# How Should a Small Business Answer AI Questions in Tenders?

Canonical URL: https://business.theaieducator.io/posts/ai-questions-in-tenders-small-business
Publication: Dan Fitzpatrick Insights
Author: Dan Fitzpatrick
Topic: AI Strategy for Business
Published: 2026-10-29T16:39:06.000Z
Modified: 2026-10-29T16:39:07.498Z

Buyers now ask suppliers how they use AI, in tenders and customer questionnaires. This guide compares how UK, Australian and US buyers ask, and gives you the Buyer's Five: the five questions to answer in writing before a form arrives.

## In brief

Answer AI questions in a tender or supplier questionnaire the same way every time: what you use AI for, what data goes in, who is accountable, who checks the output, and what happens if it goes wrong or you stop. Write it once, keep it true, and never claim a control you do not run.

## Key takeaways

- Buyers ask suppliers about AI because a supplier's use of it has become part of the buyer's own risk.
- The UK's PPN 017 does not ban AI in bids; it treats it much like using a bid writer and offers optional, unscored disclosure questions.
- Three buyer approaches checked in October 2026, from the UK, Australia and the US, overlap on the same five questions.
- The Buyer's Five are: what do you use it for, what goes into it, who is accountable, who checks the output, and what happens if it goes wrong or when we part.
- A good answer names what was done, who did it and what stayed out; a weak answer offers only tone.
- Apply the Loop Test to 'a person checks it': the check must catch the error, in time, and the work must be recoverable.
- If the honest answer is 'not yet', say so with an owner and a date rather than overclaiming.

A small business should answer AI questions in a tender or supplier questionnaire the same way every time: say what you use AI for, what data goes into it, who is accountable, who checks the output, and what happens if it goes wrong or you stop. Write that account once, keep it true, and never claim a control you do not run.

That is the short version. The rest of this article shows where the questions come from, compares three sets of them checked at source this week (last checked 29 October 2026), and gives you a named way to prepare.

## Why are buyers asking suppliers about AI now?

Buyers ask because your use of AI has become part of their risk. The evidence is thinner than the trend pieces suggest, so here is what I could verify.

In the US, the Federal Reserve Bank of San Francisco reported on 15 July 2026 that nearly 40% of small business respondents to the 2024 Small Business Credit Survey said they were using or planning to use AI. The Bank calls it a qualitative baseline, not a full estimate, and a fuller look at the 2025 survey is still to come. In the UK, a Business Matters piece of 23 September 2026 says larger clients "have started asking what AI tools their suppliers use". That is trade-press reporting with no figures behind it, so treat it as a signal, not a measurement.

Public buyers have put the question in writing. The UK Cabinet Office's PPN 017 says suppliers' use of AI is not banned in bidding, and that a buyer should weigh it much as it would weigh a supplier hiring a bid writer. My reading: if your largest customer is a council, a utility or a company with a procurement team, one of these questions will reach you before the end of the year, and the business that answers it fastest and most plainly looks like the safer pair of hands.

This is a different question from the one in [Should a Small Business Tell Customers It Uses AI?](https://business.theaieducator.io/posts/ai-disclosure-small-business). That article is about what a customer would be surprised to learn. This one is about what a buyer will ask you in writing, and what you should be able to put in a box on a form.

## What do the buyers actually ask? Three approaches compared

Three published approaches show the pattern. I opened each this week. They come from three countries and three kinds of buyer, which is why the overlap between them is useful.

