H
Howardism
Plate IIEntities中文HOWARDISM

Dan Carey

Product Manager leading product within Anthropic Labs; led Claude Design; 'Designing with Claude' talk (May 2026); ~two decades of PRDs, now replaced by prototypes

Article metadata
Publication details
Published:June 7, 2026
Filed:Entity
Domain:Entities
Tags:EntityPersonAnthropic
Reading:3 min
Source:AI-synthesised
About this piece

Articles in this journal are synthesised by AI agents from a curated wiki and are refreshed automatically as new concepts arrive. Topics, framing, and editorial direction are curated by Howardism.

Illustration for Dan Carey

Sources#

Summary#

Product Manager who leads product within Anthropic Labs. Led the Claude Design product from idea to launch (~three people, ~ten weeks). Source of the May 2026 conference talk "Designing with Claude: From prompt to production," which is the wiki's most detailed account of how a Labs bet is actually built. A PM who has "been writing [PRDs] for almost two decades," Carey now uses prototypes in their place.

Notable claims and framings#

  • Prototype over PRD. "This has effectively replaced PRDs for me… Now I just do prototypes." Talk about why, not what; record and transcribe; hand the transcript to Claude; ask for a few variations (see Prototype Over PRD).
  • Optimize the whole loop. "Why are you doing this work that Claude could do for you? Why haven't you built your own tooling?" Every optimization pays back ×(50–100 iterations per project) (see Compounding Loop Optimization).
  • Prototype the thing that almost works. "You do not want to work on the thing that already works… The next model may just fix the issues you cannot solve via engineering. We fixed [Claude Design's] with Opus 4.7 coming out. The model releases are a tide that lifts all boats" (see Build for the Next Model).
  • Lift the floor, not the ceiling. A tool should raise craft for everybody, not just add power-user controls — the lesson from ripping out Claude Design's advanced controls in a week.
  • Iterate in small enough cycles to find out you're wrong fast. "It's not necessarily can you always go fast; it's can you always iterate in a small enough cycle that you can very quickly find out when you're wrong."

Three things to try tomorrow (his closing)#

  1. Skip the PRD — talk through the why, transcribe, ask Claude for three prototype variations.
  2. Build the internal tool you've been waiting for in one afternoon — "scratch your own itch."
  3. Turn one real feature request around in 24 hours — the hardest of the three, because it surfaces every roadblock in your existing deploy/review process.

"I wouldn't try to do all three at once — layer them on."

Connections#

Sources#

§ end
Cited by 21
Related articles
  • Claude Design

    Anthropic Labs product for collaborating with Claude on polished visual artifacts — designs, prototypes, slides, decks,…

  • Claude Code

    Anthropic's agentic coding product; created by Boris Cherny late 2024; TypeScript/React on Bun (itself Claude-rewritten…

  • AI Native Product Cadence

    Cat Wu's 6mo→1mo→1day cadence at Anthropic: research-preview branding, mission-as-tiebreaker, evergreen launch room, li…

  • Prototype Over PRD

    Dan Carey's prototype-replaces-PRD method: record a why-not-what conversation, transcribe it, hand the transcript to Cl…

  • Anthropic Labs

    Anthropic's internal incubator — a 'bet factory' of ~a dozen tiny teams exploring the model frontier with lean-startup…