See what a change will do before you make it.
Tavuni turns the spreadsheets and rules your business runs on into a working application. It shows how your information connects, what a change would affect, and checks every number against the source you already trust.
No account needed. It runs in your browser, on real numbers from a real workbook.
Every number is checked against your workbook.
5/5 Decide with confidence.
What if only orders over $25,000 needed a manager?
All of these depend on that one number.
One value to look at. Everything else agrees.
Clear. I can approve this.
Orders over this amount need a manager’s approval$10,000$25,000
| Order | Customer | Total | Approval |
|---|---|---|---|
| SO-5000 | Canada Enterprise Co 1 | $63,054.97 | Manager |
| SO-5001 | UK Government Co 8 | $46,732.57 | Manager |
| SO-5002 | Canada SMB Co 15 | $36,612.93 | Manager |
| SO-5003 | UK Mid-Market Co 22 | $30,202.89 | Manager |
- 120 orderseach total is compared with it
- The approval queue89 orders waiting for a manager today
- Every order totalbuilt from price, discount, tax and freight
21 orders would no longer wait for approval
- SO-5087$24,407Auto
- SO-5062$23,734Auto
- SO-5048$22,472Auto
- + 18 more
2,869 of 2,870 values agree with your workbook
- 1 rounding edge, shown for a person to look at
- 0 different
Proposed: orders over $25,000 need a manager
- 21 fewer orders waiting for approval
- $3,209,673 of orders would still get a manager’s review
- Numbers checked against the workbook
A draft, waiting for a person. Nothing changes until someone publishes it.
Every number on this page was computed by Tavuni from one real operations workbook.
Look inside. The real application, step by step.
For the technically curious: the same orders in the live application: the rule, the evidence for every step of an order total, and the check on every value. It plays by itself; choose an order to take over. Then nine more things the platform does.
Orders over $10,000 need a manager’s approval
89 of 120 orders
| Order | Customer | Total | Approval | Checked |
|---|---|---|---|---|
| Canada Enterprise Co 1 | $63,054.97 | Manager | 8/8 | |
| UK Government Co 8 | $46,732.57 | Manager | 8/8 | |
| Canada SMB Co 15 | $36,612.93 | Manager | 8/8 | |
| UK Mid-Market Co 22 | $30,202.89 | Manager | 8/8 | |
| Canada Enterprise Co 29 | $0.00 | Auto | 8/8 | |
| UK Government Co 36 | $55,108.42 | Manager | 8/8 |
Pick a rule. See everything it depends on.
Tavuni found 7 record types and 6 connections in the workbook, and compiled 18 formulas into rules about orders and customers, not cells. Choose one.
Every rule reads other values. Pick one and follow it.
- The ruleOrder Total is calculated for every order.
- Reads directly
Order.statusOrder.netAmountOrder.taxAmountOrder.freight - Through other rulesOrderLine, Product, PricingRuleMinSubtotal, Customer, PricingRuleUplift, TaxRate
- In the applicationThe Order Total field on every order: 120 of 120 equal the workbook.
- Evidence · SO-5000 · Canada Enterprise Co 1Tavuni$63,054.97Workbook’s stored value$63,054.97
IF(Order.status = "Cancelled", 0, ROUND(Order.netAmount + Order.taxAmount + Order.freight, 2))from the spreadsheet formula=IF(D4="Cancelled",0,ROUND(I4+K4+L4,2))Checked against an answer key Tavuni didn’t write.
The workbook stores what every formula produced, computed by an independent engine. Tavuni sets those answers aside, computes everything itself, then compares 2,870 values. All four outcomes are shown, not just the passes.
2,869 agree. I flag the one that doesn’t.
Show me the rounding edge.
2,870 values: 18 calculated fields, checked on every record. The table shows one field for the first 10 orders.
