1) The AI question your bid team can’t avoid much longer

Procurement law doesn't ban AI in bids. It expects you to know what you're doing with it. Here's where the rules sit, where they're heading, and what I'm actually finding useful.

Every bid I've worked on this year has had the same conversation in some form. Sometimes in the kick-off, sometimes a quiet aside from a reviewer, sometimes a panicked Friday-afternoon message: are we using AI on this bid, and if we are, what do we say about it? Most teams have settled somewhere between "we don't talk about it" and "we assume the evaluator can't tell." However, the rules are clear enough now that you need a position, plus the people awarding contracts are getting better at asking. Here's where things sit, and what I'm finding works.

Where the UK rules sit today

Cabinet Office's headline position is simple: using AI in a bid isn't banned. PPN 017, which replaced PPN 02/24 for procurements starting on or after 24 February 2025, treats AI assistance the same way it treats using a bid writer. It's a tool. Responsibility for what lands in the submission sits with you.

What the PPN does is encourage authorities to ask. Annex B gives them example disclosure questions to drop into ITTs. They're information-only, not scored, and authorities are told not to penalise bidders who say yes. But the disclosure becomes part of the audit trail. Three things follow that bid teams need to watch:

• Higher volumes, more clarifications. Authorities are being told to expect more bids and plan for more clarification questions. Use AI to bid more, and expect to be asked more.
• Enhanced due diligence. Authorities are told to dig harder precisely because of AI's risks: factual errors, fabricated case studies, invented statistics. Post-tender clarification is getting more searching, not less.
• Keep confidential documents out of public tools. Tender documents shouldn't be uploaded into public AI tools as training data. This is the line the PPN draws hardest, and the one most likely to catch out teams using consumer ChatGPT or Gemini on live bids.

The Procurement Act 2023 doesn't single out AI, but it sits underneath all of this. The shift to "most advantageous tender," the central digital platform, the push for consistent feedback: all of it rewards the quality of your evidence and the auditability of your claims over slick prose.

Where the EU is going

The EU AI Act is the bigger structural shift. AI systems that contracting authorities use to evaluate tenders are classified as "high-risk," which brings requirements on human oversight, transparency and auditability that will change how authorities can use AI on the marking side.

For now, most UK and European authorities are nowhere near deploying AI evaluators at scale. The advice from procurement lawyers is consistent: automated marking of qualitative answers is too risky and too easy to challenge. So the picture is lopsided. Suppliers race ahead on preparation while authorities move slowly on evaluation, and that gap is where most of the live tension sits.

What is happening in the Middle East

There's no GCC equivalent of PPN 017 or the EU AI Act governing AI in bids yet. Saudi's SDAIA ethics principles and the UAE's frameworks remain guidance rather than binding law. But this is the most AI-forward government environment given Saudi has made 2026 its Year of AI, and from January the UAE seats a national AI system as an advisory member of Cabinet. Watch this space.

What I'm seeing on bids

• Evaluators are getting sharper. AI drafts have tells: bland generality, commitments with no evidence, the same sentence shapes throughout, the smoothness of one voice rather than a team. Where two bids look alike, it's collusion that gets investigated, and the supplier loses either way.
• Data protection is the real question. For anyone handling client data, pricing assumptions or confidential documents, where that information goes when you type it into a prompt is no longer theoretical. Controlled tools and managed content libraries are becoming a buying criterion.
• Retrieval is the shift that matters. The useful movement is tools that pull from your own bid library, approved case studies and tested method statements rather than writing from a blank page. The best teams treat AI as knowledge management first and a writing tool second.

