8090 raises $135M for AI 'software factories' — the catch is the spec

The pitch writes itself: describe the app you want in plain English, and AI agents build it. On June 26, that pitch pulled in real money — 8090 raised a $135 million Series A led by Salesforce Ventures for its "Software Factory," a platform where people and AI agents build and modernize enterprise software together. If you run a business that's been quoted six figures for custom software, this sounds like the escape hatch. It's worth understanding what actually got funded — and where the work really lives.

What actually happened

8090, founded in 2024 by venture capitalist Chamath Palihapitiya and headquartered in Menlo Park, closed a $135 million Series A led by Salesforce Ventures, with participation from Palo Alto Networks CEO Nikesh Arora, Quora co-founder Adam D'Angelo, and others (per SiliconANGLE, BusinessWire). Palihapitiya is taking the CEO role himself.

Here's how it works, in their own framing: developers author natural-language Requirements and technical Blueprints, and AI agents convert those documents into working code — building new apps or modernizing existing ones. The company is aiming squarely at regulated industries: healthcare, insurance, aerospace, financial services, and government. Notably, the humans don't disappear. The whole design assumes people writing the specs and reviewing the output, not a prompt-and-pray button.

Why it matters for your business

Read the mechanics again: the AI writes the code, but a human writes the Requirements and the Blueprint. That's not a footnote — that's the job. Anyone who has shipped software knows the expensive failures don't come from bad syntax. They come from a spec that didn't capture how your business actually works: the edge case in your refund policy, the one report your bookkeeper needs, the integration nobody mentioned until launch week. Agents generate code fast. They do not know your business.

So the honest read on a well-funded "software factory" is this: it makes the building cheaper and faster, which is real and good. It does not make the deciding go away. Someone still has to translate "how we run" into precise requirements, then check that what came back matches reality. That translation — and the accountability for getting it right — is what you're actually paying a builder for. The code was never the hard part.

That's the work we do: we sit with how your business runs, write the spec that captures it, use AI to build fast, and own the result — instead of handing you a generated app and wishing you luck.

Sources: SiliconANGLE, BusinessWire.