- 2,869Pass: equals the stored result
- 1Rounding edge: a midpoint where Excel and IEEE-754 round differently; reported on its own
- 0Discrepancy: pinned to a field, a rule and a cell
- 0Not measured: hand-entered values (0) and uncompiled rules (0); never counted as a pass
| Order | Answer key (workbook) | Tavuni | Result |
|---|---|---|---|
| SO-5000 | $63,054.97 | $63,054.97 | Pass |
| SO-5001 | $46,732.57 | $46,732.57 | Pass |
| SO-5002 | $36,612.93 | $36,612.93 | Pass |
| SO-5003 | $30,202.89 | $30,202.89 | Pass |
| SO-5004 | $0.00 | $0.00 | Pass |
| SO-5005 | $55,108.42 | $55,108.42 | Pass |
| SO-5006 | $73,245.63 | $73,245.63 | Pass |
| SO-5007 | $42,685.44 | $42,685.44 | Pass |
| SO-5008 | $56,432.48 | $56,432.48 | Pass |
| SO-5009 | $0.00 | $0.00 | Pass |
The one rounding edge: Order.freight at Orders!L102, Value sits on a rounding midpoint; expected 1109.78, produced 1109.79. Matches under IEEE-754 rounding, differs under Excel 15-digit rounding.
Say what the business needs. Get a reviewable change.
A requirement in plain words becomes a proposed change to the model, with the assumptions it made written out, so a person can disagree before anything happens.
- Operations on the model, not code: a rule and an approval process added.
- Assumptions stated, so they can be corrected.
- Produced here by deterministic code (local-heuristic (rule-based-v1)). No AI call.
“Add a manager approval when order value exceeds $10,000.”
- AddedAdded rule "Orders over $10,000 need Manager approval": Order.orderTotal > 10000
- AddedAdded process "Order approval" with 4 steps
- The threshold applies to Order.orderTotal (Order Total).
- Reusing the existing "Manager" role.
- The comparison is strictly greater than the threshold.
Change a rule. See exactly which orders move.
Before anything is published, Tavuni runs the proposed rule against the records that exist today and names the orders that newly need approval or no longer do.
What if the limit were $25,000?
Show me who it affects first.
At $25,000, 21 orders that need approval today would be approved automatically. 68 of 120 would still need a manager, worth $3,209,673 together.
| Order | Customer | Total | Today | Proposed |
|---|---|---|---|---|
| SO-5087 | UK Mid-Market Co 10 | $24,407 | Manager | Auto |
| SO-5062 | Canada SMB Co 35 | $23,734 | Manager | Auto |
| SO-5048 | Canada Enterprise Co 17 | $22,472 | Manager | Auto |
| SO-5071 | UK Mid-Market Co 18 | $21,578 | Manager | Auto |
| SO-5096 | Canada Enterprise Co 33 | $20,633 | Manager | Auto |
| SO-5073 | UK Government Co 32 | $20,255 | Manager | Auto |
| SO-5038 | Canada SMB Co 27 | $20,165 | Manager | Auto |
| SO-5102 | Canada SMB Co 35 | $19,103 | Manager | Auto |
and 13 more
Simulated on the 120 real orders in the example workbook. Nothing changes until a person publishes the new version.
A rebuild stops at the screens. Tavuni proves the numbers.
“From next month, orders over $10,000 need a manager’s approval. How many orders will that hold up, and can we trust the totals?”
Weeks of re-typing, and a guess.
- The logic is re-typed by hand2,993 formulas, copied into a new tool one at a time.
- Screens get builtAn order form, an approval queue.
- A few orders are spot-checkedThe rest are assumed to be right.
- The threshold goes liveHow many orders it catches is found out afterwards.
Live, and nobody can say whether the totals match.
Five steps, each with evidence.
- Reads the workbook2,993 formulas · 7 record types · 6 relationships
- Compiles the logic18 rules about orders and customers, not cells
- Checks every value2,869 of 2,870 agree · 1 rounding edge, reported
- Measures the rule89 of 120 orders · $3,566,077
- Waits for a personVersion 2 stays a draft until someone publishes it
The answer before the change: 89 orders, on proven totals.
For processes where being right matters more than being first to a screen.
The questions each kind of team brings to it. Typical questions, not customer quotes.
