Examples explicitly distinguished from complete calculation inputs

Missing exact cell evidence disclosed for aggregate-only AI answers

Duration units and ambiguous day/month dates require explicit review before supported calculations

Existing format and data-definition tools opened without automatic approval

Source evidence remains tied to its original answer rather than replacement worksheets

English, Turkish, German, Spanish and French evidence controls

Downloadable, versioned analysis plans with up to eight count, sum or average KPIs

Explicit field mapping, numeric units, decimal formats and blank-value rules

Exact-value filters and reviewed total-row exclusions

Optional aggregate baseline for the next period; no raw rows stored in plan files

Fresh approval required after source, mapping or calculation changes

Previous imported results identified separately from newly calculated results

Visible imported-row limits and source fingerprints

Absolute and percentage changes with undefined zero-baseline percentages

Verified Excel comparison reports and complete source-cell evidence JSON

User-downloaded plans; original source files remain unchanged

Changing analysis controls clears previous conversion previews, approvals and results

Mode and cross-tab calculations disable duration normalization immediately

Reconciliation results from replaced sources are hidden and pending calculations invalidated

19 advanced-analysis checks passed; calculation formulas and source data remain unchanged

Flexible 2% simple APR; 30 / 60 / 90-day terms at 5% / 8% / 12% simple APR

Immutable rates and 115 million ARXAI reward cap; no automatic compounding

Flexible rewards calculated for complete 24-hour periods from each deposit

Voluntary locked early exit returns principal and cancels the entire position reward

Principal withdrawal and allocated flexible claims remain independent of batch settlement

Permanent emergency closure controlled by a 2-of-3 Safe; earned liabilities preserved and only unallocated rewards burned after settlement

At activation on 7 October, 115M of the original 575M reserve funded staking; the reserve was 460M then. See the 13 October contribution allocation update for the current 431.25M reserve.

Activation transaction: 0xa706e9a69422ec3ddb4fd449952f457c39bd4a334f7f12565470a5fe2fd82740

Not-yet-due flexible positions excluded from normal reward calculations

Proportional allocation retained when reward capacity may be insufficient

Original 24-hour anchors, fixed APRs and the planned 115 million ARXAI cap retained

Principal exits and allocated reward claims remain independent of batch settlement

Local sparse, dense, reserve-exhaustion and emergency tests cover more than 1,000 positions

109 local contract tests passed; independent security review remains pending

Recorded staggered-deadline catch-up uses less gas; dense queue maintenance can increase total gas

Emergency closure still requires bounded settlement before surplus rewards are burned

Principal-only exit available during either checkpoint pass and emergency settlement

Already allocated flexible rewards claimable without global catch-up

Flexible earning stops at the recorded exit time; complete days remain eligible for settlement

Voluntary locked early exits still forfeit rewards, including before a later emergency closure

Final withdrawal cannot pay principal twice

Immutable 2%, 5%, 8% and 12% simple APRs and planned 115 million ARXAI reward cap retained

103 local contract tests passed; no independent audit or public deployment implied

Global reward allocation still requires bounded scans and further scalability review

Source matrices retained within the existing import limits

Suggested vertically separated and blank-column-separated table areas with manual range controls

Suspected total rows shown for explicit exclusion; original cells unchanged

Exact-value and confirmed date-range filters with visible included and excluded records

Median, tied modes, type-7 percentiles and bounded category cross-tabs

Approved ISO, day-first, month-first and Excel 1900 date interpretation

Approved numeric, clock and labeled duration interpretation into seconds

Invalid values require explicit exclusion; source row and column offsets preserved

List reconciliation identifies missing records, unequal values, blank keys and ambiguous duplicate keys

Exact matching avoids automatic financial netting or currency conversions

Downloadable local analysis and reconciliation evidence reports

New controls translated into TR/DE/ES/FR

Browser-session worksheet definitions with explicit confirmation

Definitions cleared on a new upload or worksheet change

Separate row counts and exact unique-value counts; blank keys reported

User-defined status-code labels with unknown codes kept visible

Numeric unit labels and explicit exclude, zero or clarify blank policies

Confirmed counts, code distributions, sums and averages calculated locally

Original source-row evidence and visible calculation assumptions

Ambiguous filters, incompatible values and currency conflicts require clarification

