David Cleevely

Serendipity doesn’t happen by accident

David Cleevely on networks, ecosystems, programmable chemistry and the architecture of innovation

Innovation is often described as a moment: a breakthrough, an invention, an inspired leap. David Cleevely sees something larger.

For him, innovation is not only about the brilliance of individuals or the novelty of ideas. It is about the conditions that make unexpected connections possible. It is about networks, meeting places, recycled expertise, patient systems and the ability to bring people from different worlds into productive contact.

This is why his recent book, Serendipity: It Doesn’t Happen by Accident, feels like more than a title. It is a philosophy of innovation. Chance matters, but chance can be prepared for. Unexpected encounters cannot be fully commanded, but the conditions that make them more likely can be deliberately designed.

Few people embody this idea as strongly as Cleevely. A serial entrepreneur, investor, ecosystem builder and chairman of Chemify, he has played a central role in the Cambridge innovation story. He has been connected to organizations such as Cambridge Network, Cambridge Wireless, Cambridge Angels, Cambridge Ahead, the Raspberry Pi Foundation, the Cambridge Science Centre, and several deep technology ventures.

But the most interesting thing about David Cleevely is not simply the list of institutions around him. It is how he thinks about the invisible architecture underneath them.

Where others see companies, he sees systems. Where others see luck, he sees conditions. Where others see networking, he sees the creation of new possibility.

The architecture behind the accident

At the beginning of the conversation, Cleevely turns quickly to serendipity. He recalls unexpected encounters, the kind of meeting that seems accidental at first, but later opens a new line of thought, connection or action.

This is a recurring pattern in innovation. Someone meets someone. A conversation creates a new path. A problem in one field becomes meaningful in another. A person with unusual experience appears at the right moment. Something happens that could not have been planned in detail, yet was made more likely by the existence of a network.

This is the central tension of serendipity. It cannot be reduced to a process, but it is not pure randomness either.

Cleevely’s argument is that people, organizations, cities and countries can create environments where valuable accidents become more probable. They can increase the surface area for unexpected encounters. They can connect people who would otherwise remain separate. They can build institutions that keep ideas, capital, trust, and expertise circulating.

In this sense, serendipity becomes infrastructure. Not infrastructure in the physical sense alone, but social, intellectual, and institutional infrastructure.

Ecosystems evolve

When asked how innovation ecosystems emerge, Cleevely does not reach for a simple business framework. He reaches for evolution.

An ecosystem is not a linear machine. It is not built by adding one ingredient after another in predictable order. It is more like life itself: an evolving system in which many elements interact over time.

Ideas, capital, talent, infrastructure, role models, institutions, investors, universities, physical spaces and social norms all influence one another. The relationships are nonlinear. Small changes can have large effects. Long periods of apparent preparation can be followed by sudden bursts of complexity.

Cleevely compares this kind of emergence to the history of life on Earth, including moments like the Cambrian explosion, where complexity appeared to accelerate dramatically.

This perspective also connects to his wider systems-thinking work, including his book Systems Behaving Badly which will be out next year. The phrase captures something important about innovation: complex systems rarely behave the way simple models predict. They resist linear explanations. They produce unintended consequences. They reward people who can think in interactions, not just components.

Innovation ecosystems therefore cannot be understood only by counting startups, patents, investors, or research grants. These matter, but the deeper question is how they interact.

The system is the story.

Cambridge as a networked organism

Cleevely’s most concrete example is Cambridge. In the 1980s and early 1990s, many interesting things were already happening in and around the city. There were talented people, scientific ideas, entrepreneurs, companies and university research. But much of it was not yet sufficiently connected.

The later strength of Cambridge was not simply that it had brilliant individuals. It was that deliberate efforts were made to connect them.

Organizations such as Cambridge Network, Cambridge Wireless, Cambridge Clean Tech, Cambridge Angels, and Cambridge Ahead helped create connective tissue between academia, entrepreneurs, investors, technologists, policy actors and international partners.

