In 2025, AI startups raised almost half of all international enterprise 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 final yr have been constructing one thing with AI. Which signifies that having AI now not differentiates you.
In 2026, saying your startup “makes use of AI” will sound identical to claiming to make use of databases or cloud computing. Everybody does this. And that makes me hopeful. We’ve got lastly erased the benefit of shouting “AI, AI, AI” and compelled founders to compete on one thing extra substantial.
We’ve got watched activity managers and spreadsheet instruments like Airtable and Taskade fully rebrand themselves round AI. Customers don’t care. What clients truly care about is functionality: what can they obtain with this AI? Not whether or not AI exists someplace within the stack.
Claude Cowork is an ideal instance of this. In a form of AI-building-AI second, Anthropic’s staff constructed Cowork in underneath two weeks utilizing Claude Code to jot down the code. OpenAI would have talked about AGI milestones. As an alternative, Anthropic centered on what customers can truly do with it, similar to organising information, compiling analysis, or drafting displays from scattered notes. The main focus is on outcomes, not the underlying know-how.
The moat isn’t the mannequin
There’s a curious paradox. ChatGPT has almost a billion weekly customers. Lots of my buddies, together with non-developers, are paying for Claude Professional and Max subscriptions and constructing apps with it. But many others are very vocal about hoping the AI bubble will burst and get irritated when one other product advertises “AI chat”. They don’t see a lot worth in it. I perceive why, as a result of most AI implementations are unexciting and generic. Simply take a look at Microsoft Copilot. I’m not positive they themselves perceive what the purpose of it’s.
Founders have to return to fundamentals. Clearly talk the issues you’re fixing for customers. Whether or not it’s constructing funding decks or automating tax credit score functions, what issues is the aptitude, not the truth that AI is concerned. Deal with this query: why can’t another person, armed with the identical OpenAI API, do what you do?
The one sturdy benefits now are usually not that you simply use AI, however how you utilize it to create outcomes, prices, or experiences that rivals can not simply copy. So reply the query each investor and buyer is asking anyway: “Why ought to I consider you remedy this drawback higher than anybody else proper now?”
Perhaps you’re constructing for a extremely particular area like rental property markets, tax credit for startups, or area of interest regulatory compliance. Perhaps you have got created a whole bunch of verified, superior workflows and integrations with confirmed outcomes, similar to “Our predictive gross sales platform cuts forecast error to underneath 10% in 90 days.” Or possibly you’re so assured in your worth that you’ve rebuilt your corporation mannequin round it: “We solely cost 10% of the financial savings our platform creates.”
That is why lazy LLM wrappers are virtually useless. They don’t have any moat. In the meantime, Manus constructed a defensible enterprise round deep analysis, one thing you’ll be able to technically get from any AI chatbot. Their differentiator is the superior high quality of their analysis and presentation, which is a part of the explanation Meta spent $2 billion to accumulate it.
Anybody has entry to the identical fashions. Anybody can use these fashions to vibe-code a whole product over a weekend. That is why actual defensibility now lies within the area experience, workflows, and proprietary knowledge you have got constructed and gathered. Manus is once more a robust instance. The corporate just lately partnered with Similarweb to combine its knowledge into the platform. You need to use Claude Code to construct an alternate app, nevertheless it won’t assist you to entry that knowledge. Their announcement focuses particularly on giving the Manus AI agent entry to Similarweb knowledge on net site visitors and engagement, permitting clients to place the agent to work on data-driven digital advertising evaluation and optimisation.
Discover an enemy, present the worth
Founders have to grasp how they speak about their firm. The very best positioning has at all times been about discovering an enemy. It may be an business dinosaur that no person actually loves, like Salesforce, or it may be messy spreadsheets that no person needs to take care of. Concrete enemies are way more persuasive than summary issues. That is the place specializing in a selected vertical will help tremendously. Constructing “AI for regulation companies” could also be slim, nevertheless it clearly defines what you do, not like “AI for professionals”.
Don’t put AI in your organization title. It dates you immediately and can look odd in a yr or two when AI is solely assumed. Every little thing comes all the way down to the product and its capabilities, so lead with the job, not the know-how. “We assist gross sales groups forecast income”, explains actual worth, whereas statements like “We’re an AI-powered gross sales intelligence platform” deal with know-how that’s now not a differentiator. Against this, “Our platform helps builders ship production-ready apps 3 times sooner” is clearly compelling.
