NEW Build calculation models with your AI agent over MCP

The subledger that runs on your rules

Describe it. Build it. Prove it.

Design it visually or with your AI agent. Run it at scale. Trace it to the cent.

Book a demo See how it works
Your AI agent · via MCP
Add a prepayment adjustment step
Regression suitev8 vs v7
✓ 48 / 48 cases unchanged
Logic Studio / Loans / EIR fee amortisation
Events Attributes Rules Models Regression ● Diagnostics
Steps● live test passing
01eir = RATE(term, pmt, −netCarrying)
02interest = opening × eir
03feeAmort = interest − coupon
04if prepaid → accelerate(feeAmort)AI agent draft
05post(DeferredFees, FeeIncome)
Schedule · LN-0042
Dr Deferred Fees412.18
Cr Fee Income412.18
ASC 606 · RevenueIFRS 15IFRS 9 · ImpairmentCECLASC 310-20 · Loan feesASC 842 · LeasesIFRS 16IFRS 17 · Insurance+ your own policies
Why Fyntrac

The world’s most-used accounting system is still a spreadsheet. Not out of habit — because nothing else lets accountants own the math.

Spreadsheets give control but buckle under volume, audit and key-person risk. Packaged software scales, but locks you into someone else’s reading of the accounting. Fyntrac keeps the control and adds the infrastructure.

The false trade-offcontrol × scale
CONTROL → SCALE & GOVERNANCE →
neither
Spreadsheetscontrol, no scale
Packaged softwarescale, no control
Fyntracboth
What each option gives you
Spreadsheets Packaged software Fyntrac
Accountants own the logic ✓ — ✓
Any standard, your interpretation ✓ — ✓
Every number traceable ✓ — ✓
Transaction-level scale — ✓ ✓
Audit trail & version control — ✓ ✓
Posts journals to the GL — ✓ ✓
Changes without a services project ✓ — ✓
How it works

Define. Compute. Deliver. All three are yours.

Inputs, logic and outputs are all configured by you — and every stage is versioned and lineage-linked.

01

Define

Model your world — the events you post and everything that describes them.

Transactions Attributes Balances & metrics Custom tables Chart of accounts & mappings
02

Compute

Write the logic in Logic Studio — or have your AI agent write it for you over MCP.

Logic
Studio
schedules conditions iterations chaining
Your AI agent, plugged in via MCP
03

Deliver

Journals, a subledger at any granularity, and analytics — straight out of the engine.

Dr Interest Receivable2,940.00
Cr Interest Income2,940.00
JournalsSubledgerReporting & analytics

Every stage versioned and lineage-linked — the audit story a spreadsheet can never tell.

Logic Studio

Build it. Ask for it. Prove it still works.

A computation studio made for accountants — with your AI agent as a co-builder and a regression suite that guards every change.

Bring your own AI

Connect your AI agent over MCP

Plug your AI agent into Fyntrac through the Model Context Protocol. Describe the treatment in plain language — it defines the events, writes the steps and rules, dry-runs them on sample data and explains any error. You review and approve.

Works with Claude, ChatGPT and any MCP-ready agent.

Amortise origination fees on the commercial book using the effective interest method, monthly.
✓create_event_definitions3 events
✓create_saved_rule5 steps
✓dry_run_rulebalances tie out
The rule is ready for your review in Logic Studio. Shall I add these loans to the regression suite?
Visual DSL

A language accountants actually read

Drag and drop or write in the playground, using the accounting functions you already know. Test in real time as you build.

Stepsone calculation each
Rulessteps that post
Modela full treatment
Model Seta portfolio
Regression suite

Change the model without breaking the past

Capture test cases — with their own input data and expected results — while you build. Every time the model changes, regress the whole suite and see exactly which numbers moved, so backward compatibility is never a guess.

Multiple cases per model Version-to-version diffs Step-level diagnostics
Run · model v8 against v7 baseline 5 cases
Bullet loan · fixed rate✓ match
Amortising · rate reset✓ match
Partial prepayment · M14◆ expected change
Modification · term ext.✓ match
Write-off · stage 3✓ match
Fyntrac Insight

BI built on your subledger. With a copilot that does the math.

Datasets, reports and dashboards on the same lineage-linked data your engine produced — and an agent you can question in plain language.

Insight Copilot

runtime math agent
What’s our weighted-average yield by segment this quarter, and how much did prepayments add?
→ building dataset · loans × segment × Q3
yieldseg = Σ(interest + feeAmort) ÷ avg(balance)
prepayLift = feeAmortaccelerated ÷ avg(balance)
✓ computed on 3 segments · traced to rule v8

Commercial leads the book, and accelerated fees from early prepayments account for most of its lift this quarter.

Save as KPI Open in Prism Forecast
COMMRETAILSME

Datasets

Slice any attribute into clean, reusable datasets.

Reports, KPIs & dashboards

Tables, bars, pies and pivots, assembled for each audience.

Prism

A query browser — explore the subledger with simple SQL.

Reconciliation

Match two datasets, or a dataset against a CSV, on your own logic.

Close & consolidate

Built for the close. Not just the calculation.

JAN
FEB · restated
MAR
APR · open
restatement flows forward into the open period

Close, rewind, replay.

Lock a period and reported history stays put. Need to restate? Rewind a closed period and Fyntrac replays the adjustment into the current open period — automatically, with the full trail.

Revenue Lending Leases
Consolidated journals + reporting

Many tenants, one view.

Run each use case in its own tenant with its own logic, then consolidate journals and reporting across all of them.

Own the logic.
Not just the output.

Bring one of your real treatments. We’ll build it with you in Logic Studio — and show it running end to end.

✓Your accounting treatment, modelled live ✓Your AI agent connected over MCP ✓Journals and Insight on your own data

Book a demo

Prefer email? Write to us at info@fyntrac.com

We’ll only use your details to arrange the demo. See our privacy policy.