Not every feature is worth building
The surgical rule for LLM products
Twenty years in the operating room taught me one thing: not every solution is useful. Same applies to building LLM products.
People come to you saying:
- "Let's add multimodal search"
- "Let's connect to Slack"
- "Let's do ingestion for everything"
And you ask: Why?
The Surgical Rule
When I decide to operate on a patient, I don't use every tool in the room. I choose the proven one — not every tool available.
With LLM: every new feature request = a surgical decision.
Three Questions to help you decide what to cut:
- Does this solve a real problem — not an imaginary one?
- Will users actually pay for it?
- Can I build it in a way that scales?
Three yes answers = build it.
When I built Surgical Edit, I didn't do everything. I chose bilingual content — Arabic and English — because I saw the value. I didn't chase every platform or every format.
Result: a platform that speaks directly to the GCC medical community.
Quality over quantity.
Today's Lesson
Not every feature is worth building. The best LLM builders know what not to automate.