Personal LLM tooling is the next IDE-shaped category

The claim

Every developer who writes prompts daily will own a personal prompt library the same way they own a personal IDE configuration — and the catalog layer wins on speed of recall, not feature count.

The evidence

GitHub Copilot reached 1.3M paid seats by late 2023 (GitHub disclosure, Nov 2023). Cursor passed $100M ARR in 2024 with a single-developer pricing model (Cursor public memo). Anthropic’s Claude API issued ≥4B tokens/day by mid-2025 (Anthropic earnings note). Each of those numbers is a developer running prompts at desk-rate cadence. None of those tools index the developer’s own prompt corpus — they generate, but they do not curate.

The unmet job is curation: the daily user accumulates 10–100 working prompts across six months and currently keeps them in a scratch directory or a notes app. Both lose at lookup speed once the corpus crosses ~30 entries.

The comparable

Closest category prior: the developer’s .dotfiles repo. GitHub hosts ≥500k public dotfiles repositories (GitHub search, 2024). That is the existence proof that single-user-shaped tools generate durable engagement when the ergonomics match the workflow. EVA is dotfiles for prompts.

What we ask for

A pre-seed conversation framed around “prompt-curation tooling for the daily LLM user,” not “yet another prompt marketplace.” We are not asking for a marketplace thesis — we are asking for a tools thesis with a single-user wedge.

timeline
  title Developer-tooling categories that started single-user
  2008 : GitHub : single-user repos before teams
  2015 : VS Code : single-user editor before Live Share
  2021 : Copilot : single-user completions before Copilot Workspace
  2024 : Cursor : single-user IDE at $100M ARR
  2026 : Personal prompt catalogs : the next single-user wedge