Tuning AI to Win Transport Bids

Grey AI is killing your bidding

The latest thinking on artificial intelligence in the transport industry

Machine writing has been here for a while, but machine reading is arriving now. Everyone is using the same generic AI to write, meaning most bids are reading the same, but that is a problem that you can solve quickly, to get ahead of the competition

Over 100 transport company websites were checked as I was writing this. Less than 50% of them were even machine readable and only 7% had features and pages that brief AI when they visit.

Below are some practical tips from what I have seen in the market.

Preparing your business

Be ready for the disclosure question. Authorities can now ask for evidence of where AI was used on your bid. Keep an AI log across your team on every live bid. Track which tools are used, sections touched and how answers were verified.

Make your public footprint machine legible. When your company is named in a bid, any AI assisted check can only verify what it can find and read. Ensure your website is robot friendly, otherwise, your website and key pages don’t exist.

Ensure you use consistent company facts by building a canonical facts page. Simple things like a bid stating 450 vehicles and your website stating 400 vehicles is a check a human may catch, but a machine catches in seconds. Variations of company name can also hurt. Keep that canonical page up to date.

Put the facts you want to present in the machine layer. Structured data markup for your organisation and an llms.txt for AI readers. Its an hour or so of developer time and hardly anyone in transport does it (I’ve checked over 100 transport company websites)

Writing and submitting your bid

Reverse the prompts and don’t hand the AI the exact question from an ITT. Your competitors will all be doing the same. Instead, write from your own hands first, the evidence, the claims, the prose. Let the AI tool then tighten it. The answer will be much more individual and powerful.

Keep your claims atomic, with one claim per sentence and the evidence beside it and give the AI direct items to cross reference, as every public fact that can be checked strengthens your evidence chain.

Submit machine parsable documents, such as text based PDFs and never scans. Ensure headings as real headings, not just a larger font. Tables are great, so use them and don’t lock key numbers within graphics.

This list is not exhaustive and we are always happy to give away deeper detail. Get in touch if you want the full list

Benjamin Ross · 10+ years a commercial lead in public-transport technology · September 2026

For the fundamentals underneath all of this, read our guide to winning public transport tenders. For the buyer’s side of the same shift, read preparing for AI on the buying side.