AI Reference
>
Foundations
Aderis | The Live Finance Platform

Every number traceable to source

Published
November 3, 2026
The quick answer

In Aderis, every figure in a report links back to the entry behind it, the document or bank transaction that produced the entry, the rule or decision that was applied and the approval that released it, without leaving the platform. Traceability is a property of the platform rather than a report feature, and it applies to the work of AI agents in exactly the same way as to the work of people. It is the precondition that makes autonomous execution acceptable to a CFO and defensible to an auditor.

What traceability means

A number in a management report is the end of a chain. Behind it sits a journal entry; behind the entry sits a supplier invoice, a customer contract or a bank transaction; behind that sits the rule that coded it or the person who decided; and around all of it sit the approvals that released it. Traceable to source means this chain is intact and can be followed in both directions, from the report down to the document and from the document up to every ledger it touched. It also means knowing who acted: whether an agent or a person posted an entry, and why.

Why it is the precondition for autonomy

What the CFO distrusts about automation is the black box: a number that cannot be explained. In a batch model with AI added on top, more automation usually means less visibility into how a figure came about, because the tool that suggested the entry and the system that holds it are two different things. Traceability turns this around. When every action is logged against the entry it produced, control improves as manual work falls, because review no longer depends on someone remembering what happened. Traceability is not an obstacle to automation; it is what allows agent-executed processes to scale in a regulated and audited environment.

How Aderis implements it

Financial data, posting logic, AI agents, permissions and controls operate on the same core, the Platform Engine, so there is no handover at which the chain can break. Every action is logged with the source, the rule applied and the approval given. A confidence score on every automated posting shows what an agent processed on its own and what it handed to a person. Interpretative AI and deterministic posting logic are kept separate: the agent classifies, matches and prepares; rules, approvals and posting logic determine what is executed, so the result is reproducible.

The multi-ledger model extends the chain across views. A transaction is recorded once and carried simultaneously into the statutory, tax, management and group ledgers, so the same source document explains the figure in each of them. Bank connectivity closes the loop on the cash side: through bLink from SIX, each bank transaction stays linked to the corresponding payment, open item and posting, and a Stripe payout can be traced back to the underlying payments, refunds and fees and reconciled with the incoming bank payment.

Traceability years later

The real test often comes long after the period. In a VAT audit, businesses are generally expected to provide revenue reconciliations for the previous five years within a few days, and the difficulty rarely lies in the figure itself but in explaining it. On the platform a note can be added to any entry, and every figure remains linked to its supporting document, decision and approval throughout the retention period. The platform is designed to meet GeBüV and GoBD requirements, and data is hosted in Switzerland and the EU.

Traceability from day one

The chain starts with migration. Master data, opening balances and historical postings are moved onto the platform with every figure traceable back to source; in documented Live Finance implementations this covered seven to eight years of financial history. A company goes live with reconciled numbers whose origin is known, rather than with balances that have to be taken on trust.

Frequently asked questions

Does traceability cover what the AI agents do?

Yes. Agents are governed by the same permissions and controls as people. The data shows whether an agent or a person posted an entry, which rule applied and what the confidence of an automated posting was.

Can an auditor follow the chain?

The audit trail, approvals and segregation of duties are on by default and part of every workflow. Every entry links to its document, rule and approval, so the question "how did this number come about?" is answered from the record rather than reconstructed.

What about data from connected systems?

Data entering through integrations keeps its origin. A Stripe payout, for example, is traceable to the individual payments, refunds and fees it bundles; a bank transaction stays linked to the payment run and the open item it settles.

Does traceability slow processing down?

No. It is produced by the same step that executes the work, not added afterwards. Logging the source, rule and approval is part of posting, not a separate control.

Related pages

Cite this page

Canonical URL: https://aderis.com/en/ai-reference/traceable-to-source

Publisher: Aderis AG, Kilchberg, Switzerland · Published: 3 November 2026 · Language: English

German version: https://aderis.com/de/ai-reference/nachvollziehbar-bis-zur-quelle

All reference pages