Where it actually helps me, task by task

Anyway, enough theory. Here's what I'm finding useful across a live bid, and where it falls down:

• Research. Great for getting up to speed fast on a new authority, a policy, a market. The catch is that you have to check everything. I treat every fact it gives me as a claim to verify, not an answer to trust. One specific watch-out: on transport bids, AI sometimes struggles with basic facts like getting the number of stations right, so always check the detail.
• Writing. Useful for drafting case studies. Restructuring is where it struggles. Reworking a generated draft into something that actually answers the question is often slower and more frustrating than writing it myself.
• Evaluation. The easiest win, and the most mature. Scoring your own draft answers against the marking scheme is something several tools now do well. It's the task I'd point a sceptical team to first.
• Finance. I'm not the finance specialist, but I've yet to see AI add much on the pricing and financial side of a bid. If someone has cracked it, I haven't seen it, though at this pace that may well have changed by the next issue of Surbon Signal!

The real difficulty is keeping up. The tools change every few weeks, the rules shift underneath you, and you're doing all of it against a submission deadline. For a small team, that pressure is the real constraint.

Where this is heading

Three things to watch over the next 12 to 18 months:

• Disclosure goes from optional to expected. Authorities that aren't asking now will start, and some will move from information-only to scored questions about how AI use was governed.
• Volume keeps rising. The tools that let suppliers bid for more also bury authorities in poorly-targeted bids. Bid/no-bid discipline matters even more in an AI-enabled market.
• Governance becomes a buying criterion. Authorities buying AI-driven services (demand-responsive transport, scheduling, predictive maintenance) are starting to ask how you govern your own AI use. Your bid will need a convincing answer.

None of this changes what wins bids: targeted bid/no-bid calls, deep buyer knowledge, verifiable evidence, a clear win theme, lean financials. AI speeds up teams already doing those things well, and exposes the ones that aren't.

What to do now

If you take one thing from this, make it a position, not a policy document:

• Write your AI disclosure answer before an ITT asks for it. The honest version: yes, AI supported drafting, and named experts reviewed and verified everything.
• Get confidential material out of public tools. Pick one approved tool and make it the only one used on live bids.
• Treat your bid library as the asset. Approved case studies, verified social value commitments, tested method statements. That's what turns AI from generic to genuinely useful.
• Hold your bid/no-bid discipline. More capacity to bid is not a reason to bid for more.

2) Further links

✈️ Keflavík Airport launches cultural tour programme for departing passengers — Passenger Terminal Today Iceland's main airport now offers free 30 to 45 minute guided tours of local art, food, music and drink for departing passengers. A neat example of airports leaning into experience and destination branding.

⛴️ Hyke's electric passenger ferry completes 14 months of public service — electrive Norway's electric F-15 Shuttle carried over 41,000 passengers in Fredrikstad with up to 88% lower energy use than diesel. electrive's the honest read: the municipality flagged it was underutilised and demand declined.

🚆 Saudi Arabia's 'Dream of the Desert' luxury train on track for December launch — Arabian Business Saudi Arabia Railways and Italy's Arsenale Group launch the Kingdom's first ultra-luxury train this December, running north from Riyadh with 33 suites for 66 guests.

🚄 GCC Rail plans The GCC Railway, a 2,177 km line linking all six Gulf states from Kuwait to Oman, is targeted for completion by December 2030 and reported to be around half built

3) Upcoming events

🚇 23–25 June — UITP Urban Mobility Training, Dubai A UITP Academy course in Dubai, delivered through its MENA Centre for Transport Excellence with the RTA, covering the fundamentals of planning and running urban public transport.

🎓 22–24 September — UITP Sustainability Training, Madrid A three-day in-person course run by UITP Academy with its regional training centre EMT Madrid, from 22 to 24 September 2026, covering sustainability strategy, reporting and double materiality, climate risk, sustainable finance and procurement for public transport. I'm one of the trainers, so yes I'm biased, but it's focussed on the business benefits of sustainability, with strong trainers and great case studies from EMT Madrid.

About Surbon Consulting

Surbon Consulting helps public transport operators, authorities and suppliers win contracts and deliver social value, with 18 years of experience and over 40 bids across the UK, Europe and the Middle East. Want to know more? Book a call or explore articles and resources.​

You are receiving this newsletter because you subscribed via our website or LinkedIn. You may unsubscribe at any time and are always welcome back.