Sensitive identifiers excluded from mapped distributions and arithmetic

Earlier worksheet context excluded from follow-up history after switching sources

Five-language definition controls and starter calculations

No source-cell changes, automatic unit conversion or raw definition upload

Tentative header-based context hints for call records, accounting, personnel, inventory and surveys

Mixed or unknown structures remain general tables; domain meaning is not assumed

Identifier, numeric status, duration, date, measure and category roles

File-specific structural, category-distribution and data-quality questions

Browser-local category counts with missing-value totals and original source-row references

Ambiguous performance questions ask for a metric instead of guessing

Numeric identifiers and status codes excluded from automatic measures

Non-financial numeric fields receive generic descriptive averages

Categorical-only tables remain available for count and quality analysis

Workbook structural checks keep worksheet results separate; unavailable or ambiguous scope requires clarification

TR/DE/ES/FR starter questions and context labels

AI receives only protected field roles and tentative context, not raw cell values

Existing correction approval and multi-file relationship review remain unchanged

Flexible 2%, 30-day 5%, 60-day 8% and 90-day 12% simple APR; no automatic compounding

Independent positions and principal-only voluntary early exits from locked positions

Flexible rewards settle after complete 24-hour periods

Locked rewards reserved upfront; bounded proportional flexible settlement

Permanent emergency closure controlled by a verified 2-of-3 multisignature wallet

Unused emergency rewards burned while principal and earned reward liabilities remain protected

Public testnet deployment and independent security review remain pending

Application controls, help and common error messages follow the selected application language

Original file names, source column identifiers and cell values remain unchanged

AI explanation and workbook-planning requests carry the selected application language

Unicode-capable PDF reports and localized Excel summary headings

Localized correction manifests and multi-file report metadata

Verified calculations and source-cell evidence remain unchanged

Trust Wallet EVM namespace compatibility with callable-provider checks

Wallet announcements remain active after first opening the connection dialog

Fresh discovery requested before selecting a wallet

Missing extensions show recovery instructions instead of automatically opening a download page

Desktop extension and unavailable phone QR connection clearly distinguished

Existing BNB Smart Chain checks and purchase approvals retained

Two to five browser-local files with worksheet selection

One to four many-to-one relationships with explicit preview and approval

Duplicate lookup keys block calculations to prevent record multiplication

Unmatched and blank keys remain visible; every main record is retained

Grouped sum, weighted overall average and main-record count

Measures come from the main table to prevent repeated lookup totals

Ambiguous numeric text is excluded and counted

Approval is bound to a SHA-256 source and relationship fingerprint

File, worksheet and cell evidence for every contributing record

Complete downloadable JSON evidence report

Worker-backed import and calculations with cancellation

8 MB per file; 20 sheets, 1,000 records and 100 fields per sheet

Original files remain unchanged; raw rows stay in the browser

Separate browser-local correction preview and approval flow

Leading and trailing whitespace cleanup

Unambiguous decimal text conversion in supported business measures

Exact duplicate-record removal without shifting worksheet coordinates

Individual change selection with visible before-and-after values

Projected imported-row quality score and record counts

Missing values, ambiguous numbers, dates, outliers and sensitive fields remain for manual review

Workbooks containing formulas, macros, external links or pivot dependencies are blocked in this initial correction release

Protected, merged and table worksheets are excluded

Source fingerprint rechecked before applying approved changes

Original file never overwritten; untouched package content preserved

Corrected Excel copy, change manifest and integrity report

Re-upload the corrected copy to recalculate analysis

Raw workbook rows remain browser-local

Interactive metric and percentage scenario controls

Observed values remain visibly separate from hypothetical results

Baseline, projected value and impact shown together

Exact worksheet cell range retained for every scenario

Human-readable formula, observation count and verification fingerprint

Downside and upside scenarios calculated deterministically in the browser

Dynamic decision plan and prioritized executive actions remain connected to verified evidence

