Ask most AI tools to help with an audit and each new chat starts from scratch. They can be sharp in the moment — the trouble is the moment is all they hold. Open a fresh chat and the engagement is gone: the file, the policies, the decisions, all re-explained from the top. A goldfish, every lap around the bowl a clean slate. For a one-off question that's fine. For an audit, which runs for weeks and builds on its own work, it's a problem. Auditors need an elephant, not a goldfish.
An auditor carries the engagement with them: everything from last year's file and the firm's way of doing things, to every decision made on this year's engagement so far. Relay is built to do the same. Here's how it carries all this context, and what that does for the engagement.
Engagement context
You give Relay the engagement: prior-year working papers and AFS, the trial balance, the client's accounting policies, supporting evidence, email communication and the rest. Relay reads all of it and keeps track of how it fits together — which policy governs which account, what support matches which amount, which client contact provided what support, and what last year's team concluded and why.
It builds those connections into a knowledge graph: a single connected map of the engagement, rather than a stack of documents searched one at a time.
Reading the engagement
Relay reads the whole engagement into one connected map — last year's file, the ledgers, the policies.
High Level Audit Knowledge Base Example
It's closer to how a manager holds the file in their head: ask about a revenue figure and they already know the recognition policy behind it, the risks identified during planning, the name of the client's revenue clerk, and the three key customers driving the balance. Relay builds a similar map of related items, using this context to perform testing, build workbooks and answer queries.
The map that evolves with the engagement
The context map is never frozen. As new evidence arrives throughout the engagement, Relay folds it into the knowledge graph. The picture stays current as the work progresses.
So what happens when the inevitable updated trial balance is received? Or, heaven forbid, materiality figures are revised?
Relay identifies exactly which amounts or assumptions have changed, and using its connected map of the engagement, it knows which accounts those changes affect and how they tie to work already performed. Version control becomes a problem of the past.
Your firm's own methodology
Two audit firms can run the same test and document it completely differently: different templates, different sample sizes, different thresholds, different standards for what counts as sufficient support, different types of review notes. Both comply with the International Standards on Auditing (ISAs), but each has its own way of doing things — the methodology of the firm. Relay is built to take this into account; it works to the way you do things.
Your firm likely has its own templates and procedures Relay can follow: how you select a sample, what you accept as valid support, how you test cut-off, how a working paper has to be documented before it's ready for review. The same methodology gets applied the same way across every engagement and every team, ensuring consistency across your firm.
Yet don't think for a moment that Relay can't act where firm methodology is silent. Relay has the full context of the ISAs, and everything it does is built to align with those standards.
What this looks like in practice
Take revenue testing, the kind of job that runs for weeks in every audit and trips over its own early decisions.
Relay starts with the engagement context: last year's revenue working papers, the client's revenue recognition policy, the current year's trial balance, the sales ledger and the relevant client communication, and builds a map of revenue and all its related items. But before any testing starts, Relay reaches for your firm's methodology: your templates and colour schemes, your sampling rules, your preferred procedures and standards for support.
A tailored audit workbook is built, all procedures clearly stated, and a sample of invoices selected for testing. That sample is added to Relay's Open Items, ready to be requested from the client. Once support is received, it's inspected and instantly added to Relay's web of relational data — ticked in and referenced in the workbook.
Not the support that was needed? Updated trial balance received? Material errors requiring extended sampling? Relay rolls with the punches, keeping a meticulous record of the evolving engagement and adjusting workbooks and documentation on the go.
No one had to re-explain the engagement at each step. That's the difference between a tool you prompt and a tool that's actually working the job with you.
What this means for an audit team
- Less explaining, more doing. Relay doesn't need to be told what's happening in the engagement. Your team spends its time on judgement, not on prompting and explaining.
- Consistency. The same tailored methods of your firm, applied the same way, on every engagement and by every person.
- Auditability. Every decision is recorded with its reason and where it was applied: a file a reviewer can follow, not a black box.
An audit only grows as it progresses. Relay's systems of engagement and firm context mean none of the work already done has to be rebuilt as the engagement gets bigger. Relay is an elephant, not a goldfish.