Our thesis
A technology company is proven at the second deployment.
The first deployment proves the technology works. The second proves the company does: a second site, a second business unit, a second country, done without the founders in the room. We back technical companies at that point, and we do the commercial work that gets them there: the buyers, the route in and the people who run it.
By Alessandro Singh and Rohith Varma, Last reviewed , 8-minute read
The second deployment
By a deployment we mean technology in paid use in a customer's own operation, after the pilot has ended.
There are two kinds of second. A second site with the same customer shows you can deliver again, though it can be a favour from a customer who already trusts you. A second customer who pays shows you can sell again, and it is the one an investor will ask for. We look for both, and we tell you which one you have.
This page is for you if one customer uses your technology in its own operation and the second has not happened yet, or happened only with you on site. It is also for the funds that back companies at that point.
Others call this step the move from a first customer to repeatable sales, or from one site to rollout.
Some technology spreads without a deployment at all. Software that people adopt one by one has no site to install and no second site to prove. Our test is for technology that has to be installed, integrated or approved at each customer.
Three areas
One thesis, three areas. Each has its own page: what we mean, what we look for, why now and where we could be wrong.
- Measurement, assurance and control for AI and AI agents in regulated work.
- Robots, sensors and autonomous systems in real operations.
- Quantum, new computing, materials and biology at the point they become a product someone buys.
The five tests
We test every company on the same five things, in all three areas.
- Proof in use. The technology works in a customer's own operation, and the customer keeps using it after the pilot ends.
- A buyer who pays. A named budget holder pays for it, at a price that covers the cost of delivering it.
- A way to deploy again. The company's team or its partners do the next site, business unit or country without the founders in the room, in less time and at lower cost than the first.
- A way to support it. Someone fixes it, updates it and answers the call, year after year, at a cost the price covers.
- A route through the rules. The company has mapped the approvals, certifications and procurement steps for each new market, and set aside the time and money for them.
No early company passes all five. What matters is knowing which tests it fails today, what it will take to pass them, and who pays for that work.
The evidence
The research behind the headline. Each line is one paper and what it reports.
- Buying is not using. A new information technology can be widely bought and then only sparsely deployed. The researchers who measured the difference called it the assimilation gap (Fichman and Kemerer, Information Systems Research, 1999).Source 1
- Know-how stays where it was made. The information needed to solve a problem is often costly to move, so the solving tends to happen where that information is held (von Hippel, Management Science, 1994).Source 2
- Results shrink when they travel. A result from a first trial can fade at scale, because the people and the situation change (Al-Ubaydli, List and Suskind, International Economic Review, 2020).Source 3
- Most of the pilots stayed pilots. In Uganda in 2008 and 2009, about 23 of 36 mobile health projects did not move beyond the pilot, as reported in a peer-reviewed essay (Tomlinson and others, PLOS Medicine, 2013).Source 4
- Complexity stops the spread. In a study of six health technology programmes over up to three years, those that were complex in several ways at once rarely, if ever, became routine (Greenhalgh and others, Journal of Medical Internet Research, 2017).Source 5
- Keeping it running costs more than it looks. Machine learning systems commonly carry massive continuing maintenance costs (Sculley and others, NeurIPS, 2015).Source 6
- Selling is learned before it scales. A company sells efficiently only after it has learned how its customers buy and use the product. Hiring ahead of that learning spends cash without the revenue (Leslie and Holloway, Harvard Business Review, 2006, a practitioners' article).Source 7
Only the last of these is about startups, and none proves that two is the number. Together they say why a first success does not spread by itself: the know-how stays with the first team, the conditions change, and the cost of keeping it running arrives later. The second deployment is the first place all three show up.
Where this could be wrong
Two is where the proof starts, and it does not end there. In 2023 the US Department of Energy said repeat deployments of advanced nuclear reactors are expected to help cut their overnight capital costs by 40%. It called for committed orders for five to ten deployments of one design. So we watch the deployments after the second: each should take less time and money than the last.Source 8
Some pilots are right to stop. A pilot that ends with a clear no is a result. And a first deployment that the founders stretch to another site is not a second. So we count a second deployment only when the company does it without its founders.
Founders stay in the room for longer in some fields. In complex technical sales the founders often sell and deliver the first several deployments, and that is how the company learns. So we ask what was written down at each one, and we count as the second deployment the first one the company does without them.
Selling the technology can be the right outcome. More than 80% of European deep technology exits come through mergers and acquisitions, as counted in the 2026 European Deep Tech Report, and buyers from the US accounted for most of the value of those acquisitions. So when a sale is the better route, we say so, and the founders choose.Source 9
Some markets have one buyer, and some start with a regulator. Where one public body is the market, the first contract is the test. In food, medicine and energy, approval comes first. So we agree which proof counts before we start.
What would change our mind: if by our review in May 2027 the companies we work with still need their founders on site at the third deployment, the second is too early to count as proof, and we will say so here, with how many companies we counted.