| Source and what I checked | Who it is for | What it asks of a supplier | Limits |
|---|---|---|---|
| UK Cabinet Office, PPN 017, version of 17 February 2025 (gov.uk) | Central government buyers; other public bodies may follow it | Three example disclosure questions: did you use AI to help write the bid, was it checked for accuracy, and is AI part of the service you will deliver | Annex B says the questions are for information only and should not be scored. The note does not mention small firms |
| Australian Government, "Questions to ask AI suppliers" (ai.gov.au, undated, opened 29 October 2026) | Any organization buying an AI product or service | Nine groups: risks, transparency, accountability, data, testing, oversight, fairness, exit, sustainability | Written for buyers of AI products, so it is stricter than a firm that simply uses AI needs to expect. It does not mention small business |
| US General Services Administration, GSAR 552.239-7001, effective 19 October 2026, as summarized by the consultancy Winvale on 9 October 2026 | Federal schedule contractors that sell large language model products | Protect government data, pass the clause down to subcontractors, report incidents | Applies when an LLM product is delivered to government; internal back-office use is excluded. I read a summary, not the clause itself |

A short verdict. The UK note is the one most small suppliers will meet, because it governs how a buyer reads your bid. The Australian list is the best map of what a careful buyer wants to know, so use it as your study guide. The US clause matters only if you sell LLM-based products to federal buyers, and it has moved: a March 2026 draft, as described by the law firm Gibson Dunn, would have required disclosing every AI system used in contract performance, while the final version Winvale describes centers on data safeguards and incident reporting. Rules shift. The five questions underneath them do not.

## What is the Buyer's Five?

The Buyer's Five are the five questions behind every customer, tender and supplier questionnaire about your use of AI, however each is worded. I am offering this as my suggested way to prepare, not as research.

The Buyer's Five are: What do you use it for? What goes into it? Who is accountable? Who checks the output? What happens if it goes wrong, or when we part? A business that can answer all five in writing, the same way each time, is ready for any questionnaire.

Here is where each one shows up in the three sources.

- **What do you use it for?** PPN 017 asks whether AI helped write the bid and whether it is part of the service. The Australian list calls this transparency.
- **What goes into it?** The Australian list asks, "Will any of our data be used to train AI models now or in the future?" The Business Matters article adds where data is processed and which plan type you use. The GSA clause is mostly about this one.
- **Who is accountable?** The Australian list asks who is accountable once the system is live. Business Matters says to name one person who approved the use.
- **Who checks the output?** PPN 017 asks you to confirm that AI-assisted tender responses were checked and verified for accuracy. The Australian list asks whether the system can be paused, overridden or switched off.
- **What happens if it goes wrong, or when we part?** The Australian list covers exit, data deletion and continuity. The GSA clause adds incident reporting.

If your answer to any of the five is "it depends who you ask in the office", that is the finding. A buyer reading a form cannot tell the difference between a business that has no answer and a business that has not written it down. Both look the same, and both get a clarification question, a delay or a quiet drop down the shortlist.

## What does a good answer look like?

A good answer is short, specific and true. Take a hypothetical 18-person marketing agency bidding for work with a regional housing association. The form asks: "Have you used AI or machine learning tools, including large language models, to assist in any part of your tender submission?" That wording is Example Question 1 from PPN 017.

A weak answer reads: "We take AI very seriously and use it responsibly." It says nothing a buyer can check. A reader finishes it knowing less than before.

A strong answer reads: "Yes. We used an AI assistant to produce first drafts of two of the six method statements. Our bid lead rewrote both, and our operations director checked every factual claim, certification and date against our records before submission. No housing association information was entered into the tool." It names what was done, who did it and what stayed out. Every clause in it is something the agency could prove if asked.

The strong answer is not longer because it is clever. It is longer because it contains facts. The weak one only contains tone.

## Does "a person checked it" count as an answer?

Only if the check would catch the error. This is where the Loop Test earns its place. The [Loop Test](https://business.theaieducator.io/posts/ai-for-retail-and-hospitality-small-business) is three questions I ask before accepting "a human checks it" as an answer: Would that person notice a wrong result without being told to look for it? Would they notice in time to stop it mattering? And if they missed it, could the work be put back? Oversight that survives all three is a safeguard. Oversight that fails any of them is a signature.