Original workbook is never changed by scenario testing

Raw workbook rows remain browser-local

Coinbase Wallet DApp URL carries both a presale query signal and the presale anchor

Presale section is restored when a wallet browser removes the URL fragment

MetaMask and Trust Wallet continue to receive the same official presale destination

No wallet credentials, balances, seed phrases or private keys are requested

Missing-cell detection with worksheet and source-row references

Exact duplicate-row detection

Mostly numeric columns checked for incompatible values

IQR-based statistical outlier review

Explainable confidence and recommendation for every finding

Per-finding Review or Dismiss decision

Workbook data-quality score and issue counts

Review decisions added to the downloadable manifest

No data value is automatically corrected in this controlled release

Raw workbook rows remain browser-local

Live before-and-after preview using privacy-safe workbook columns

Original values remain visible and unchanged during preview

Controlled change log with operation type, reason and scope

Individual cleanup, formula, structure and visual operations can be removed before approval

Removed operations do not mutate the original plan or source workbook

Approved operations are recorded in the downloadable change manifest

Existing formula, macro, external-link and scope integrity gates remain active

Original workbook is never overwritten

Raw workbook rows remain browser-local

Preflight scan for formulas, macros and external workbook links

Explicit acknowledgement before revising workbooks with protected assets

Macros preserved without execution

Existing external links preserved exactly and new links rejected

Formula additions require separate approval

Original workbook is never overwritten

Revision blocked when formulas, sheets, protected assets or approved scope change unexpectedly

Downloadable change manifest and post-build integrity report

Optional KPI, executive summary, pivot-style summary, comparison and dashboard outputs

Raw workbook rows remain browser-local

Guided metric and period clarification before calculation

Dynamic KPI, driver, comparison, forecast and anomaly answer layouts

Executive and Analyst presentation levels

Context-specific follow-up actions

One-click verified PDF and Excel output

Percentage-scale and duplicate-identifier consistency checks

Session-scoped metric, period, currency and filter memory

Explain this finding in plain language action

Memory resets when a new file is loaded

Raw workbook rows remain browser-local

Decision-ready summary limited to three short sentences

No more than three key findings shown at once

One recommended next step separated from optional follow-up questions

Full explanation preserved behind progressive disclosure

Calculation method, scope, confidence and source rows remain collapsible

AI response contract now prioritizes conclusions and avoids repeating evidence

Raw workbook rows remain browser-local

Semantic roles for measures, rates, targets, actuals, identifiers, dates and dimensions

Safe default aggregation rules such as sum for measures and average for rates

Currency, percentage, count, date and text unit inference

Mixed 0–1 and 0–100 percentage-scale warning

Repeated-identifier warning for potential join multiplication

Unknown-unit review warning before business totals are trusted

Workbook readiness state: ready, review or blocked

Readiness score with explicit blockers, warnings and assumptions

Compact semantic field map in Analysis details

Raw workbook rows remain browser-local

Progressive upload summary limited to three essential facts and three focused next steps

Deeper KPIs, anomalies, forecasts and technical evidence remain available on demand

Verified what-if calculations for percentage increase and decrease questions

Scenario results separated from observed data and forecasts

Automatic actual-versus-target and actual-versus-budget variance analysis

Decision confidence decomposed into data quality, sample adequacy, calculation verification, relationship strength and forecast fit

Dynamic multi-step decision plan adapted to available workbook evidence

Prioritized executive actions for target gaps, anomalies, data-quality risks and positive drivers

Assumptions, limitations and calculation inputs shown explicitly

Every new decision calculation retains a reproducible local verification fingerprint

