About Agathon
Agathon gives independent AI judgement for expensive, hard-to-reverse decisions. Its main service is AI Due Diligence for investors and acquirers who are about to put serious money behind a company whose value depends on its AI, and cannot check the AI claims themselves.
What Agathon does
AI Due Diligence for investors and acquirers
Independent technical assessment of whether the AI is worth the price, and whether the advantage will last. The assessment covers six dimensions: technical reality, defensibility and moat, scalability, team capability, data and governance, and commercial alignment. The question underneath all six is whether you are buying an asset or a story.
A written report with a severity rating on every finding, and a live debrief with your investment or operating team. You can then go ahead, re-price, set conditions or walk.
The broader practice
AI Leadership Advisory
Independent judgement for leadership teams who want a joined-up strategy for their AI and their people.
Case study: AI strategy and implementation roadmap for a major global foundation
Fractional AI CTO
Part-time AI CTO or Head of AI for a small number of companies at a time.
The one service where I make the technical decisions for you, as part of your leadership team.
Case study: Fractional CTO building patent-pending AI product from zero to beta
Independent AI Architect
An independent read on the AI tool, agent or app your team built with AI tools, once people rely on it, starting with whether it gives the right answers on your real cases. I don't build, fix or maintain it.
What makes Agathon different
No stake in the answer
Agathon does not sell implementation or remediation on anything it assesses, and has no vendor affiliation. AI Due Diligence is charged as a fixed fee. An assessor who profits from finding work has a reason to find it. My only incentive is to tell you what I find, whichever way that cuts.
In due diligence: AI-specific depth and human judgement
Generalist technical due diligence treats AI as one checkbox among many. Agathon's assessment tests the AI itself, down to exposure to foundation-model commoditisation and evaluation metrics that only work on a curated test set. Code scanning is part of it, but a tool can't tell you whether a moat will hold or a team can deliver its roadmap. Those answers come from architecture and data-room review and from interviews with the founders and CTO.
Research depth and hands-on experience
I hold a PhD in Natural Language Processing from Cambridge and read Mathematics and Computer Science at Oxford. My technical work in AI spans IBM, PA Consulting and two VC-backed AI scale-ups, and I've taken an AI product from concept to public beta, so I know where production systems break and which claims to test first.
The person you speak to does the work
There's no partner who sells and junior team who delivers. I scope and run each engagement myself, and I'll stand behind the findings.
Who uses Agathon
Agathon works for investment professionals backing or acquiring AI-led businesses. The best fit is a small investment team writing a large cheque without an internal CTO or specialist AI diligence capability.
- Partners, investment directors, principals and deal leads at small to mid-sized PE firms, VC and growth funds, family offices and co-investment vehicles
- Corporate development and M&A teams assessing an acquisition whose value depends on its AI
- Investment advisors acting for any of the above
- Private equity firms and operating partners assessing AI in companies they already own
- Exiting founders and their M&A advisors who want the AI capability verified and documented before a sale process opens
When they get in touch
- Actively evaluating an AI company, or already in due diligence
- Considering an investment or acquisition of $2m or more
- Proprietary AI is central to the valuation
- Technical claims you cannot challenge, with an investment committee or transaction deadline approaching
The broader practice
Outside due diligence, Agathon works with the people accountable for an AI decision inside their own business.
AI Leadership Advisory
Leadership teams accountable for an AI plan who want independent judgement on it
Fractional AI CTO
Seed to Series B companies and non-technical founders who need someone to own the technology decisions part-time, where AI is central to the product or the plan
Independent AI Architect
Owners, MDs and founders whose team built an AI tool, agent or app with AI tools such as Lovable, Replit, Cursor or Claude, which people now rely on and which has outgrown the way it was built
It's not a fit if you want sign-off on a decision that's already made.
The team behind Agathon
Agathon is Dr Colin Kelly, a solo, senior-led practice. For larger engagements I bring in specialists, and I keep direction and quality control.
I have a PhD in Natural Language Processing from Cambridge and 15+ years in applied AI, at IBM, PA Consulting, Adarga, Wipro and Unlikely AI.
I started Agathon because I kept meeting founders and executives asked to make AI decisions with no independent way to check them. Big consultancies sent junior staff, and senior AI expertise was priced for enterprises. Agathon now leads where that gap costs most: the point where an investor or acquirer is about to put money behind an AI claim.
The name comes from the Greek word for "excellence", and from the Athenian poet who hosted the dinner party in Plato's Symposium.
More about my background is on LinkedIn.
Career
- 2003-2006
University of Oxford: BA in Mathematics & Computer Science
Where I first encountered machine learning.
- 2006-2012
University of Cambridge: Masters and PhD in Natural Language Processing
Four years developing systems that could extract common-sense knowledge from text.
- 2012-2016
IBM: Strategic Analytics Consultant
Identified over €50m in AI value across telecoms, financial services and automotive.
- 2016-2019
PA Consulting: Analytics Principal Consultant
Led data science strategy for government, policing, and financial services clients.
- 2019-2023
Adarga: Head of Applied AI Research
Built and led the applied research team at an AI scale-up and took its proofs of concept into production. Developed entity recognition and an in-house ontology, with a benchmarking framework to evaluate them. Founded and chaired the Committee for Responsible AI.