This matters because innovation does not move only through formal transactions. It moves through trust. It moves through recommendations, introductions, stories, shared experiences and repeated encounters. It moves when someone who has already built something helps someone just beginning. It moves when capital is recycled from one success into new ventures. It moves when expertise, confidence, and networks are passed forward. Over time, this creates a flywheel.

A successful entrepreneur becomes an angel investor. An investor becomes a mentor. A mentor introduces a founder to a customer. A company exit funds another generation of companies. A university spin-out attracts people who later start something else. The ecosystem learns by reusing its own success.

This is why physical and social meeting places matter. Places such as Innovate Cambridge, the Glass House and earlier spaces like Library House are not peripheral. They are part of the machinery of serendipity. They reduce the friction between people. They make it easier for useful collisions to happen.

Pekka Ketola observes during the conversation that networking is not only about transferring ideas. It is about creating new ones. That distinction is crucial. A weak network passes information around. A strong network generates possibilities that no single participant could have produced alone.

The Lunar Society and the long memory of innovation

To explain the power of curated networks, Cleevely reaches back to the Lunar Society of the 18th century. The Lunar Society brought together people from manufacturing, engineering, medicine, chemistry, arts and science. Its members did not belong to a single discipline or profession. Their value came from the interaction between different kinds of knowledge.

One of the great examples is James Watt. The improvement of the steam engine was not only a technical achievement by an isolated genius. It was supported by a network of people, ideas, skills, capital and encouragement. By improving the efficiency of earlier steam engine designs, Watt helped make steam power economically transformative, contributing to the Industrial Revolution.

The lesson is not that today’s innovators should copy the Lunar Society. The lesson is that breakthrough innovation has long depended on cross-disciplinary networks that are actively maintained.

But the story also carries a warning. Networks are fragile. The Lunar Society eventually suffered under social and political backlash. Radical ideas attract resistance. Ecosystems can be damaged by fear, distrust, and instability.

Networks therefore need care. They need curation, protection, renewal and memory. They are not self-maintaining machines.

From possibility space to programmable chemistry

The conversation then moves from ecosystems to AI and chemistry, but the underlying theme remains the same: how do we explore spaces of possibility too large for the human mind to navigate directly?

Cleevely uses the metaphor of Borges’s Library of Babel. Imagine a library containing every possible book of a fixed length. Somewhere in that library are all the great works of literature, but also every possible meaningless variation.

Chemistry has a similar structure. There is an enormous space of possible molecules. Starting from atoms and rules of combination, one can imagine an astronomical number of potential compounds. Some are impossible. Some are unstable. Some are useless. But hidden within this vast possibility space are molecules that may solve real problems.

This is where Chemify becomes fascinating. Chemify is working on programmable chemistry: using automation, AI and machine learning to help design and synthesize molecules. The ambition is not merely to speed up laboratory work. It is to make it possible to explore chemical possibility space in a fundamentally new way.

A universal chemical machine can, in principle, synthesize molecules from design instructions. AI can help identify which molecules are feasible, which pathways might produce them, and which regions of the space are worth exploring.

This does not replace scientists. It changes what scientists can do. Discovery becomes less like wandering blindly and more like navigating an enormous, rule-bound landscape with better instruments.

Here again, Cleevely’s thinking is systemic. The breakthrough is not a single molecule or machine. It is the connection between computation, chemistry, automation, human judgment and the logic of exploration.

AI as a way to move between worlds

Cleevely also speaks about AI as a way to translate problems between domains.

This is an important idea. Sometimes a problem is difficult in one domain but easier in another. Engineers and scientists have long used mathematical transformations to move problems into a form where they can be solved more easily. AI expands this ability.

He describes work by Foster Technology (sic) in Bristol, where AI was used to translate between domains such as hydraulics and electronics. By moving the problem into a different conceptual space, designing there and translating back, they were able to create a train suspension system that was significantly cheaper and better at managing vibration.

This is more than optimization. It is a new kind of intellectual mobility. AI does not merely give answers. At its best, it helps us move through possibility spaces, compare structures, find analogies, and cross boundaries that human specialists may not easily cross alone.

