What works when the firm has no AI key

Which Achekam features still work for a firm with no AI key, which fall back to keyword rules, fixed rules or standard wording, and which simply refuse.

By Achekam · 11 October 2026 · 5 min read · For Accountants & audit firms in South Africa, Nigeria, Kenya

Not every firm wants AI in its workflow, at least not yet. Some partners are waiting for their professional body's guidance. Some clients have asked that their information not be sent to AI vendors. Some firms simply want to see the system working on its own terms before adding anything. And some have not got round to setting up a key.

Whatever the reason, the practical question is the same: if we do not configure AI, what still works, and what will staff see? A firm deciding whether to switch AI on, or to leave it off for a while, should know exactly where the edges are.

This article goes through the features that touch AI in Achekam and what each does when there is no AI key.

Every AI feature says so

Achekam's AI features all run through one AI engine, which uses a firm's own API key first and the platform's key only where allowed. Administrators manage keys under Settings > "AI models, keys, routing & spend".

With no key at all, every AI feature says so and drafts nothing. Nothing pretends to have worked. A staff member who presses an AI button sees a message, not an empty draft that looks like a finished one.

That honesty matters more than it sounds. The worst outcome would be a feature that quietly produces nothing, or produces a placeholder that someone mistakes for real output. Without a key, the system tells you plainly that the AI part is not available, and the rest of the work carries on.

Trial balance mapping: keyword rules

Mapping a client's trial balance is where many firms first expect AI to help. Achekam suggests where each trial balance line belongs from four sources: the file's own mapping column, earlier mappings, AI, or keyword rules. Without an AI key, keyword rules are used, and the suggestions are labelled so staff know where they came from.

For established clients the gap may be smaller than it first appears. Earlier mappings are one of the sources, so a client's own history still counts, and a file that arrives with its own mapping column is used too. Keyword rules then cover what neither source answers.

What does not change is the control: no suggestion becomes an approved mapping until a person approves it. With or without AI, mapping is a person's decision.

Financial statements: standard wording

When an AI key is set up, AI writes narrative text in the financial statements, such as the nature of business and note wording, which is checked in code and labelled "AI draft - review before issue". Without a key, standard wording is used instead.

The figures are unaffected either way, because they come from code, not from AI. A firm without a key gets statements built from the mapped trial balance in exactly the same way; the difference is that the narrative starts from standard text rather than a tailored draft.

Standard wording still needs reviewing. It is a starting point written for the general case, and the reviewer and partner should adapt it to the client just as they would an AI draft. The safeguard in both cases is the review stages and the partner's signature.

Grading and drafting

Lead grading: fixed rules

In the CRM, staff can have AI grade a lead from A to D against a weighted rubric. When AI is not available, fixed rules grade the lead instead. Staff can override any criterion either way, and the grade is saved on the lead for the whole firm.

The grade is advice. Nothing acts on it automatically, so a firm without AI loses some nuance in the grading, not any control over what happens to the lead.

Working-paper drafting: refused

Some features have no non-AI fallback, and the clearest is drafting on working papers. With a key, staff can ask AI to draft a conclusion or a procedure's response, suggest review notes or explain an automated observation, and a draft enters the file only when a person saves it. Without a key, those actions refuse and point to Settings.

That is the right behaviour. There is no sensible "standard conclusion" for an audit area, and a fallback that produced one would be dangerous. Staff simply write the conclusion themselves, as they always have.

Reading documents without AI

Document reading is split. Scanned or photographed bank statements are read by text-recognition software on Achekam's own server, which is on by default, and checked against the running balance; that does not need an AI key.

Supplier invoices and receipts are different. Achekam reads scans and photos of those only when an outside OCR service or an image-reading AI key is set up. Without either, a review item called OCR_ENGINE_NOT_CONFIGURED is raised, so a firm without AI should either set up an OCR service or plan for staff to capture those documents by hand.

A summary for the partners

Choosing whether to use AI is a decision about the firm's resources and its risks, both of which a system of quality management has to address. ACCA's article on ISQM 1 explains the eight components such a system must address, starting with the firm's risk assessment process. Record the decision, and the reasons, wherever the firm keeps its quality management documentation.

If the firm later adds a key, what the firm's AI use costs explains how to watch spending, and what Achekam's AI can see, do and refuse sets out the controls that come with it.

Sit down with the firm's administrator this week and confirm, for each row of the table, what your staff will see today.

General guidance only, not advice on your own circumstances. Tax and payroll rules change: check the current position with the authority or your adviser.