161 application verification checks passed

Raw workbook rows remain browser-local

Safe cross-sheet joins through verified shared keys

Minimum key-overlap threshold before a join is allowed

Ambiguous dimension keys rejected

Grouped sums, averages and counts across related sheets

Multiple business dimensions considered during contribution analysis

Visible calculation operation, inputs, filters and safeguards

Source-sheet and source-row references retained

Sensitive join keys and fields excluded

AI interpretation separated from browser-local calculation

Raw workbook rows remain browser-local

155 application verification checks passed

Category-based sum, average, count, minimum and maximum calculations

Highest, lowest, top and bottom category ranking

Optional result limits for focused answers

AI selects only the calculation plan and approved column names

All grouped values recalculated from browser-local rows

Source-row references retained for every returned group

Sensitive columns rejected before grouped analysis

Invented filter values rejected before calculation

English and Turkish ranking questions supported

154 application verification checks passed

Raw workbook rows remain browser-local

Conversation promoted to the primary workspace

Focused dark navigation rail and calm analysis canvas

Session-only recent chat list

New analysis flow without losing recent in-session conversations

Fixed responsive composer with upload, scope and analysis-mode controls

Analysis details remain available without interrupting the conversation

Mobile full-screen chat flow with slide-in data navigation

Chat history clears when a new file is loaded to prevent cross-dataset confusion

No browser-persistent chat storage added

Existing verified KPI, forecast, workbook creation and redesign features retained

Privacy-minimized AI summary upgraded to v2

Locally verified KPI values available to AI explanations

Period-change context supplied without category values or raw rows

Forecast estimates include observation count, R² and uncertainty ranges

Forecasts must remain clearly labeled as estimates

Mathematical contribution is never presented as proof of causation

AI evidence values cross-checked against the expanded local summary

Sensitive columns removed before Analysis Engine calculations enter AI context

File names, sheet names and source rows remain browser-local

Automatic browser-local fallback retained

CoinGecko Preview Only profile published

Public profile: coingecko.com/tr/coins/arexai

ARXAI contract shown as 0xeaA12b3bE7CDEC7749b972eD5C342f4e933CcBb3

BNB Smart Chain network identity supplied

Maximum supply supplied as 1,000,000,000 ARXAI

Project website, GitHub and whitepaper information supplied

PancakeSwap public trading remains scheduled for 27 November 2026 at 18:00 UTC

No claim of guaranteed listing price, liquidity or return

Column-level type, completeness, uniqueness and invalid-value profiling

Duplicate-row and IQR outlier detection

Automatic KPI discovery for revenue, cost, profit, orders, rates, actuals and targets

Derived average order value, gross margin and target-attainment calculations when matching fields exist

Period-over-period contribution analysis by a detected business dimension

Safe cross-sheet relationship mapping without exposing join-key values

Robust contextual anomaly detection with source-row evidence

Forecast estimates separated from actuals with residual uncertainty ranges

Clear separation between mathematical contribution and causal claims

Deterministic local verification records and calculation fingerprints

Multi-sheet engine result with workbook-wide quality totals

Analysis details panel for data readiness, KPIs, relationships, anomalies, forecasts and change drivers

Approved ArexAi Executive Report sheets now include verified KPIs, change drivers, anomalies and forecast ranges

152 application verification checks passed

Raw workbook rows remain browser-local

Trust Wallet mobile DApp routing aligned with the current official deep-link format

MetaMask mobile routing safely preserves the official presale destination

Legacy injected-provider discovery added for Trust Wallet, Binance Wallet and Coinbase Wallet

Duplicate providers removed when several wallet extensions are installed

MetaMask, Trust Wallet and Coinbase Wallet deep-link regression checks added

BNB Smart Chain network verification retained before purchase

No seed phrase, private key or wallet credential is requested or stored

TypeScript, lint and production-build gates passed

Scheduled start: 27 November 2026 at 18:00 UTC

Türkiye time: 21:00 TRT

Venue: PancakeSwap

Network: BNB Smart Chain

Planned primary pair: ARXAI/USDT

Official token contract remains 0xeaA12b3bE7CDEC7749b972eD5C342f4e933CcBb3

Pair address will be published only after the official pool is created

No claim of guaranteed price, liquidity or return

52 of 52 automated contract tests passed

ARXAI Token, PresaleRound and TeamVesting covered

Listing Reserve and Liquidity Reserve added to the active test scope

Invalid deployment configurations and zero-address dependencies tested

Allowance, balance, allocation and rollback failure paths tested

START and END sale boundaries tested

Ownership transfer and renunciation consequences documented

Fixed-supply and token-burning permissions verified

Team vesting funding, tranche boundaries and cumulative release behavior verified

