The verdict in three sentences
An IT services firm bidding on 70 public tenders a year spends about 1,750 hours on written responses, more than one senior consultant's full year. A RAG-based AI assistant, which searches your 400 case studies and CVs before drafting, cuts a response from 25 to roughly 14 hours, provided a human owns the win strategy and the final review. AI does not win contracts for you: it frees the time to qualify opportunities better and sharpen the answers that actually score.
Where the 25 hours of a bid go
For a London IT services firm bidding through Find a Tender, Contracts Finder, Crown Commercial Service frameworks and borough portals, quality questions often carry 60 to 70 % of the evaluation. Since the Procurement Act 2023 came into force on 24 February 2025, buyers assess the most advantageous tender and publish more detailed assessment summaries, which raises the bar on tailored answers.
| Step | Current time | With AI assistant | AI role |
|---|---|---|---|
| Reading the ITT, specification and criteria | 3 h | 1.5 h | Requirements summary and compliance matrix |
| Bid or no-bid decision | 1 h | 0.5 h | Comparison with past wins and losses |
| Selecting case studies and CVs | 4 h | 1 h | Semantic search across 400 case studies |
| Drafting method statements | 8 h | 4.5 h | First draft from winning responses |
| Tailoring to the buyer's context | 4 h | 3.5 h | Still mostly human |
| Review, formatting, procurement forms | 5 h | 3 h | Consistency checks, forms kept separate |
| Total | 25 h | 14 h | 2026 estimate |
The clearest saving is finding the right case studies and CVs. Bid teams often spend hours locating the similar project with the right figure and the right referee.
Architecture: why RAG and why UK hosting
A language model on its own invents plausible-sounding case studies. RAG (retrieval-augmented generation) changes that: the assistant first queries your document base, winning bids, case studies, current CVs, certifications such as Cyber Essentials Plus or ISO 27001, then drafts while citing its sources. Every paragraph links back to an original document, which makes review fast.
For a firm selling to the public sector, hosting matters. CVs are personal data under UK GDPR, and some contracts require data to stay in the UK or meet specific security standards. Deploying in a UK cloud region or on your own infrastructure, with a model accessed under terms that exclude training on your data, is the safest setup.
| Option | 2026 cost estimate | Strengths | Limits |
|---|---|---|---|
| Consumer AI assistant | 18 to 25 GBP per user a month | Instant | No secure access to your library, leakage risk |
| Specialist bid SaaS | 120 to 500 GBP per user a month | Tender alerts and document analysis built in | Library hard to customise |
| Freelance bid writer | 400 to 700 GBP per day | Public procurement expertise | High cost per bid |
| Kolonell custom RAG assistant | Simple 3,810 to 6,100 EUR, medium 6,100 to 10,670 EUR, complex 10,670 to 22,870+ EUR | Your data, your hosting | Library needs structuring first |
| Kolonell maintenance | 53 / 114 / 183 EUR a month | Index and model updates | Level set by usage |
| Model API costs | 80 to 350 GBP a month for 70 bids | Usage-based | Depends on the model chosen |
See our approach to AI projects.
Limits to understand before you start
First limit: the quality of your library. If your 400 case studies are inconsistent Word files with no value, date or buyer name, the assistant will produce vague text. Plan two to three weeks of structuring, often the best investment in the whole project.
Second limit: evaluators spot generic text. A response that repeats the same wording for a borough council and an NHS trust loses marks on understanding of requirements. AI produces a first draft; the team turns it into a specific answer.
Third limit: accountability. Commitments in a tender response become contractual. No sentence about timelines, named staff or service levels should go out without human sign-off. Mandatory exclusion questions and the procurement declaration forms stay outside automated drafting.
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
Mini case study
James, sales director of a 90-consultant IT services firm in London, bids on 70 public tenders a year at 25 hours per response and an internal cost of 65 GBP an hour. Total: 1,750 hours, or 113,750 GBP a year. With the RAG assistant, each response drops to 14 hours: 980 hours, or 63,700 GBP. Time saved: 50,050 GBP a year.
He reinvests part of the 770 freed hours to bid on 10 more tenders and qualify the rest better. If his win rate rises from 18 to 20 % over 80 bids, he wins 16 contracts instead of 12.6. At an average margin of 22,000 GBP per contract, the commercial effect approaches 75,000 GBP. A medium-complexity build between 6,100 and 10,670 EUR, plus about 200 GBP a month of API costs and 114 EUR of maintenance, pays back within three months on time savings alone.
FAQ
Can a public buyer reject a bid written with AI?
The Procurement Act 2023 does not ban AI use in 2026, though some buyers now ask suppliers to declare it. The bid is judged on content, but commitments must be accurate or the contract can be terminated.
How many case studies does the assistant need?
Search becomes useful from 100 to 150 structured case studies. With 400 case studies and current CVs, the assistant finds in under a minute what used to take an hour.
Can it connect to tender alerts?
Yes, it can read notices from Find a Tender and Contracts Finder and assign a fit score. That cuts qualification time by 30 to 40 %.
Can we use consultants' CVs?
Yes, under legitimate interest with staff informed, UK hosting and a defined retention period. CVs of candidates who were not hired should be removed after a set period, often 6 to 24 months.
How long until it goes live?
Allow 8 to 12 weeks: 3 weeks of library structuring, 4 to 6 weeks of build and testing on 5 real bids, then rollout.
Let's scope your project. We define your document library, hosting and a first assistant tested on your real bids with you, delivered in weeks. Detailed quote within 48 h. WhatsApp +221 77 596 93 33.
Mohamed Bah
Fondateur, Kolonell
Passionate about digital and entrepreneurship in Africa, Mohamed has been helping Sénégalese businesses with their digital transformation since 2020. Founder of Kolonell, he believes every SME deserves a professional and accessible online présence.