This connects closely with Brian Arthur’s view of technology as combinatorial evolution: new technologies are built from existing technologies, and future possibilities depend on the paths already taken. Once a society goes deep into one technological trajectory, it may become difficult to shift into another. AI may help reveal paths that were hidden by disciplinary habit.

In that sense, AI becomes not only a productivity tool but a bridge between ways of thinking.

The grasshopper mind

When asked what advice he would give young innovators, Cleevely does not begin with a business plan. He begins with the mind.

He speaks warmly about the value of a “grasshopper mind”, the kind of restless intelligence that jumps between fields, ideas, books, problems and conversations. In school, such a mind may be treated as a problem. It may look unfocused. It may resist staying within one box. But for innovation, this kind of connective intelligence can be a gift.

The grasshopper mind notices patterns across domains. It reads something in physics and sees its relevance to economics. It studies history and recognizes a modern organizational problem. It hears a technical argument and connects it to politics, culture or design.

Cleevely encourages young people to read widely: literature, Greek history, physics, philosophy, systems theory. His own discovery of Norbert Wiener’s Cybernetics as a teenager became intellectually formative. Such encounters matter because they give the mind new shapes to think with.

But breadth alone is not enough. Cleevely is equally clear about the need for depth.

To innovate seriously, one must go deep in at least one domain. It is not enough to collect labels, names, trends, or references. One must understand mechanisms. How does the thing actually work? What are the constraints? What are the feedback loops? What happens when conditions change?

He refers to Darwin, who spent years studying barnacles before writing On the Origin of Species. The point is not the barnacles themselves. The point is that deep attention to a specific system can support a much broader theory.

Innovation requires both movement and grounding. A grasshopper mind needs somewhere to land.

Against stamp collecting

One of Cleevely’s warnings is against what he calls “stamp collecting.” Knowing the names of things is not the same as understanding them. A person may know the names of birds, theorems, startup concepts or technologies and still not understand how systems behave.

This is especially relevant today, when information is abundant and fluency can be simulated. People can speak the language of innovation without understanding the mechanisms. They can reference AI, ecosystems, venture capital, deep tech, and systems thinking without seeing how the pieces interact.

Cleevely’s advice is to go beyond labels. Ask how things work. Ask what depends on what. Ask what happens if one part changes. Ask where the feedback loops are. Ask whether the explanation is mechanistic or merely descriptive.

This discipline is increasingly important in an age when AI can generate plausible language at scale. The ability to sound informed is becoming cheaper. The ability to understand remains rare.

The social life of ideas

What emerges from the conversation with David Cleevely is a view of innovation as deeply social. Ideas matter. Technologies matter. Capital matters. Talent matters. But none of them move alone. They move through people.

They move through networks of trust, shared curiosity, unexpected encounters, and institutions that make collaboration easier. They move through places where people can meet informally and through formal structures that recycle success back into the system.

This is why Cleevely’s career is so instructive. He has not only built companies. He has helped build the environments in which companies can be born.

There is a difference. A company can succeed once. An ecosystem can learn to produce many successes over time.

And if Cleevely is right, the difference often lies in whether we treat serendipity as luck or as something we can prepare for.

The designed accident

The phrase Serendipity: It Doesn’t Happen by Accident may seem paradoxical at first. But after speaking with Cleevely, its meaning becomes clearer.

The accident itself cannot be predicted. The exact conversation, introduction, discovery, or insight cannot be scheduled. But the conditions can be designed: diverse networks, porous institutions, meeting places, patient capital, intellectual breadth, deep expertise and a culture that values connection.

Innovation ecosystems do not emerge simply because a city has a university, money or ambitious founders. They emerge when these elements begin to interact in self-reinforcing ways. They emerge when people create the possibility of useful surprise.

David Cleevely’s message is therefore both practical and philosophical. If we want more breakthrough innovation, we should not only ask how to fund startups or commercialize research. We should ask how to increase the number of meaningful collisions between people, ideas, disciplines, and problems.

We should ask how to build systems that behave better. We should ask how to create environments where a grasshopper mind can find both freedom and depth. And perhaps most importantly, we should stop treating serendipity as something that merely happens to us.

The future may depend on learning how to conduct it.