Pricing, quoting and order totals
Can we trust the totals the new system calculates?
Discount bands, tax lookups and freight minimums in a workbook several teams depend on.
Every value checked against the workbook before switching overQuality & testingTesting, inspection, certification
Which version of the procedure was this tested against?
Measurements judged against limits; two departments signing off.
Verdicts calculated by explicit rules, signed off in parallelOperationsApproval-heavy processes
Who approved this, against which rule?
Purchase, discount and exception approvals living in email.
Decision, approver and version in the audit trailAny team with a shared sheetRegisters that outgrew a spreadsheet
Can we keep the sheet's logic but add access control and history?
Customers, assets or cases tracked in a shared sheet that now needs a real application.
Proposed objects shown for correction before anything is createdOne model in the middle. Everything else compiled from it.
Sources are read into the Business Graph; the compiler turns the graph into the running system; verification closes the loop.
Change the model once; every screen and rule follows.
- Isolation
- Every organisation separated in the data layer; another tenant’s record answers “not found”.
- History
- A tamper-evident audit chain, including what automations did.
- Identity
- SAML and OIDC sign-on, SCIM provisioning, multi-factor authentication.
- Integration
- Scoped API keys, signed webhooks, OAuth 2.0 connectors with PKCE.
- Change
- Versions with authors; configuration promoted between environments.
- Runtime
- Python backend on PostgreSQL or SQLite: one test suite, both engines.
Engineered and tested. Not yet in production.
Over three thousand backend tests on two databases, browser tests on a production build, and no business running on it yet. We would rather say so than imply otherwise.
Read the full statusWhat buyers ask first.
Anything else: talk to us. The full status is on the About page.
Plain answers, including what Tavuni does not do yet.
Has anyone used Tavuni in production?
Not yet. The product is engineered and tested (over three thousand backend tests on SQLite and PostgreSQL, browser tests against a production build), but no business runs on it today. We are looking for one real process to pilot, and we say so on every page rather than imply otherwise. Talk to us about a pilot.
What if the AI gets something wrong?
The AI never edits your system. It may only propose operations from a fixed vocabulary: add a field, add a rule, add an approval step. Deterministic code validates them, applies them to a copy, runs them against your records and shows you the difference. Publishing is a person’s decision.
When the AI describes your business, every fact is traced to the sentence it came from, and it cannot claim that something was derived by the platform. How the AI is governed.
Can I see what a change will do before it goes live?
Yes. A change arrives as a diff in business terms (“added rule: orders over $10,000 need manager approval”) together with how many of your existing records it affects and what they add up to. Nothing is live until a person approves it as a new version. See the example.
Does it replace my spreadsheet, or sit beside it?
It replaces the spreadsheet as the system people work in, and keeps it as the answer key. Tavuni reads the workbook’s structure and formulas, compiles them into a model and rules, and checks the result against the values the workbook stored. Spreadsheet bugs are reproduced faithfully (the business runs on those numbers today) and that is said out loud, not hidden.
Which spreadsheet functions are supported?
178 of Excel’s roughly 500 functions, covering the lookups, aggregations, conditionals, rounding and dates that operational workbooks rely on. Every import produces a coverage report that names anything unsupported, and an unsupported formula is reported as such, never silently dropped.
Who can see what?
Access is decided by the server, not hidden by the screen. Roles carry explicit actions; record access can be all, own, team or none; individual fields can be restricted and are then never sent to a browser that may not read them. Organisations are isolated structurally, and asking for another organisation’s record answers “not found”. Governance in detail.
What does it cost?
There is no public pricing yet. During the pilot stage we agree terms with each pilot directly. Tell us about your process.
Does my data train an AI model, and which provider is used?
The public demonstration sends nothing to any AI service. Inside the platform, AI is switched off until an administrator registers a provider and model for their organisation, with a monthly budget and every call metered. Whether that provider trains on your data is governed by your agreement with that provider. Tavuni does not train models.
See what your own workbook computes.
Send one workbook that runs part of your business. We run it through the compiler and show you the coverage report and the verification result, including anywhere it disagrees with itself.