Ask HN: If the models are so good, why doesn't OpenAI just start 100 companies?
TL;DR Highlight
HN answers: code is easy, knowing what to build isn't, so they sell tokens instead
Who Should Read
Developers and founders building on LLM APIs who wonder what is left for their business if AI can do everything
Core Mechanics
- No article body, only comments
- The post body is empty and only the question and comments exist.
- This is a collection of community opinions, not a claim from an article.
- The question asks why not start companies directly
- The title is the question itself.
- No separate data or figures were provided.
Community Reactions
- Writing code is the easy part, one view says
- A design-services worker said products fail from wrong requirements and poor market fit.
- They argued LLMs can write code all day but lack the agency to know what to build.
- Selling tokens is risk-free revenue, some say
- Several comments used the gold-rush shovel analogy.
- One compared it to Apple not building every app in the App Store.
- Competing with customers would drive them away, one view says
- A comment said customers might stop sharing their knowledge and regulators could act.
- Another noted that training benefits from external users feeding in content.
- An AI-run cafe experiment was linked
- Andon Labs put a Gemini-powered agent named Mona in charge of a cafe in Stockholm.
- The articles mention questionable decisions like ordering thousands of rubber gloves.
- Focus and resource limits, some argue
- One comment said OpenAI and Anthropic are in a race for the best model, so downstream businesses make no sense.
- Another said model intelligence is jagged, so it is weak at chained skills like running a business.
How to Apply
- Audit your product's moat
- Spend 30 minutes today listing your customer relationships and domain knowledge beyond code.
- That list is what agents will find hardest to replace.
- Split what you hand to agents
- Keep requirements and market judgment with humans.
- Delegate implementation to the LLM in your workflow.
Terminology
Related Papers
OpenShell – A safe, private runtime for fleets of autonomous AI agents
OpenShell uses policy files to control agent file, network, and credential access at kernel level.
Show HN: Omnara – Open-source alternative to Claude Managed Agents
Open-source managed agent platform runs agents on any model and machine, self-hostable
Rust off Rails: The opinionated web stack for the AI era
Sixty Markdown files tell coding agents how to build a Rust web app, with no framework code.
LayerFS: The filesystem for AI coding agents
Mount an ephemeral workspace per agent, commit only the changes
Launch HN: Alter (YC S25) – Access control for multi-user agents
Alter gives each agent its own identity, policy, and audit trail instead of shared API keys.
Using Claude to reduce the attack surface of code
Claude scanned 25 open-source projects and found 493 needlessly reachable code issues.
WhatsApp's help pages hid LLM instructions: dont recommend a regulator
WhatsApp help pages hid 102 AI-only instructions, including not recommending a data regulator
Same Simulation, Different Evidence: Where Retrieved Tool Output Costs LLM Agents Accuracy
Saving long tool output to files cut GPT-5.1 accuracy by 11.3 points; naming the authoritative file restored it.