Canopy

Turn conversations into commitments.

Canopy is an AI-powered meeting-to-action platform. It transforms meeting transcripts into traceable commitments, tracks them over time, and delivers personalized audio briefings, so teams know exactly what needs to get done.

Inspiration

Meetings generate plenty of notes, but notes don't create accountability. After a meeting, teams often lose track of who agreed to do what, when it's due, and whether it was ever completed.

We built Canopy to bridge the gap between conversation and execution. Instead of another meeting summarizer, Canopy turns conversations into commitments and shows the health of a project as a living tree.

What it does

Canopy turns meeting conversations into commitments that are owned, tracked, and auditable.

Paste in a transcript and Canopy reads it the way a good chief of staff would: it pulls out every decision, who owns it, and when it's due, then renders the result as a living project tree. Each branch is a workstream and each leaf is a commitment someone made out loud. The canopy fills as work lands and thins as deadlines pass, so the state of a project is visible before anyone has to ask for a status update.

Extraction that understands context. Gemini identifies decisions, action items, owners, and deadlines. It resolves relative dates like "by Friday" against the meeting date, and leaves ownership blank rather than guessing.

Every commitment traces back to the room. Each item keeps the line of the transcript it came from, and you review every item before it's added. Anything on the tree can be checked against what was actually said.

Health you can read in a second. Leaves are green when on track, yellow as the deadline approaches, and wilting once overdue. At 3+ days late they fall to the ground, and when done they bloom.

A record, not a snapshot. Tiger Data stores every change as a timestamped event, so you can reconstruct how a deadline moved, who reassigned what, and how long a project held green.

A briefing that actually reaches people. ElevenLabs delivers a 60-second personal audio briefing each morning: what's overdue, what's due today, and the one thing that matters most.

Update it by voice. Say "Push the vendor quote to Friday" and Canopy updates the commitment, records the change, and confirms out loud.

Mood Mirror. A private, per-person view of how a meeting felt. If someone agreed to a deadline but sounded uncertain, they can choose to flag it as at-risk before it becomes late.

How we built it

Layer	Role
Gemini	Canopy's intelligence layer: turns unstructured conversation into structured commitments and understands voice requests
Tiger Data	Canopy's memory: stores timestamped events (`created`, `reassigned`, `deadline_moved`, `done`, `overdue`, and more) to reconstruct each commitment's lifecycle
ElevenLabs	Canopy's voice: turns daily priorities into natural audio briefings and spoken replies

The frontend is built with Next.js, React, and TypeScript, with a custom interactive tree visualization of project health. Mood Mirror runs as a Python (FastAPI) service.

Challenges we ran into

Turning natural conversation into reliable structure. People rarely phrase responsibilities as clean task assignments, so we designed our Gemini workflow to understand context, ownership, and deadlines, and added a human review step before anything reaches the tree.

Tracking history, not just status. We wanted Canopy to show more than a task's current state. Integrating Tiger Data let us preserve the history of each commitment and visualize how it changes over time.

Accomplishments that we're proud of

We're proud that Canopy goes beyond meeting summarization and focuses on what happens after the meeting.

We built a cohesive system where each technology has a clear purpose:

Gemini understands
Tiger Data remembers
ElevenLabs speaks
Canopy brings it all together

Most importantly, the living tree turns something normally hidden in task lists into an intuitive, visual picture of team accountability.

What we learned

Extracting information with AI is only the first step. For AI to be useful in team productivity, its output has to be structured, traceable, and actionable.

We also gained hands-on experience with generative AI, time-series data, text-to-speech, and integrating multiple APIs into a single product.

What's next for Canopy

We want Canopy to become an intelligent accountability layer across the tools teams already use:

Integrations with Slack, Teams, Zoom, and Google Meet
Commute Mode for completely hands-free daily briefings
AI insights that spot recurring bottlenecks and overdue patterns
An organization-wide forest, growing today's Grove of project trees into a view across every team