Apply it to a bid. Would your bid lead notice an AI-drafted sentence that claims an accreditation you do not hold, without being told to look for accreditations? A skim of a smooth draft often will not catch it. Would they notice before the bid goes in? Only if the check happens before submission, not after the award. Could the claim be put back once a buyer relies on it? Often not: PPN 017 itself tells buyers they may run extra clarification and due diligence checks to confirm that tender commitments can be delivered. A claim you cannot back up is the one that surfaces there.

So the honest version of "we check it" names the check: who, against what record, and before when. Knowing which record settles each claim is the same discipline as knowing where your facts live, which is the subject of [Is Your Business Data Ready for AI?](https://business.theaieducator.io/posts/is-your-business-data-ready-for-ai)

## What I Tell Business Owners

Across the business teams I train, the first gap is almost always the first of the five. Ask three people in the same office what the business uses AI for, and you will hear three answers, because nobody ever decided; people simply started. That is the pattern behind [shadow AI](https://business.theaieducator.io/posts/shadow-ai-small-business), and it is why a questionnaire feels like an ambush when it should feel like a form.

My advice is the same each time. Do not write the answer for the buyer. Write the answer for the business, then lift it into the form. The organizations I work with include Glasgow City Council, Ealing Council, Welsh Water and Findel, and the preparation that matters is identical whether you sell to a council or to a company: one agreed account, one named owner, one place it lives. The business that has that can answer a questionnaire in an afternoon. The business that does not spends the week arguing about what is true.

## What if the honest answer is "we have not decided yet"?

Say so, and say what you will do by when. A buyer can work with "we use AI to draft internal documents and have not yet set a rule for client data; we will have one in place by 30 November, owned by our operations director". A buyer cannot work with a confident answer that falls apart on the first follow-up.

Three rules keep you out of trouble. Never describe a control you do not run. Never promise a date you have not put in someone's diary. And when a question asks about something you do not know, such as how a tool's provider treats your inputs on your plan, find out from the provider's own terms before you answer, not from memory. If a contract or regulated sector is involved, ask your solicitor or compliance adviser to read the answer; I am not a lawyer and this article is not legal advice.

## What should you do this week?

You can do all of this without a project. Block two hours.

1. **List the AI you actually use.** Ask each team lead for the tools and the jobs, including the ones nobody approved. Write one line each.
2. **Write the Buyer's Five on one page.** One short paragraph per question, in plain words, naming a person for the third. If a paragraph cannot be written, mark it as the first job.
3. **Run the Loop Test on your check.** For every "a person checks it" on the page, write who, against what, and when. Fix any that fail.
4. **Agree who may send it.** One named person owns the page and approves every answer that leaves the building. Put a review date in the diary, no later than the next quarter.
5. **Align it with your policy and your customer promises.** If you have no policy yet, the [UK small-business AI policy template](https://business.theaieducator.io/posts/uk-small-business-ai-policy-template) or [Does a Small Business Need an AI Policy?](https://business.theaieducator.io/posts/does-a-small-business-need-an-ai-policy) will get you started.

If you are thinking about selling the business one day, the same page does double duty. A buyer of the company asks the same five things, and the [Sale-Ready File](https://business.theaieducator.io/posts/ai-and-selling-your-business) is the longer version of the page you just wrote.

## Where to go next

If you want to know where your business stands before a questionnaire lands, take the free [Business AI Readiness self-assessment](https://theaieducator.io/business-ai-readiness?utm_source=business.theaieducator.io&utm_medium=referral&utm_campaign=ai-questions-in-tenders-small-business): twelve questions, about five minutes, and a 90-day plan at the end. If your team is using AI but nobody has agreed how to talk about it, that is exactly what my [AI training for teams](https://theaieducator.io/ai-training-for-business?utm_source=business.theaieducator.io&utm_medium=referral&utm_campaign=ai-questions-in-tenders-small-business) is built for. And if you run a chamber, trade association or industry event where members are starting to face these questions, my [business keynotes](https://theaieducator.io/business-keynotes?utm_source=business.theaieducator.io&utm_medium=referral&utm_campaign=ai-questions-in-tenders-small-business) cover this ground.

