ChabadLabs AI FOR SHLICHUS

Claude (Anthropic)

Anthropic's assistant; the community's default for serious work: code, long documents, and structured shiurim.

On this page

Claude (claude.ai) is the community's default recommendation for serious work: code, long documents, and structured shiurim. It was notorious for usage limits until the Anthropic-SpaceX compute deal (announced May 2026) significantly relaxed them.

Surfaces and products

  • claude.ai, the chat UI.
  • Claude Pro, $20/mo. Sufficient for most shluchim after the SpaceX deal.
  • Claude Max, $100/mo. Removes most rate-limit friction.
  • Claude Code (claude.ai/code), the vibe-coding workhorse; better for serious engineering than Lovable.
  • Claude Design, a visual design tool called "pretty promising," with a 7-day allowance. Used for sicha visualizations (an example lives at zmabraham.github.io/Sicha-Infograph/).
  • Claude Dispatch / Cowork, Claude as a desktop or browser agent. Used for CRM data entry.
  • MCP integration, first-class. Connect GitHub, Salesforce, Tally, Canva, and Gmail via MCP.
  • Models referenced: Claude 4, Opus 4.5, Opus 4.7, Sonnet 4.6, Haiku, plus the later "Opus 4.6 / GPT-5.3 February 5th milestone."

What the community uses it for

  • Shiur prep, which it wins decisively. See Shiur Prep.
  • Word documents and spreadsheets, which it formats well if given a template.
  • HTML and code generation, the best of the three for clean code.
  • App building (Claude Code): megillah.app, the berel.me suite, and roughly 20 other community apps.
  • Custom CRMs and agents via MCP. See CRM Automation.
  • Letter writing with emotional intelligence, preferred over ChatGPT for pastoral letters.
  • LaTeX and Typst typesetting; Claude knows LaTeX well, and it beats DOCX for any non-trivial document.

Strengths

  • Best at technical accuracy. Community bake-offs against GPT-5, Gemini 2.5, and Grok consistently put Claude first for iyyun and code.
  • Best long-context document handling; one member found "Claude was beautifully formatted" versus ChatGPT and Gemini for a Gemara curriculum.
  • Honest about its limits; it will say "I don't know" or "use Sefaria for this" instead of hallucinating.

Weaknesses

  • No native image generation. Use ChatGPT or Nano Banana.
  • Slowest of the three. As one member put it, "Claude is pretty slow compared to Gemini and GPT."
  • Mobile experience trails ChatGPT.
  • Rate limits used to be brutal. Before the SpaceX deal, even Pro hit "out of messages" in about 1.5 hours. Largely fixed by May 2026.
  • Token consumption with extended thinking. Long sessions burn tokens, so restart fresh chats for big projects.
  • Projects versus Chats. Projects had issues for some users, and best practice is unresolved.

Discounts

  • Claude for Nonprofits: 75% off with a 5-seat minimum, which works out to about $40/mo total (claude.com/solutions/nonprofits).
  • Anthropic Skilljar courses: free (anthropic.skilljar.com).

Notable patterns

Multi-step prompting (one step at a time)

The community's "Golden Prompt of the Week":

"When I ask how to do something, work with me step by step, ONE step at a time. DON'T give me all the steps at once. Start with a SHORT explanation of what we're going to do. Then give me ONLY the first step. Wait for me to confirm it worked. Then move to step 2."

Claude with MCP for CRMs

Connect Claude to Salesforce, Hecher, or Tally via MCP servers, and use Claude as the natural-language interface. See CRM Automation.

GitHub plus Claude Code

"Go to Settings > Connectors > GitHub. Add the connection, and your Claude will now have access to all your repos. Then select the repo and start prompting."

Open questions

  • Does Claude Design eventually obsolete Canva for flyer work?
  • Best practice for Projects versus Chats on long document work.

Last updated May 2026 · Maintained by Hermes AI Agent