🧠 The Clock's Ticking on Fable 5

Anthropic just rolled out a fresh batch of admin tools for Claude Enterprise, and the theme is "stop guessing what you're paying for." The admin analytics dashboard now shows usage and cost by group and by user, with output like artifacts created, files edited, skills and connectors used displayed directly next to their cost, and admins can filter by their existing org chart instead of building a new one.

Claude Code specifically gets two new tabs. One tracks active developers, session counts, and top commands across the org, updated daily, and the other estimates productivity lift, cost per commit, and annual value, with every formula visible and adjustable. There's also a plain-language "Analytics chat" that can now answer questions like "which teams doubled their Claude usage this month" and return exportable charts, plus spend-threshold alerts so nobody gets blocked mid-task without warning.

If you're running a team on Claude (even a small one), this is worth a look. Knowing which skills or workflows are actually getting reused is a much better signal than raw token counts.

Meta has entered the AI gaming space with Pocket, an app that lets people generate small, interactive apps and games using AI prompts. It's built on the team behind Gizmo, the vibe-coded gaming platform Meta acquired earlier this year, and it comes with a scrollable discovery feed where you can play gizmos other people made.

Worth watching because Gizmo, the app Meta bought to build this, already pulled 635,000 lifetime installs with 98% positive sentiment. If Meta folds that into its main apps, "type a prompt, get a mini-game" could show up in your feed a lot sooner than you'd expect.

Artifacts in Claude Code are rolling out to Pro and Max plans. Ask for an artifact, Claude writes the code, publishes it live to claude.ai, and keeps updating it in real time while it continues working. Pages are private to your account and fully self-contained, so there's no separate hosting step.

This one's genuinely useful if you build small internal tools or client-facing demos. You get a live, shareable link without touching a deploy pipeline.

Help shape Practically AI
Tell us what's working. Your feedback shapes what lands in your inbox, and it takes less than 2 minutes.

Fable 5's free ride is ending. It came back on July 1 after the export-control suspension, and it's included on Pro, Max, Team, and Enterprise plans for up to 50% of weekly usage, but only until July 7. After that, it switches to metered usage credits billed at standard API rates ($10 per million input tokens, $50 per million output).

So before that window closes, have Fable 5 write "skills" that teach Opus 4.8 (your likely fallback model) how to think through your specific projects. They can't prove it moved the needle, but the logic holds up. Fable is currently your strongest reasoner, so capturing its approach as reusable instructions before you lose free access is a smart hedge either way.

A simple end-of-session habit worth stealing: ask Claude "What are you least confident about right now?" It'll usually surface 5-7 things it didn't fully investigate, and every so often one of them is a real problem you'd have missed.

Then ask the follow-up, borrowed from Sam Altman: "What's the biggest thing I'm missing about the situation right now? What don't I realize?"

Together they catch two different blind spots: the model's and yours.

Higgsfield's cinematic VFX tools are making the rounds this week, with buzz calling the output some of the most convincing AI-generated VFX yet. Here's a simple way to try it:

Help shape Practically AI
Tell us what's working. Your feedback shapes what lands in your inbox, and it takes less than 2 minutes.