See more about FusedFrames
Recorded walkthroughs of capture, distillation and querying, plus images across the desktop app, dashboard, API and CLI.
See FusedFrames in motion
macOS app demo
A walkthrough of the FusedFrames macOS app, the menubar companion that runs on each team member's machine and captures the actions, decisions and reasoning behind their day-to-day work so your AI agents can learn from it. The tour moves through the main surfaces of the app: the connection screen where a fresh install pairs with the dashboard in a single click from a shared link; the menubar dot that holds the start and stop capture controls and shows an offline indicator when the network drops; the floating question overlay that appears at the bottom of the screen when the app spots a decision worth asking about, with a single field for the answer and a skip button when there's no time; the main window that lists every action the app has observed in chronological order with a status for each one; and the action details view that lays out the full sequence of events behind a single action alongside the question asked and the response captured.
Dashboard demo
A walkthrough of the FusedFrames dashboard, the web app where captured actions are turned into a queryable library of workflows for your AI agents. The tour moves through the main surfaces of the dashboard: the pattern libraries index where each library groups patterns extracted from a chosen set of categories, tags and applications; the library detail view with its data sources, processing mode and sortable table of distilled patterns; the pattern detail view that lays out a single workflow as an ordered sequence of actions with the decisions, conditions and alternatives noted at each step; the context graph that arranges every pattern in a library as a node connected by labelled edges marking alternatives and common sequences; the sequence graph that zooms into a single pattern and shows every branch the team has taken through it; the connections and API keys pages used to wire the library into agents over API or CLI; and the exports view for pulling a library out as JSON or Markdown.
Introducing FusedFrames
Introducing FusedFrames: a Desktop app that runs on your team's devices, observes how they work, captures the context behind their decisions and turns it all into repeatable patterns your AI agents can follow. This is exactly what AI agents have been missing, because the bottleneck to automation is no longer model intelligence, it's access to the tribal knowledge that runs your organisation. Stop waiting for a smarter model and start capturing the context your AI agents need.
Look around the product
Pattern library page in the dashboard
The detail view of a single FusedFrames pattern library in the dashboard. Cards across the top configure the data sources, categories, tags and processing mode (off, automatic or review) the extraction pipeline runs against, and the patterns tab below lists every workflow that pipeline has distilled out of the team's captured actions, with sortable columns for actions, connections, applications and last seen.
Pattern detail page in the dashboard
A single distilled pattern open in the FusedFrames dashboard, with the category and tags assigned during distillation and the structured What and why card filled in. The Behaviour field describes what the team consistently does when this pattern fires and the Reasoning field describes the rationale behind it, so an AI agent reading the pattern at runtime acts the way the team would.
Question window in the Desktop app
The FusedFrames Desktop app's question window floating beside Stripe in the background, asking one AI-generated question about a recent captured action and capturing the user's one-sentence reasoning. The window is non-activating so it does not steal focus from the user's work, Esc dismisses the prompt without responding and the up arrow sends the answer back to the team's pattern libraries.
Context graph in the dashboard
The context graph view in the FusedFrames dashboard, with every pattern in the selected library laid out as a node and connected by labelled edges marking where one pattern is an alternative to another or often follows another. Filters at the top scope the graph by pattern library, category, tag, application and relation, and a mini-map in the bottom corner keeps you oriented as the graph grows.
Querying patterns through the API
Two curl requests to api.fusedframes.com authorised with a bearer token. The first hits /search/patterns to look up patterns matching the keyword 'incident' and lists the three matches by id and title. The second hits /patterns/{id} to fetch the structured pattern with its trigger and the first three SOP steps, each with stepNumber, application and instruction. This is the same JSON an AI agent consumes at runtime.
Querying patterns through the CLI
Three commands from the FusedFrames CLI piped through jq. ff patterns get returns the standard operating procedure steps for an incident-response pattern, ff patterns evidence returns the captured events that prove it including PagerDuty, Grafana and VS Code interactions, and ff traverse walks the pattern graph one hop outward to list related patterns by title.
Sequence graph in the dashboard
The sequence graph view in the FusedFrames dashboard, where every captured action is rendered as a card with its events inside and connected to other actions in the order the user performed them. The query card on the left filters the graph by connection, status, application and event type, and any combination of filters can be saved as a reusable query.
Actions page in the dashboard
The actions page in the FusedFrames dashboard, with every distilled action captured by the team's Desktop apps in a sortable table. Each row shows the AI-generated title, the connection it came from, the status of the action (observed, responded, dismissed or partially responded) and the moment of capture, filterable by connection, status and date.
Action detail page in the dashboard
A single captured action open in the FusedFrames dashboard. The Question and response card holds the AI-generated question that asks about the user's reasoning and the one-sentence response the user gave from inside the Desktop app, and the Event timeline below lists every click and keystroke the action is built from with timestamps and the application each event happened in.
Speak to the founder
Get a demo. Watch a live capture, then an AI agent query the result.
Ask anything. Pricing, security or integrating with your stack.
No purchase obligation