Skip to content
Claude Code Guide

Phase 4: The "Indispensable" Prompt

This is the single most powerful question you can ask during a build:

"What would turn this from impressive to indispensable?"


Why this question works

"Impressive" is table stakes. It means the demo looks good and the core features work. But impressive projects get forgotten the day after the presentation.

Indispensable means someone would be upset if you took it away. It means the tool solves a real pain point so well that going back to the old way feels broken.

The gap between impressive and indispensable is usually one or two features that nobody asked for explicitly — but everyone immediately recognizes as essential once they see them.

Try this prompt
Look at what we've built so far for [project]. It works, and the core flow is solid. Now I need you to think critically:

What would turn this from impressive to indispensable?

Consider:
- What's the most annoying friction point in the current UX?
- What data or insight is one click too far away?
- What would make a user say "I can't go back to the old way"?
- What's the feature that would make someone recommend this to a colleague unprompted?

Give me 3-5 specific, actionable suggestions ranked by impact-to-effort ratio.
Ask this after the core build is working but before you start polishing. Timing matters.

Human oversight is critical here

This is where your domain expertise and taste become the filter. Claude will generate ideas — some brilliant, some mediocre, some overengineered. Your job is to:

  1. Evaluate each suggestion against what you know about the users
  2. Pick the one or two that would genuinely change the user experience
  3. Reject the rest — even if they're technically interesting

Tip

During the zenith.chat build, the "indispensable" feature turned out to be contextual conversation memory — the ability to reference earlier parts of a conversation naturally. Nobody on the hackathon brief asked for it. The judges specifically called it out as the differentiator.

The financial expert lens

Coming from a finance background, I evaluate "indispensable" features through a specific filter:

  • Does it reduce time-to-insight? Financial professionals are paid for decisions, not for navigating dashboards.
  • Does it surface risk proactively? Don't wait for the user to find the problem — bring it to them.
  • Does it build trust? In finance, a tool that's wrong once is a tool that's never used again. Accuracy and transparency aren't features, they're prerequisites.

When to ask

The "indispensable" prompt works best at a specific moment:

  • Too early and you don't have enough context — the suggestions will be generic
  • Too late and you've already committed to an architecture that can't accommodate the insight
  • Just right is when the core flow works end-to-end but you haven't started polish

That usually means after Phase 3's stories are 60-70% complete.


Next: Browser-Based Visual QA →