The TTL platform

One shared processing logic for four specialised products.

TTL connects data import, normalisation, classification, rule engine, review, and reporting in one end-to-end process. Product-specific specialist logic stays separate. Governance, audit trail, and data model are shared.

Request a platform demo →Data & integrations
Architecture

Eight layers. Each one documented.

The products share the following layers; only the rule sets are product-specific. Every result stays linked to source, rule, calculation, user or agent, and approval.

01
Data import
Files, API data, bank reports, ERP exports, and documents are ingested technically. Hash, source, parser version, import time, and original data are retained.
shared
02
Normalisation
Fields, keys, amounts, currencies, date types, and references are mapped into a unified data model.
shared
03
Reference & profile master
Instruments, accounts, counterparties, countries, fund types, tax profiles, reporting profiles, balance-sheet assignments, and exchange rates are maintained with validity periods.
shared
04
Classification engine
Source events are assigned to a canonical specialist class. Known cases run deterministically. Unclear documents or free text can be interpreted technically as structured candidates.
shared
05
Product rule sets
Product-specific rule packs process Depot, AWV, FX, or Private Tax matters. Rules consider profile, period, instrument, event, country, amount, and history.
per product
06
Review & governance
Conflicts, missing data, or defined special cases become tasks, statuses, and approvals. Every manual decision requires a rationale and an accountability reference.
shared
07
Reporting & export
Booking entries, filing data, tax reports, itemised evidence, and management overviews are produced from structured result objects.
shared
08
Audit & reconciliation
Every result stays linked to source, rule, calculation, user/agent, and approval. External holdings or reports can be used for reconciliation.
shared
Use of AI

Rules first. AI afterwards.

TTL uses AI where source texts, documents, or inconsistent labels cannot be processed in a purely deterministic way. AI supplies candidates, evidence, missing data, and confidence. It does not produce unchecked final bookings, tax decisions, or AWV filings.

Unknown bank description
AI candidate · confidence 0.82
review
Known source key · BUY_EQUITY
Deterministic rule DEP-CLS-0412
automated
Reviewed candidate
Approved by role: Tax Lead
approved
Governance by design

Five properties that every result carries.

Versioned rules

Every decision references a rule ID, version, and validity period.

Validity periods

Master data, profiles, and parameters apply for defined periods.

Idempotence

The same input with the same rule version produces the same result.

Reversal instead of silent change

Corrections are historised.

Manual review list

Unclear cases stay visible with category, priority, evidence, and approval status.

See how TTL turns raw data into an auditable process.

Request a platform demoExplore products