Earlier 37-test scope recorded 100% line and statement coverage

Independent professional audit remains pending

Six fixture-backed question types checked against known workbook totals

Exact column names take priority over shared words such as amount or target

Regional and monthly sums calculated locally

Ambiguous metrics or workbook sheets trigger clarification

Source references retain actual Excel row numbers when header rows or blanks precede the data

Workbook rows remain in the browser

Workbook-wide analysis uses local evidence from the available sheets

Highest and lowest value questions retain source-row references

AI explanations and evidence values are both checked against local summary calculations

Analytical AI replies require reproducible evidence

Browser-local answers are excluded from later AI conversation-history requests

No raw workbook rows are sent to the AI analysis endpoint

Removed the temporary 20,714-day countdown flash

Presale clock remains disabled until the current time is synchronized

Round totals show a synchronization state before contract reads complete

Purchase submission stays disabled during initial clock synchronization

Allocation cards distinguish synchronization from unavailable data

Live reads continue to recover the confirmed 10,000 ARXAI Round 1 purchase

Four-provider BSC RPC fallback and last-verified cache retained

Removed the connected-wallet balance panel

Removed the shortened wallet address from the purchase card

Stopped periodic balance reads for the connected account

Retained only the network identifier required for safe BNB Smart Chain purchases

Preserved wallet connection, network switching and purchase behavior

Reduced the mobile purchase panel height

Honeypot.is public result linked from the security page

Low honeypot risk and passed simulation status disclosed

0% buy, sell and transfer tax results disclosed

Simulated-liquidity limitation clearly stated until a live DEX pair exists

GoPlus tax, sellability and gas-abuse findings published

GoPlus source-verification index mismatch disclosed

Manual GoPlus metadata review marked as pending

Slither 102-detector result linked as a dated technical record

Professional independent audit status remains pending

Automated checks explicitly separated from an audit certificate

Exact USDT allowance shown before purchase

Expected ARXAI delivery shown beside the allowance

Round-specific presale contract displayed with BscScan verification

Configured payment treasury displayed with BscScan verification

Explicit confirmation that unlimited approval is never requested

Official domain-to-contract proof under the arexaidata.com domain

Successful 100 USDT to 10,000 ARXAI mainnet purchase evidence

Post-settlement USDT allowance verified as zero

Blockaid review dossier published as plain text

Natural-language workbook revision requests from chat

Workbook-aware optional KPI, summary, pivot-style, dashboard and comparison choices

Explicit before-and-after revision preview

Separate approval for calculated KPI formulas

Final Approve & create copy gate

Original workbook remains unchanged

Selected-sheet or eligible-workbook revision scope

Unselected worksheets verified byte-for-byte

Original formula preservation checks

Macro, media and embedded-asset preservation checks

External relationship comparison

Approved ArexAi Report sheet verification

Eight-point revision integrity report

Automatic download blocking on integrity failure

Browser-local workbook transformation

30 workbook regression fixtures across 10 business use cases

Real Office package extraction during quality tests

Required XLSX part verification

Worksheet name and count checks

Approved row and column dimension matching

Formula range verification

Dropdown validation count checks

Frozen-header and filter checks

Macro and VBA package blocking

External-link and embedded-object blocking

Network-formula safety scan

10 MB uncompressed package safety limit

Visible workbook verification score

Download automatically blocked on validation failure

Generated files remain browser-local

Free-form AI workbook planning

Strict structured-plan validation

Workbook plan preview before file creation

Choice buttons for missing requirements

Explicit Approve & create gate

Dynamic columns and data types

Fictional sample-data safeguards

Choice-field dropdown validation

Currency-aware number formatting

Optional dashboard and executive-summary sheet

Local summary formulas generated after approval

CRM, sales, inventory and name-list local planning fallback

Direct downloadable XLSX output

Raw uploaded workbook rows remain outside AI planning requests

New Excel workbooks created directly from chat

Multi-turn request context retained for file name, row count and table type

Direct XLSX download card inside the conversation

Professional title and table header styling

Alternating row treatment and readable automatic column widths

Frozen title and header rows

Built-in filters and print-ready landscape layout

Browser-local generation for supported requests

No manual copy-and-paste workflow

Regression tests for Turkish creation requests and XLSX package output

Workbook-aware recommendation cards

User-selectable KPI, summary, pivot-style, dashboard and comparison options

Calculated profit, margin, variance or attainment formulas when matching columns exist

