It’s Topical
← The AI Builder Roundup archive
The AI Builder Roundup

Find the AI recommendations worth shipping

What's good, devs, if your AI keeps suggesting fixes that never get shipped, today's workflow can help you sort the list. We also have a cheaper Claude model and a way to publish apps from your existing AI tool.

In today’s edition

  • Claude Fable 5.2 costs less than its predecessor.
  • Spacefast publishes apps from your existing AI.

Claude Fable gets cheaper and faster

Anthropic’s Fable 5.2 cuts API token prices by 20% from Fable 5.1 and generates output up to 30% faster. That makes its model for ambitious technical work less expensive to use.

  • API pricing is $8 per million input tokens and $40 per million output tokens, down from $10 and $50.
  • In Anthropic’s FrontierCode 2.0 test, Fable at High effort scored about level with Opus 5.5 at Max effort for roughly half the cost per task.
  • Fable 5.2 is available on paid Claude plans, the Claude Platform, and major cloud platforms.

What it means Try Fable for an ambitious, open-ended technical project. For most daily coding and knowledge tasks, Anthropic recommends starting with Opus 5.5.

Source: claudeai

Put apps online without switching your AI

Automattic launched Spacefast, a publishing platform that works with AI tools you already use. It gives your agent a database and a place to put an app online.

  • It works with tools including ChatGPT, Claude, Cursor, and Codex. You can share a private link and have your agent turn team comments into fixes.
  • It’s free to start. Personal plans include $1 a month in usage; the $5-a-month Go plan adds unlimited spaces and more usage.
  • Built-in app authentication, transactional email, and shared team knowledge are planned for the next few weeks.

What it means Try it if you want to publish and share an app while keeping your current AI tool. Ask your AI about setting up Spacefast.

Source: www.prnewswire.com

Quick hits

GitHub repo of the day

Agent Skills: plan before your coding agent builds

  • What it does: The maintainers provide reusable workflows for turning an app idea into a spec, breaking it into tasks, and checking the result before shipping.
  • How to start: In Claude Code, run /plugin marketplace add addyosmani/agent-skills, then /plugin install agent-skills@addy-agent-skills. Use /spec to define your project before moving to /plan.

AI workflow of the day

How to find which AI recommendations to ship

Shared by Noah Kagan (@noahkagan) · View original post ↗

If your team has a backlog of AI-generated suggestions, this prompt helps you check what’s live and rank the remaining work. Noah Kagan reports using the approach to sort nine SEO fixes and identify three worth pursuing.

  1. Give Claude access to, or provide records from, the relevant calendar, docs, Slack, Stripe, and analytics sources for the last 90 days.
  2. Paste the prompt below. The dollar impact and chance of success are estimates to check, especially where the source data is missing.
Original prompt
More from The AI Builder Roundup →