Corporate housekeeping with AI: resolutions, minutes and registers
By the ChamberOne team 5 minute read
Corporate housekeeping is the work nobody brags about and everyone needs: board resolutions, minutes, registers, annual filings. It is repetitive, deadline-driven and easy to get slightly wrong. That makes it a good candidate for templates, and a reasonable candidate for AI assistance. It is also an area where a small error, such as the wrong threshold for approval or a missing signatory, can undo a transaction. This article sets out a division of labour: what to leave to templates, what AI can usefully draft, and what a person must check every time.
Key points
- Use approved templates for fixed wording and let AI help fill and adapt the variable parts.
- Confirm the approval route, notice, filings and thresholds against current law and the constitution each time.
- Treat AI-drafted minutes as working papers and compare them to notes and attendees' recollection.
- Reconcile registers against resolutions and filings regularly, with a person deciding on discrepancies.
- Use a second reader who works from source documents rather than from the draft.
Why housekeeping suits automation
Most resolutions follow a pattern. Approve a document, appoint an officer, authorise a signatory, confirm a prior act. The wording varies in a handful of places: names, dates, amounts, the document being approved. A template with merge fields handles that variation well and produces consistent output.
AI adds something templates cannot: it can read the source material and fill the variable parts. Given a term sheet or an email thread, it can propose the recitals, extract the parties and draft a first resolution. That saves the corporate secretary from retyping and reduces transcription slips, though it adds a new kind of risk: the tool may confidently insert something that was never agreed.
- Handle variation in names, dates and amounts
- Produce consistent output
- Can be reviewed once
- Reads source material and fills the variable parts
- Can propose recitals and extract the parties
- May confidently insert something never agreed
Templates first, AI second
Start with the firm's approved templates. They carry the structure and the wording the firm has decided to use, and they can be reviewed once. Use AI to populate or adapt them, not to write from a blank page.
A sensible hierarchy looks like this: a locked template for the standard wording, merge fields for facts held on the matter, and AI for the parts that vary in substance, such as describing the transaction in the recitals. The more of the document that is fixed, the less there is to check.
What AI can usefully do
- Read a set of emails or a term sheet and list the facts a resolution will need.
- Draft recitals that describe a transaction in neutral language, for a person to correct.
- Turn rough meeting notes into a first draft of minutes in the firm's format.
- Compare a draft against the previous year's version and highlight changes.
- Check a register entry against the underlying documents and flag mismatches in names or dates.
- Produce a checklist of documents that a particular kind of transaction usually involves, for a lawyer to adjust.
What must be checked every time
Statutory requirements differ by country, by company type and by the company's own constitution, and they change. This article does not say what any of them require. What it can say is that the following items should be confirmed against current law and the constitutional documents each time, never assumed from a previous matter or a tool's output.
- Who is entitled to pass the resolution and by what majority or approval route.
- Whether any notice period, quorum or conflict-of-interest procedure applies.
- Whether the matter needs a statutory filing, and the form and deadline for it.
- Whether any threshold, such as a value limit, triggers a different procedure.
- Whether the company's own constitution adds restrictions.
- Whether the signatories shown are the right people and still hold office.
Minutes and the record of what happened
Minutes are a record, not a story. They should say what was decided, by whom and on what basis, without embellishment. AI is good at tidying rough notes but it will sometimes smooth over a disagreement or invent a tidy conclusion. Compare the draft to the notes and to the memory of those who attended.
If a meeting was recorded or transcribed, check the privacy and consent position before feeding the transcript to a tool, and decide whether the transcript itself is kept. Keep the approved minutes as the record and treat the AI draft as a working paper.
Registers and the problem of stale data
Registers of members, directors or charges are only useful if they are current. AI can help reconcile a register against recent resolutions, but it can only work from what you give it. If an allotment never reached the register, the tool will not know.
A good habit is a periodic reconciliation. Once a year, or after any major transaction, compare the register to the resolutions, the filings and the signed documents. Ask AI to list apparent discrepancies and have a person decide which are real.
- 1Compare the register with resolutions, filings and signed documents
- 2Ask AI to list apparent discrepancies
- 3A person decides which are real
- 4Correct the register
- 5Repeat each year or after a major transaction
- Then back to step 1.
A hypothetical workflow
Suppose a three-person corporate team handles housekeeping for a few dozen small companies. For a share transfer, the paralegal opens the matter and the template for the board approval. A tool reads the signed transfer form and proposes the party names, date and number of shares. The paralegal confirms each against the form. The associate checks the approval route against the company's constitution and current law, and confirms what must be filed. The partner reviews the final pack, and the register is updated only when everything is signed.
The tool saved typing. It did not decide anything.
Filing deadlines and the human calendar
Housekeeping fails most often on dates. A filing deadline missed by a week can mean a penalty or a gap in the public record. Keep a compliance calendar for each company, owned by a named person, and enter dates from the source documents rather than from anyone's memory or from a tool's suggestion.
AI can help by reading a set of resolutions and listing the events that may trigger a follow-up, such as an appointment or an allotment. Treat the list as a prompt for a person to check what, if anything, is due and by when. The check must use current law and the registry's own instructions.
When a deadline is close, work from the official form or portal, not from a template written last year.
Where it goes wrong
Typical failures are quiet. A resolution that refers to the wrong class of shares. A date copied from last year's document. A signatory who left the board. AI can introduce or repeat these errors as readily as a tired human. Reduce the risk with locked templates, a short pre-signing checklist and a second reader who works from the source documents, not from the draft.
This article is general information about the law, legal practice and technology, current when it was published. It is not legal advice. Laws and guidance change: check the sources linked above and take advice on your own circumstances from a lawyer qualified where you are.
Read next
AI contract review: a first-pass checklist for lawyers
What to ask an AI to flag on a first read of a contract, what it tends to miss, and where a lawyer must read every word.
5 minute readWhat AI document analysis extracts well, and what it misses
What AI document analysis does well on parties, dates, clauses and citations, and what it misses, with ways to spot the gaps.
5 minute readAI-assisted engagement letters: what to automate and what stays human
Merge fields and templates handle the routine parts of an engagement letter, while scope, fees and limits of liability need a lawyer's decision.
5 minute read