Last reviewed We review this page every six months. When we change our view, the old view stays here, with the date and the reason.Before 12 November 2026 this page set out a thesis about machines and industrial technology alone, and said it is won at the second site. We changed three things. The test now covers three areas, because the same five tests held in each. The tests ask for more: the customer keeps using it after the pilot, the price covers the cost of delivering it, and the company does the next deployment without its founders. And "won" became "proven": the research shows that a second deployment is needed and is not enough.
What we bring
For companies, we do the commercial work, in three parts.
- The buyers. We map the buyers in your sector and your next country, and set up meetings with the people who decide.
- The route in. Procurement, local partners and compliance in regulated markets, across the UK, Europe, the Gulf, Asia-Pacific and the US.
- The people who run it. Senior leaders in sales, growth, technology, operations, market access and partnerships, part-time until your permanent team is in place.
Client resultFounder advisory, 2026
30x pipeline. A B2B company in AI-agent security
- Starting point
- $1M of pipeline, with their founders doing the selling.
- What we did
- We became their go-to-market team and built the revenue engine with them: we helped hire three salespeople, set up the sales processes and systems, coached the team and provided market access.
- The result
- $30M of pipeline and $1.6M in revenue during 2026. We closed the sales together with their founders and helped them raise their Series A.
By November 2026 their own salespeople were closing without the founders.
Two founders lead our engagements. Three strategic advisors add investing and fundraising experience, technical and industrial AI depth, and connections to buyers from Asia-Pacific and the Gulf to Europe and the Americas. Both of our 2026 advisory engagements were paid in equity, which one of our founders holds.
Our work is done when your team runs the next deployment without us. If you still need us in the room for the third, we have not done our job.
When it is time to raise, we help you raise: the story, the numbers, the data room, an investor strategy and support through the round. Introductions and transaction mandates are arranged through regulated partners.
For funds and corporate venture teams, we scout and screen companies against your thesis across the UK, Europe, the Gulf, Asia-Pacific and the US. An operator checks each one before you see it. You receive a written note on each company: what checks out, what does not, what is still open, and where it stands on the five tests. It is factual and commercial, with no recommendation to invest or pass.
For family offices, we build and run an investment programme on this thesis: research and screening of companies, operator diligence and support after the deal, with no recommendation to invest or pass. Decisions stay with you or your regulated manager. If you own operating companies, they can be the second site: we match a company to a business unit that would buy, and agree what it needs to see before a pilot.
We also invest our own money when the fit is right, and can take part of a fee in equity. Size and terms are agreed case by case, in writing. Every stake is disclosed in writing to every client involved.
If we have an interest, we say so. If it conflicts, we step back.
Newsletter
One email a month
Our newest notes, and what we are seeing in AI, machines at work and deep technology.
Sources
In the order we cite them, with the day we read each one.
1. The Illusory Diffusion of Innovation: An Examination of Assimilation Gaps (opens in a new tab)
Robert G. Fichman and Chris F. Kemerer, Information Systems Research 10(3), , Accessed , Back to the text, where source 1 is cited
Eric von Hippel, Management Science 40(4), , Accessed , Back to the text, where source 2 is cited
Omar Al-Ubaydli, John A. List and Dana Suskind, International Economic Review 61(4), , Accessed , Back to the text, where source 3 is cited
4. Scaling Up mHealth: Where Is the Evidence? (opens in a new tab)
Mark Tomlinson, Mary Jane Rotheram-Borus, Leslie Swartz and Alexander C. Tsai, PLOS Medicine (an essay; the figure is from J. Lemaire, 2011), , Accessed , Back to the text, where source 4 is cited
Trisha Greenhalgh and nine others, Journal of Medical Internet Research 19(11), , Accessed , Back to the text, where source 5 is cited
6. Hidden Technical Debt in Machine Learning Systems (opens in a new tab)
D. Sculley and nine others, Advances in Neural Information Processing Systems 28 (NIPS 2015), , Accessed , Back to the text, where source 6 is cited
7. The Sales Learning Curve (opens in a new tab)
Mark Leslie and Charles A. Holloway, Harvard Business Review (a practitioners' article, not peer-reviewed), , Accessed , Back to the text, where source 7 is cited
8. Commercializing Advanced Nuclear Reactors Explained in Five Charts (opens in a new tab)
US Department of Energy, Office of Nuclear Energy, , Accessed , Back to the text, where source 8 is cited
9. The European Deep Tech Report 2026 (opens in a new tab)
Dealroom, with Lakestar, Walden Catalyst and Hello Tomorrow (a report by a data provider and investors; the exit figures are in its executive summary, page 3 of the full report), , Accessed , Back to the text, where source 9 is cited
10. The 2026 European Deeptech Report: Sector reaches $690B as VC share hits record (opens in a new tab)
tech.eu (press report of the same report), , Accessed , Back to the text, where source 10 is cited