- 2023-2024
Wipro: Managing Consultant, AI Advisory
Ran the analytics, intelligent document processing and decisioning AI workstreams on a UK retail bank's insurance transformation programme, and led Wipro's UK AI Advisory practice.
- 2024-2025
Unlikely AI: Language Engineering Manager
Led development of language reasoning systems within a neuro-symbolic AI framework.
- 2023-now
Agathon: Founder
Independent AI judgement for expensive, hard-to-reverse decisions, led by AI Due Diligence for investors and acquirers.
Key facts
| Fact | Detail |
|---|---|
| Company name | Agathon Limited |
| Type | Independent AI judgement for expensive, hard-to-reverse decisions, through an AI due diligence and advisory practice registered in England and Wales |
| Founded | 2023 |
| Founder | Dr Colin Kelly |
| Headquarters | London, UK |
| Website | agathon.ai |
| Core offering | AI Due Diligence for investors and acquirers |
| Pricing | AI Due Diligence: fixed fee, agreed before the work starts. From $15,000 for the two-week assessment. For larger mandates the fixed fee is set at a fraction of a percent of deal value. AI architecture and readiness review: $8,000 fixed, one week. Optional extras, priced separately: redesign and business case, only if the review finds the tool needs redrawing, $8,000 fixed; finding a developer, a fixed fee agreed before it starts; the monthly check, a fixed monthly fee with an end date. AI Leadership Advisory and Fractional AI CTO: priced per engagement, agreed before work starts. |
| Timing | AI Due Diligence: the core assessment takes two weeks and typically starts within one week of engagement confirmation. Larger scopes run up to six weeks. AI architecture and readiness review: one week. Redesign and business case, only if the review finds the tool needs redrawing: about one week. |
| Contract terms | Every engagement under NDA and fully confidential |
| Services | AI Due Diligence; AI Leadership Advisory; Fractional AI CTO; Independent AI Architect |
| Communication | Contact form or a booked 30-minute call. I reply to every message myself, usually within one business day. |
| Notable clients | Cardinal40 (fractional CTO) |
| Competitors | For AI Due Diligence: generalist technical due diligence firms, Big Four technology practices and automated code-scanning tools. For AI Leadership Advisory: strategy consultancies and boutique AI consultancies that also sell implementation, and hiring a Head of AI. For the independent AI architect: development agencies and freelance developers that review AI-built apps and then sell the rebuild. For Fractional AI CTO: fractional CTO networks and interim or full-time CTO hires. |
| Social | linkedin.com/in/hellocolin/ linkedin.com/company/106004277/ |
Frequently asked questions
Does Agathon sell implementation or remediation work on anything it assesses?
Agathon does not sell implementation or remediation work on anything it assesses, including after an AI due diligence assessment. Agathon has no vendor affiliation, so the assessment is the product.
Who will I work with on an Agathon AI Due Diligence assessment?
On an Agathon AI Due Diligence assessment you work directly with me, Dr Colin Kelly. Agathon is a solo, senior-led practice. For larger engagements I bring in specialists, but I run the work and stand behind the findings.
What does AI Leadership Advisory involve?
AI Leadership Advisory gives leadership teams independent judgement as they work towards a joined-up strategy for their AI and their people. Agathon's engagement starts with a stocktake of the AI your people already use and a short plan with owners and dates. Ongoing advisory follows while the plan runs. An advisory engagement typically involves 2 to 4 hours a week of direct work plus async support, with more during hiring or major technical decisions.
What is a Fractional AI CTO?
A Fractional AI CTO is a part-time CTO who owns the technology decisions, AI and beyond, with a seat on the leadership team, for companies where AI is central to the product or the plan. Agathon also offers the role scoped to the AI function, as a Fractional Head of AI. Agathon's engagements typically take 1 to 3 days a week, with a three-month minimum.
What does an Independent AI Architect service include?
Agathon's Independent AI Architect service gives an independent read on an AI tool, agent or app your team built with AI tools, once people rely on it: whether it gives the right answers on your real cases, what it can do on its own, where your data goes and whether text can make it ignore its instructions. The main step is an AI architecture and readiness review, $8,000 fixed, one week. Then you, or whoever on your team built it, make the fixes, or a developer takes it over; I don't build, fix or maintain the tool. Optional extras are priced separately: a redesign and business case ($8,000 fixed, about one week, only if the review finds the tool needs redrawing), finding a developer (a fixed fee agreed before it starts) and the monthly check (a fixed monthly fee with an end date).
Which Agathon service do I need?
The Agathon service you need depends on the decision in front of you. If you are about to invest in or acquire a company whose value depends on its AI, start with AI Due Diligence. If you lead a business deciding what to do next with AI, AI Leadership Advisory fits, and if you need someone to own the technology decisions part-time, a Fractional AI CTO does. If your team built an AI tool, agent or app with AI tools and people now rely on it, the Independent AI Architect service gives an independent read on it, starting with whether it gives the right answers on your real cases.
Talk to me before the money moves
A short confidential call will show whether an assessment is worth doing. I'll tell you quickly if it's not a fit.
Book a confidential discussionWork across the broader practice starts the same way: a short call about the decision in front of you.
Start a conversation