The bid-writing AI race is solving the wrong problem

AI can now produce a credible first draft faster than most teams can schedule a review. That matters. It is also the smallest part of the opportunity.

The expensive failures in public-sector pursuit usually begin before the response is written. An opportunity enters the pipeline because the revenue is attractive. Optimism hardens into commitment. Buyer and incumbent knowledge is incomplete. Evidence gaps are discovered late. Price becomes the only remaining lever. By the time AI is asked to draft the answer, the organisation may already be executing a weak decision faster.

The market needs a different use of AI.

Stop asking AI to sound confident. Ask it to test whether confidence is justified.

The first job is not prose. It is challenge.

What does the market evidence actually say? Which buyer needs are sourced and which are inherited assumptions? What advantage does the incumbent hold? Which proof points survive scrutiny? Where does the commercial model depend on another party? What must be true before this opportunity deserves more capital?

Those questions are repetitive enough for AI to help at scale and important enough that the answers must remain auditable.

The market is not bid writing

The UK's 37 largest strategic suppliers spend an estimated £0.8bn-£1.0bn a year on public-sector pursuit activity, including the bids they lose. Around 70% is sunk on unsuccessful bids.

Some of that spend is the necessary cost of competition. The category error is to treat all of it as an unavoidable overhead and then optimise only the production layer.

The real market is the capital suppliers allocate to win public-sector revenue.

The tender is often too late

Advantage is created in account knowledge, framework choice, early market engagement, evidence readiness, stakeholder understanding, partner strategy and commercial positioning. The notice compresses time; it does not create context.

That is why pursuit intelligence must begin before the formal procurement. Identify the opportunity earlier. Qualify whether it is viable. Shape the evidence and path to win. Execute with an integrated solution, commercial and response position. Then learn from the outcome.

Governed AI is more valuable than generic generation

AI should challenge and assist the whole pursuit: assemble evidence, compare sources, find contradictions, generate questions and test assumptions. Deterministic services should own arithmetic, thresholds and gate logic. Experienced people should remain accountable for go / no-go, strategy, price and risk.

That division of labour is not a limitation. It is the product.

Government should ask a better AI question

The relevant question is not whether AI touched the tender. It is whether the supplier can show which claims are sourced, which assumptions were tested, who approved the response and whether the proposed capability is real.

Cabinet Office PPN 017 gets that balance right. Supplier use of AI is not prohibited, but it can be disclosed, checked and subjected to proportionate due diligence. The standard disclosure questions are for information, not a reason to penalise a bidder simply for using AI.

That creates a positive role for governed pursuit intelligence. It can help credible challengers and SMEs compete with better evidence and lower repeatable cost while giving buyers a clearer audit trail. Better competition does not come from more machine-written bids. It comes from more credible bidders making stronger, verifiable offers.

Every loss should make the next pursuit better

Most organisations store the submission and hold a debrief. The causal trail disappears: what the team believed at qualification, when the probability changed, which evidence remained weak, how much effort was consumed, why price moved and what the evaluator ultimately said.

A pursuit-intelligence platform retains that memory as structured evidence. The next opportunity starts with the accumulated truth of the last one, not a blank folder and the recollection of whoever is still in the room.

The competitive implication

For challengers, AI pursuit intelligence buys down incumbent advantage without requiring an incumbent-sized bid machine.

For strategic suppliers, it is margin defence. If competitors lower the cost of sale, they can price sharper, retain more margin or invest more selectively. The risk is not that one platform disrupts the incumbent. It is that the economics of competing change around it.

The question for leaders

Should AI produce more bids - or make fewer, stronger bets?

The first is productivity. The second is a new operating model.