AI in public sector procurement: is it saving suppliers time?

  • Andrea Scott, Director Customer Solutions at Civica CGD
  • Dallán Ryan, Data Research & Content Manager
abstract london building

Key Takeaways

In this guest piece, Andrea Scott, Director of Customer Solutions at Civica CGD, argues that AI in public sector procurement is not removing bid work so much as moving it. Effort shifts from drafting to assurance as buyers adopt the same tools, tender packs grow and response timelines shrink.

  • Early productivity gains are eroding: buyers who assume suppliers use AI have less reason to simplify documents, and they may expect faster responses.
  • AI may make drafting easier for smaller suppliers, but larger suppliers with stronger review and governance capacity could still hold the advantage.
  • Stotles data: the UK public sector awarded 840 AI contracts worth over £1.8bn between January 2020 and July 2026, with 240 in 2025 alone. Generative AI, LLM and copilot contracts grew from near zero in 2023 to over 60 by 2026, with DWP as the lead buyer.
  • APMP data shows generative AI use among proposal teams rising from 34% in 2024 to 68% in 2025.

The open question is who captures the time AI saves, and whether procurement timelines still leave room for responsible human oversight.

AI was supposed to make bidding easier. For many suppliers, it's made the job harder.

That's the argument Andrea Scott, Director of Customer Solutions at Civica CGD, makes below. Andrea works with public sector suppliers every day, and she's watched the early productivity gains from AI start to slip as buyers adopt the same tools.

We wanted to test that theory against real numbers. So our research team pulled Stotles data on the UK public sector AI contract market and shared the analysis with Andrea. It covers every AI contract awarded since 2020: how the market has grown, where the money is going, and how fast generative AI tools are moving into government.

Written by Andrea Scott

I’ve been thinking a lot about the impact AI is having on public sector procurement in the UK.

At first, the story felt straightforward. AI gave suppliers a real productivity boost. It became easier to digest tender packs, map requirements, identify gaps and get to a first draft more quickly. For bid teams dealing with long documents and tight timescales, that felt like a genuine breakthrough.

So why does it feel harder again?

My view is that some of those early gains are already being eroded.

The reason is simple: customers are adopting AI too. And once that happens, behaviour starts to change. If buyers assume suppliers can use AI to interpret material quickly, there is less incentive to simplify the pack, remove duplication or make requirements easier to navigate. In practice, that can mean longer documents, more annexes and more content that still needs careful interpretation.

It can also create a new expectation that responses should come back faster. If drafting looks quicker, timelines can shrink, even though the real work of validating evidence, aligning delivery models, checking legal positions and securing approvals is still very human. Public sector procurement is not just a writing exercise. It is a governance and assurance exercise as well.

What this means in practice

For suppliers, I think the risk is that AI does not remove work so much as move it. It reduces time spent on drafting but increases the burden of assurance. Teams still need to check accuracy, validate interpretations and make sure polished language has not disguised weak substance. In some cases, they may also need to explain how AI was used, which adds another layer of effort.

There is a wider market effect too. AI may help smaller suppliers get to a response more efficiently, which should be a good thing. But if the overall burden keeps growing because customers expect more volume, more speed and more evidential depth, the advantage may still sit with larger suppliers that have stronger governance and review capacity. AI could democratise drafting while amplifying advantage for suppliers with larger assurance capacity.

The question I keep coming back to

The question that really interests me is what happens next.

Human-in-the-loop oversight is widely accepted as best practice, and for good reason. It preserves accountability, context and judgment. But in procurement, that principle depends on one thing above all: time. If customers assume AI can accelerate every stage, and suppliers respond by industrialising more of their process, the practical room for human review starts to shrink. At that point, the issue is not whether AI is being used responsibly in theory, but whether the timetable still allows responsible use in practice.

What the data seems to say

The available trend data broadly supports this view, even if it does not prove every part of it.

Dallán Ryan and the team at Stotles, provided me with data analysis, that reports that:

  • The UK Public Sector awarded 840 AI contracts between January 2020 and July 2026, totalling over £1.8bn, with 2025 being the standout year by far with 240 contracts in 2025 alone.
  • 2026 is set to surpass these records, with 175 contracts awarded to July already. Of these contracts, 7.1% are for Generative AI, LLMs or copilots, the exact type of tools most used in proposal and procurement management.
  • This type of AI tool is the largest growth area, exploding from near zero in 2023 to over 60 contracts by 2026, with Central government being the lead segment and DWP the lead buyer.

So AI is clearly moving into the mainstream.

There are signs the same shift is happening on the supplier side too.

  • DSIT reported that around one in six UK businesses had adopted at least one AI technology by early 2025, with higher uptake in larger organisations.
  • Proposal-management industry data from APMP suggests adoption has accelerated quickly: in 2024, 34% of proposal teams had tried generative AI, rising to 68% of teams using AI by 2025, with 70% of those adopters using it at least weekly.
  • That is not UK public sector supplier data specifically, but it does point to a market where AI use in bid and proposal workflows is moving rapidly from experimentation into routine practice.

My view

For me, AI in public sector procurement is no longer just a story about efficiency. It is becoming a story about escalation. As both government and suppliers adopt AI, each side adjusts to the other’s assumed capability.

The risk is that documents grow, expectations rise and timelines compress, while the need for human judgment does not disappear.

If AI were only inflating workload without creating any real value, adoption would probably be slowing down. Instead, organisations keep investing because there is still clear productivity value in it.

So there’s no question that AI creates efficiency, but who captures the time saved as expectations, assurance requirements and response pressures continue to rise?

That is why I think the near-term issue is not whether AI has a role in procurement. It clearly does. The real question is whether procurement is evolving in a way that still leaves enough space for responsible human oversight.

Stotles tracks public sector AI contract awards as they happen, along with the buyers, frameworks and spend patterns shaping the market. If you want to see where the opportunities are before your competitors do, talk to our team about what the numbers show for your sector.

Try Stotles for free