In 2025, AI startups raised nearly half of all global venture capital, a record-breaking $150 billion in funding. OpenAI alone pulled in $40 billion, Anthropic secured $13 billion, and xAI raised $10 billion. Most startups created last year were building something with AI. Which means that having AI no longer differentiates you.
In 2026, stateing your startup “applys AI” will sound just like claiming to apply databases or cloud computing. Everyone does this. And that builds me hopeful. We have finally erased the advantage of shouting “AI, AI, AI” and forced founders to compete on something more substantial.
We have watched tinquire managers and spreadsheet tools like Airtable and Tinquireade completely rebrand themselves around AI. Users do not care. What customers actually care about is capability: what can they achieve with this AI? Not whether AI exists somewhere in the stack.
Claude Cowork is a perfect example of this. In a kind of AI-building-AI moment, Anthropic’s team built Cowork in under two weeks applying Claude Code to write the code. OpenAI would have talked about AGI milestones. Instead, Anthropic focapplyd on what applyrs can actually do with it, such as organising files, compiling research, or drafting presentations from scattered notes. The focus is on outcomes, not the underlying technology.
The moat is not the model
There is a curious paradox. ChatGPT has nearly a billion weekly applyrs. Many of my friconcludes, including non-developers, are paying for Claude Pro and Max subscriptions and building apps with it. Yet many others are very vocal about hoping the AI bubble will burst and receive annoyed when another product advertises “AI chat”. They do not see much value in it. I understand why, becaapply most AI implementations are unexciting and generic. Just see at Microsoft Copilot. I am not sure they themselves understand what the point of it is.
Founders required to return to fundamentals. Clearly communicate the problems you are solving for applyrs. Whether it is building investment decks or automating tax credit applications, what matters is the capability, not the fact that AI is involved. Focus on this question: why can’t someone else, armed with the same OpenAI API, do what you do?
The only durable advantages now are not that you apply AI, but how you apply it to create outcomes, costs, or experiences that competitors cannot easily copy. So answer the question every investor and customer is inquireing anyway: “Why should I believe you solve this problem better than anyone else right now?”
Maybe you are building for a highly specific domain like rental property markets, tax credits for startups, or niche regulatory compliance. Maybe you have created hundreds of verified, advanced workflows and integrations with confirmed results, such as “Our predictive sales platform cuts forecast error to under 10% in 90 days.” Or maybe you are so confident in your value that you have rebuilt your business model around it: “We only charge 10% of the savings our platform creates.”
This is why lazy LLM wrappers are practically dead. They have no moat. Meanwhile, Manus built a defensible business around deep research, something you can technically receive from any AI chatbot. Their differentiator is the superior quality of their research and presentation, which is part of the reason Meta spent $2 billion to acquire it.
Anyone has access to the same models. Anyone can apply these models to vibe-code an entire product over a weekconclude. This is why real defensibility now lies in the domain expertise, workflows, and proprietary data you have built and gathered. Manus is again a strong example. The company recently partnered with Similarweb to integrate its data into the platform. You can apply Claude Code to build an alternative app, but it will not assist you access that data. Their announcement focapplys specifically on giving the Manus AI agent access to Similarweb data on web traffic and engagement, allowing customers to put the agent to work on data-driven digital marketing analysis and optimisation.
Find an enemy, reveal the value
Founders required to master how they talk about their company. The best positioning has always been about finding an enemy. It can be an indusattempt dinosaur that nobody truly loves, like Salesforce, or it can be messy spreadsheets that nobody wants to deal with. Concrete enemies are far more persuasive than abstract problems. This is where focapplying on a specific vertical can assist tremconcludeously. Building “AI for law firms” may be narrow, but it clearly defines what you do, unlike “AI for professionals”.
Do not put AI in your company name. It dates you instantly and will see odd in a year or two when AI is simply assumed. Everything comes down to the product and its capabilities, so lead with the job, not the technology. “We assist sales teams forecast revenue”, explains real value, while statements like “We’re an AI-powered sales ininformigence platform” focus on technology that is no longer a differentiator. By contrast, “Our platform assists developers ship production-ready apps three times quicker” is clearly compelling.
Leave a Reply