When every company starts to sound the same, investors need to look beyond language to what truly sets them apart.
One of the great ironies of venture capital is that the more differentiated a startup becomes, the harder it can be to explain.
When founders first raise capital, the conversation is often straightforward. There is an idea, a market, and a broad category into which the business fits.
The challenge comes later.
The convergence of language
As products mature, genuinely innovative companies often evolve beyond the categories that originally described them. Ironically, this can make them more valuable to customers while making them more difficult for investors to evaluate.
When I first raised capital for MiAI Law 20 months ago, the proposition was relatively easy to understand: we were building a legal AI research company to deliver lawyers perfect memory.
That was enough, and we raised $2 million in five days.
However, over the past two years, the language of legal AI research has converged. Almost every company now talks about retrieval from primary legal sources, verification, explainability, transparency, legal workflows and AI assistants. These have become the language of the industry. To someone looking at the sector from the outside, many of these companies now sound remarkably similar.
That creates an interesting problem. As the language converges, differentiation appears to disappear. In reality, it has simply moved deeper into the product.
Over the past week, I watched demonstrations from two of the world’s largest legal publishers. Listening to their presentations, I realised that an investor with no legal background could quite reasonably conclude that everyone in legal AI is building essentially the same product.
How is difference to be evaluated?
The obvious question then becomes: Why invest in the startup when the incumbents appear to be delivering the same thing? It is a perfectly reasonable question.
The answer, I think, lies in how technology markets evolve.
As markets mature, differentiation moves from the surface (the ‘marketing language’) to the architecture. It sits in the engineering decisions informed by domain expertise. Those differences are profoundly important to customers, but much harder for outsiders to see.
The architecture matters
I saw this distinction crystallise during a recent demonstration to the legal team of a government enterprise.
Initially, the discussion focused on features. We analysed the organisation’s standard contracts, answered increasingly complex legal questions and identified risks that had not previously been detected, including references to repealed legislation.
As the demonstration progressed, the lawyer stopped me and asked:
“Whose brain has gone into this platform?”
It was a remarkable question because it recognised something I had been struggling to articulate. The differentiation was no longer in the features. It was in the legal methodology engineered into the platform.
Over the past two years, experienced lawyers and barristers have regularly identified practical problems they wanted solved or suggested new capabilities. Some of those ideas became part of the platform. Others did not. Every suggestion was evaluated against a single architectural vision.
Nevertheless, the product was not built by consensus. It was built by applying a consistent legal methodology to problems that practitioners repeatedly encounter. Customers who live with a problem every day tend to look beneath the marketing language. They ask: How does it retrieve information? How does it reason? How does it verify its conclusions? How does it fit into my workflow?
Investors ask different questions. They need to assess hundreds of companies across dozens of industries. They cannot become domain experts in every field they evaluate. High-level descriptions and familiar categories become an efficient way for them to understand markets.
The way forward
This is not unique to legal AI. I suspect it happens in almost every emerging technology sector.
As markets mature, differentiation moves from the surface to the architecture. Customers eventually experience those differences. Investors, by necessity, often must evaluate them before they do.
That changes the founder’s challenge. Early on, the task is convincing people that your idea matters. Later, it is explaining why your company is fundamentally different from others who now describe themselves in similar terms.
Perhaps there is a lesson for investors as well?
Venture capital has always been about recognising potential before everyone else does. When every company appears to be saying the same thing, it is worth looking beneath the language.
The hardest startups to evaluate are not always the ones solving unfamiliar problems. Sometimes they are solving familiar problems in fundamentally unfamiliar ways.