New ArexAi executive report worksheet

Category-based pivot-style summary table

Professional column, bar or line chart selection

Monthly and yearly comparison blocks

Detailed before-and-after preview

Separate approval for new KPI formulas

No structural change without final approval

Local in-browser workbook generation

Original workbook remains unchanged

Removed the public external-account sign-in control

Removed saved-workspace controls from the application header

Saved-workspace routes safely return to the application

Workspace write APIs paused

Existing storage foundation retained but not publicly exposed

No user data removed

ArexAi-branded email access moved back to planned status

Sign in with an external account

User-owned private workspaces

Explicit Save workspace control

Saved aggregate analysis summaries

Saved conversation history

Dedicated workspace library

Protected saved-report view

Account-scoped access checks

Workspace deletion control

Original Excel and CSV files are not stored

Raw spreadsheet rows are not stored

Expanded Excel and CSV question regression suite

Hallucination and irrelevant-answer probes

AI evidence cross-checked against local calculations

Stronger header, delimiter and date detection

Automatic clarification for ambiguous multi-metric questions

Incorrect-answer reporting without storing workbook content

Per-response latency and token tracking

Optional cost estimates from configured model rates

Persistent aggregate AI quality telemetry

Early Access Beta product status

Live Excel and CSV analysis capabilities

Installable application experience

Unified ArexAi AI product identity

Public capability history retained

Known product limitations remain clearly documented

Workbook commands routed directly from chat

Non-destructive empty-row cleanup

Data-bar visuals for numeric columns

Professional report and print layout

Preview before every transformation

Explicit approval required before file creation

Local in-browser workbook processing

Original workbook never overwritten

Formulas and macros preserved

Separate downloadable revised copy

Up to six recent conversation turns for follow-up context

Structured question-intent routing

Conversation, explanation, analysis, comparison, anomaly, visualization, redesign and upload-help intents

