AI is changing how tenders are written but not how they’re evaluated in Ireland. That gap is becoming a problem says BidReview founder Tony Corrigan.
Until very recently, the biggest challenge in public procurement was getting SMEs to compete at all. The submission process was torturous, the success rate was low, and most business owners took one look at the process and decided their time was better spent elsewhere. I spent decades trying to streamline the workflow, but it didn’t really change anything.
Then AI arrived, and everything changed.
Through 2024 and 2025, companies that had never tendered before started feeding sales collateral into ChatGPT and passing the results into submissions. Today, a bidder can choose from dozens of AI-powered proposal platforms that will generate a full response from the barest scrap of input.
I evaluate tenders for a living, and from the buying side of the table, the volume has exploded. Competitions that attracted three or four bids in 2023 are now attracting twelve or fifteen. In one recent evaluation I took part in, roughly 30pc of the submissions were entirely AI-generated, another 40pc were largely AI-generated, and only the last 30pc had been meaningfully written by a human.
Buyer concerns
Buyers have started to notice. Two things worry them.
The first is what happens to their RFP documents once a supplier feeds them into a multi-tenant AI model they don’t control. Clauses restricting the use of AI in proposal preparation are now appearing in more RFPs. In theory, suppliers aren’t penalised for declaring it. In practice, AI-generated copy has a fingerprint, and I wouldn’t bet against some evaluators being more sceptical about claims in a submission that has been obviously ChatGPT’d end to end so that they don’t even have to get into evaluating the proposal.
The second and more serious concern is what the AI puts in that was never true. The polite term is hallucination. In a procurement context, that word is far too soft. What’s actually happening is that a supplier with real gaps in experience, capability or resource has those gaps plastered over by a model trained to produce a plausible-sounding response. The bid reads well. The business behind it may not be able to deliver. In the best case, the buyer wastes time evaluating a proposal that was never at the races. In the worst case, they award a contract to a supplier who can’t fulfil it.
The supply side of the market has been transformed. The evaluation side has not moved. Every tender I have evaluated this year has been read, scored and debated by human beings, sitting in a room or on a call, working line by line. There are no widely adopted AI evaluation tools, and I don’t expect them anytime soon. If a buyer outsourced their judgement to a model and a losing supplier found out, the fallout would be severe. AI’s well-known tendency to produce agreeable answers is a feature when you’re drafting sales copy and a catastrophic bug when you’re deciding who gets a €250,000 contract.
So, we now have a market in which it takes no time at all to produce a credible-looking bid, and exactly as long as it ever did to evaluate one properly. The volume has quadrupled. Indeed, there’s a tendency among evaluators to disqualify bids on the basis of qualification criteria and precedent contracts. And the buyer’s best defence against a flood of generic, plausible, interchangeable submissions is the one thing that has been a feature of public procurement for as long as I’ve worked in it: retreating to the suppliers they already know.
Past delivery wins
Fewer than 3,000 businesses have recorded a public sector contract win on this island in the past two years. That is a fraction of the companies that are perfectly capable of delivering. Evaluation panels score on evidence of past delivery, and past delivery almost always means past delivery for someone else. The consequence, over time, is supply chain concentration: buyers end up dependent on a narrow pool of established suppliers, pricing power shifts, flexibility reduces, and when disruption hits, there are no well-developed alternative relationships to fall back on.
The arrival of AI has not broken this pattern. It has reinforced it. When an evaluator is faced with fifteen submissions, several of which are visibly template-generated and indistinguishable from each other, the rational response is to weight known suppliers even more heavily.
None of which is an argument against AI in tendering. The first draft of a modern proposal should almost certainly be written with AI help; anyone competing without it is now at a real disadvantage. But the quality gap between “the AI wrote a comprehensive answer” and “this is a bid that will actually win” is the whole game, and nothing in the current toolkit is closing it.
In a market worth €22bn on this island, where one in four competitions still attracts only one bid or none, the problem isn’t that we need more proposals. We need a way to tell which of the ones we have are any good. Until the evaluation side catches up with the drafting side, the asymmetry will keep widening, and the companies that already had the advantage will keep extending it.
Tony Corrigan is the founder of BidReview.ai, an AI-powered platform that independently scores tender submissions against evaluators’ own criteria. He previously founded TenderScout, having started his career at IBM.
Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.



























You must be logged in to post a comment Login