Visual automation you can watch run
Build a workflow as a node graph, wire your tools and AI together, then watch every step execute on the canvas in real time.
Build the workflow. Watch it run. Step in when it matters.
Most automation is a black box. You trigger it and hope. Charms is the opposite. You build the flow on a canvas, watch it run node by node with the inputs and outputs in plain view, and put a person in the loop wherever a decision needs one.
A canvas, not a config file
Drag steps onto the canvas and wire them together. Build from a library of triggers, AI steps, decisions, and connections, group steps into subflows, and let Build with AI draft a flow from a plain description when you want to start fast.
Every run, in the open
Each run streams a live log. For every step you see what went in, what came out, and what it cost. When something breaks you see exactly which step and exactly why, not a single red X at the end.
People, in the loop by design
A workflow should pause when a person should weigh in. Approval, data input, and AI review steps hold the flow for the people who need to act on it.
Six kinds of step. One shared visual language.
Every step is colour-coded by what it does, on the canvas and in the run log, so a flow reads the same way whether you built it or someone else did. Mix them freely to model the work your team actually does.
A workflow should pause when a person should weigh in.
Three kinds of step hand control back to people at the moments that need judgment. Expose any of them as a public form, and anyone inside or outside your team can take part without an account.
- Approval. Hold the flow until someone signs off.
- Data input. Collect what the flow needs from a person.
- AI review. Have an AI step check the work, then route it to a human.
The best automation is the one your whole team can build on.
Save a working flow as a template, import flows others have built, and run on a schedule or on demand.
Save as a template
Turn a working flow into a starting point the rest of the team can reuse.
Import a flow
Bring in flows others have built and adapt them to your own work.
Run on a schedule
Trigger a flow on a cadence so routine work happens without anyone starting it.
Run on demand
Kick off any flow by hand or from a public form when you need it now.
Charms reads cleanly for a non-technical teammate and runs deep enough for a platform engineer. The canvas explains itself, and the run log proves what happened.
Build the workflow. Watch it run.
Start on the canvas, then see how Charms fits the rest of the platform.