Evidence summary cards for AI analysis

Evidence restricted to the privacy-minimized dataset summary

No chat history treated as dataset evidence

Clear upload guidance when no workbook is loaded

Strict response validation retained

No preloaded sample workbook

Upload-first empty state

ArexAi logo in the AI mode selector

Natural AI conversation without a workbook

Clear file-required responses for dataset questions

No invented workbook context

Analysis controls remain unavailable until real data is loaded

Existing privacy-minimized summary boundary retained

Installable Progressive Web App

Direct launch into the Data Analyst workspace

Android install prompt

iPhone and iPad Add to Home Screen guidance

Standalone app presentation

Safe-area support for modern mobile screens

Offline fallback screen

AI API responses excluded from offline caching

Existing local privacy controls retained

No token-purchase flow inside the app experience

Free-form visual design briefs

AI-generated custom palettes and typography

Professional workbook preview before application

Active-sheet or eligible-workbook scope

Local in-browser Excel transformation

Formulas, charts, comments and macros preserved

Original workbook never overwritten

Downloadable ArexAi-designed Excel copy

Large-workbook background processing

Private local design fallback

Background workbook processing

Visible import progress and cancel control

Early 20-sheet import boundary

1,000-row and 100-field limits applied before interface transfer

Formula inspection capped for predictable performance

Full formula count retained

Reduced browser memory transfer

Large real-world workbook regression validation

Production AI model connection enabled

Natural-language questions and answers

Context-aware follow-up conversations

Explicit user opt-in before AI analysis

Privacy-minimized analytical summaries only

No raw rows, filenames or sheet names sent to the model

Structured response validation and request throttling

Automatic browser-local fallback if AI is unavailable

Server-only AI service boundary

Strict summary-only request validation

Raw-row and unknown-field rejection

32 KB minimized request limit

Per-client request throttling

Structured model responses

Explicit AI consent before activation

Automatic local fallback

Live AI availability detection

65 automated application verification checks

Follow-up questions retain the previous metric

‘Why?’, ‘Explain that’ and ‘Compare it’ supported

Ambiguous requests trigger a clarifying question

Local mode remains active by default

AI Analyst preparation panel explains the privacy boundary

Responsive mode controls for desktop and mobile

60 automated application verification checks

No raw rows in the AI summary format

Sensitive columns excluded completely

File and sheet names excluded

Row labels and source citations excluded

Only schema, aggregate quality, trends and correlations retained

Automated leakage regression checks

55 automated application verification checks

Excel definition questions recognized

CSV definition questions recognized

Natural variants such as ‘What mean Excel?’ supported

English and Turkish explanations

Relevant upload action and follow-up suggestions

Definition questions excluded from metric analysis

Regression coverage added

53 automated application verification checks

Natural greetings and combined greeting questions

Identity, capability and help responses

Acknowledgements, thanks, apologies and farewells

Short everyday conversation support

Clear boundary for general-knowledge questions

Data questions continue to use evidence-backed analysis

Updated conversational interface guidance

51 automated application verification checks

CSV, XLSX and XLSM validation

8 MB upload limit

20-sheet workbook limit

1,000 rows and 100 fields per sheet

Sensitive-field detection and default masking

Clear browser-local processing disclosure

Data-readiness label separated from AI certainty

38 automated application verification checks

Chat-first analysis experience

Single English interface

Compact desktop and mobile data menu

Persistent question composer

Plain-language answers in the main conversation

Expandable method and evidence inside answers

On-demand analysis-details drawer

Excel, CSV and multi-sheet workbook support retained

Result-first executive summary

Strongest movement, risk and recommended-action cards

Summary, charts, signals, forecast and source-data tabs

Collapsible filters and analysis-method details

Dedicated fixed-height Analysis Copilot panel

Evidence and methods collapsed inside conversational answers

Compact dataset controls with workbook sheet switching

Responsive single-column flow on smaller screens

Active-sheet and whole-workbook analysis scopes

Named-sheet questions and cross-sheet trend comparison

Relationship answers grounded in shared columns and key overlap

Method, evidence scope and confidence shown on every answer

Fixed-height scrollable conversation with automatic latest-answer focus

Chart intent switches bar, line, scatter and heat-map views

Sheet-aware evidence links that open the cited source

Automatic sensitive-column detection

Sensitive source values masked by default

No workbook upload or external model call

Four-provider BSC RPC fallback

Independent contract-read recovery

Last-verified snapshot retained locally

Live, partial, cached and connecting states

Manual synchronization retry

Unavailable values separated from verified zero balances

All-sheet Excel workbook import

Sheet switching with independent analysis

Shared-column and key-overlap relationship detection

SUM, AVERAGE, MIN, MAX, COUNT and arithmetic formula recalculation

Cached-result fallback for advanced Excel formulas

Embedded chart and image inventory with image previews

Style, merged-cell and conditional-format inspection

Safe VBA macro detection without execution

Conversational questions over the filtered dataset

Turkish and English question patterns

Clickable source-row references

Contextual follow-up question suggestions

Automatic metric focus in the visual workspace

Explicit local-engine and generative-model boundary

Excel and CSV import in the browser

Date, category and numeric filters

Data-quality checks and automatic field typing

Bar, line, scatter and correlation heat-map views

Three-period linear projection and scenario controls

Downloadable PDF analysis report

Growth, sales, finance and customer sample datasets

Pearson correlation ranking

Data completeness and confidence indicators

Bar and line chart switching

Turkish and English interface

Source-row citations for findings

CSV parsing and numeric-field discovery

Trend and change calculations

Z-score anomaly screening

Metric trajectory visualization

Plain-text findings export

Local processing with no file persistence