## Sources and further reading

- Cabinet Office, [PPN 017: Improving transparency of AI use in procurement](https://gov.uk/government/publications/ppn-017-improving-transparency-of-ai-use-in-procurement/ppn-017-improving-transparency-of-ai-use-in-procurement-html), GOV.UK, version published 17 February 2025. Opened 29 October 2026. Paragraph 9 and Annex B.
- Australian Government, [Questions to ask AI suppliers](https://www.ai.gov.au/sites/default/files/2026-07/Questions-to-ask-AI-suppliers.pdf), ai.gov.au, undated. Opened 29 October 2026.
- Winvale, [GSA's new AI clause is here: here's what MAS contractors should know](https://info.winvale.com/blog/gsas-new-ai-clause-is-here-heres-what-mas-contractors-should-know), 9 October 2026. A consultancy summary of GSAR 552.239-7001.
- Gibson Dunn, [GSA AI procurement rules would introduce new disclosure and use rights requirements for federal contractors](https://www.gibsondunn.com/wp-content/uploads/2026/03/gsa-ai-procurement-rules-would-introduce-new-disclosure-and-use-rights-requirements-for-federal-contractors.pdf), 12 March 2026. Describes the proposed draft.
- Federal Reserve Bank of San Francisco, [AI Adoption Among Small Businesses: Qualitative Insights from the Small Business Credit Survey](https://www.frbsf.org/research-and-insights/publications/community-development-research-briefs/2026/07/ai-adoption-in-small-businesses-2024-sbcs/), Community Development Research Brief, 15 July 2026. United States, 2024 Small Business Credit Survey respondents.
- Business Matters, [The AI questions your clients are about to start asking you](https://bmmagazine.co.uk/business/the-ai-questions-your-clients-are-about-to-start-asking-you/), 23 September 2026. UK trade press, no figures cited.

*Dan Fitzpatrick is [The AI Educator](https://theaieducator.io/about): a keynote speaker, Forbes contributor and five-times bestselling author who has trained more than 150,000 educators in 30+ countries.*

## Frequently asked questions

### How should a small business answer AI questions in a tender?

Answer in plain, specific terms: what you use AI for, what data goes into it, who is accountable, who checks the output, and what happens if it goes wrong. Name people and checks you can prove. Never describe a control you do not run, and have someone qualified read answers for regulated work.

### Do small businesses have to disclose AI use in bids?

Not by default. The UK's PPN 017 says suppliers' use of AI in bidding is not prohibited and offers optional, unscored disclosure questions for buyers. Whether you must answer depends on the tender documents and contract terms, so read each form and ask your solicitor if a clause is unclear.

### Can I use AI to write a tender response?

Usually yes, unless the buyer says otherwise. UK guidance treats it much like using a bid writer, and asks buyers to consider checks for accuracy. If you use AI for drafting, have a named person verify every factual claim, certification and date against your records before you submit.

### What AI questions will customers and tender buyers ask suppliers?

Expect five themes: what you use AI for, what data goes into it, who is accountable, who checks the output, and what happens if it fails or you part ways. Wording varies by buyer, but the Australian Government's supplier question list and UK procurement guidance cover the same ground.

### What should I say if my business has no AI policy yet?

Say so, and say what you will do by when. For example, state that you use AI to draft internal documents, that a rule for client data is being written, and who owns it. A dated, owned commitment is more credible than a confident answer that fails the first follow-up question.

### Who should sign off AI answers in a supplier questionnaire?

One named person, usually an owner or operations director, should approve every AI answer that leaves the business. That person checks it against what the business really does, keeps the master version, and reviews it on a set date, no later than the next quarter.

---
Source: [How Should a Small Business Answer AI Questions in Tenders?](https://business.theaieducator.io/posts/ai-questions-in-tenders-small-business)
Publisher: [The AI Educator](https://theaieducator.io)