AI Agents Are Now Editing Your Figma Files

The AI magic trick got boring. The real work is happening in the plumbing, connecting AI to the tools we actually use.

AI Agents Now Have Read/Write Access to Your Figma Files

An open-source project just gave AI direct control over design, moving beyond read-only suggestions.

An open-source project just gave AI agents full read/write access to Figma files. The official Figma tool is read-only and requires a paid subscription, but this new community server is free, open-source, and lets AI directly create, edit, and arrange your designs. This isn't about generating code from a static image anymore; it's about giving an LLM a mouse and letting it work.

This is the next logical step in AI's evolution, moving beyond parlour tricks. Generating a pretty picture was a cool demo, but the real value comes from integrating AI into actual workflows. This server provides the essential, unglamorous plumbing required for AI to become a genuine collaborator, not just a suggestion engine. It bridges the gap between AI intelligence and design execution.

This matters for anyone building AI agents or automating design-to-code pipelines. Developers, front-end engineers, and designers get a powerful new way to accelerate tedious work. Instead of asking an AI for advice, you can now have it implement the changes for you, directly in the design file.

Read more →

Your New AI Co-pilot Is a Soft Skills Coach

AI is moving beyond raw output to refine how we communicate and create.

AI Studio: Your next fashion shoot is pixel-perfect.

AI Studio generates on-model photoshoots without the studio or the drama. It completely democratises high-end visuals, but also raises the competitive bar for every indie brand.

Clear for Slack: Stop getting left on read.

This app coaches you to write clearer Slack messages. It's less about fixing grammar and more about fixing the broken, async firehose we all drink from every day.

Lingua: Stop mumbling, start speaking.

Lingua provides a flight simulator for conversations so you can practice before you travel. It tackles the real problem: not knowing the words, but being afraid to actually use them.

Finally, Tools That Get the Job Done

The most interesting software today isn't for everyone; it's for someone specific.

70Lives: Slay your financial headaches.

This app handles finance for self-employed beauty professionals. It proves the future of useful software is vertical, solving a specific pain point that generic apps completely ignore.

CalPal: Your calculator just got a notebook.

CalPal blends a text editor with a calculation engine. It rightly assumes that numbers without context are useless, turning budgeting into a narrative.

AppWish: Turn your app idea into a real product.

This platform is basically Kickstarter for app ideas. It tackles the core developer risk—building something nobody wants—by letting the community validate ideas before a single line of code is written.

Quick hits

World TripPin: Your digital scratch-off map for travel bragging rights.
A dead-simple way to map your travels, winning because it's frictionless: no login, no setup, just pins.

FinancialAha! Spreadsheets: Budgeting templates that put you in control.
This privacy-first budgeting tool uses Google Sheets, a direct reaction to the data-hungry subscription apps that keep disappearing.

Timelines Time Tracking 4: Finally, a time tracker that doesn't feel like a boss is watching you.
This visual time diary for iOS is for personal use, not corporate surveillance, which is its entire appeal.

My takeaway

The era of AI novelties is over; the era of AI plumbing has begun.

We're moving past the standalone magic tricks that generate an image or a block of text. The real, unglamorous work is now about integrating these capabilities deeply into existing workflows. This is why infrastructure plays like the community Figma server are more important than the next flashy image generator.

This shift means the most valuable skill is no longer just prompting an AI for a cool result. It's about building the connectors and agents that let AI perform actual work inside the tools we already use. The next billion-dollar companies won't be selling magic; they'll be the invisible infrastructure that makes the magic useful.

Are we building AI assistants, or are we just building better plumbing for the tools we already have?

Drop me a reply. Till next time, this is Louis, and you are reading Louis.log().