All 9 domains from Anthropic's "How people ask Claude for personal guidance" study (April 2026), read as a founder's demand map.

Compiled by The Diff. Source data comes from Anthropic's study of 1M claude.ai conversations (~639K unique users, ~6% seeking personal guidance), cross-referenced with Anthropic's careers page, product announcements, and the funding landscape.

How to read each domain — Share tells you demand volume. Stakes tell you willingness to pay. Sycophancy and conversation patterns tell you where the generic chatbot fails (your product spec). Anthropic's build status tells you platform risk. The startups tell you what's validated and what's still open.

The three-filter score for any sub-topic: (a) stakes high enough that people pay, (b) behavioral data shows generic AI struggling, (c) no lab building for consumers at the sub-topic level. Build where all three are yes.

1. Health & Wellness — the proven one

• Share of guidance conversations: 27% (#1)
• High-stakes rating: 81%
• Sycophancy rate: 2.9% — the lowest in the study. People get straight answers and act on them.
• Sub-topics named in the study: interpreting test results, chronic condition management, injury triage, calorie/macro tracking for body composition, sleep, fitness programming
• Anthropic's build status: Enterprise-only. Claude for Healthcare + Life Sciences target hospitals, insurers, pharma. Consumer app got health-data integration (Jan 2026) as a feature, not a vertical.
• Who's already mining it: Cal AI (~$50M run rate, 7 employees, acquired by MyFitnessPal), Grow Therapy ($3B valuation), Talkiatry ($210M raised), Slingshot AI (Ash therapy tool)
• Open wedges: lab-result interpretation between doctor visits, chronic condition companions (one condition, deep), post-injury triage + recovery tracking, women's health guidance
• Read: The proven bucket. Trust is already established (lowest sycophancy anywhere), so the wedge is depth and persistence in one sub-topic, not breadth.

2. Career — the tricky one

• Share: 26% (#2)
• High-stakes rating: high (flagged as huge yet understudied in the report)
• Sub-topics: job search, career transitions, offer evaluation, salary negotiation, interview prep
• Anthropic's build status: Cowork is live and actively staffed — but for within-job productivity. The between-job layer (finding, weighing, negotiating) has no lab product.
• Who's mining it: Teal (400K interviews landed), Jack & Jill ($20M seed, conversational AI recruitment), Global Work AI (13x growth in 10 months)