Turn every meeting
into shipped code.

Reverbic distills any conversation (standup, planning call, Slack thread) into a structured, context-aware spec your AI tools can execute immediately.

Solo devs free forever · No credit card required

app.reverbic.ai / meetings / sprint-planning
Reverbic

Sprint Planning · Q2 Cycle 3

42 min · Today 10:00 AM · 5 participants

Processed
Summary
Transcript
Task Distiller
Action Items
Context: NextJS API Gateway v2

Goal

Add rate-limiting middleware to the API gateway before auth endpoints go live.

Requirements

100 req/min per IP (token bucket)
Redis-backed, configurable TTL
429 + Retry-After header on limit
Skip /health and /auth/* routes

Export to

Claude

XML format

Cursor

Markdown

Codex

Sys prompt

Open in Claude

Exports directly to the tools you already use

Claude·Cursor·Codex·GitHub Copilot·Windsurf·VS Code

Three steps from
conversation to commit.

01

Capture

Record a meeting, paste a Slack thread, or upload any audio. Both voice and text feed the same pipeline.

02

Distill

AI extracts a structured task spec (goal, context, requirements, acceptance criteria), enriched with your active Context Pack.

03

Execute

One-click export formatted for Claude (XML), Cursor (Markdown), or Codex (system prompt). Or open in Claude instantly.

The full context layer

Every feature built for how AI-native teams actually work.

Task Distiller

Any conversation → structured task in seconds.

Record a standup, paste a Slack thread, or upload a meeting. Reverbic extracts the goal, requirements, and acceptance criteria, enriched automatically with your Project Context Pack. Hand it straight to Claude, Cursor, or Codex.

Feature planning · 28 min

4 participants · Today

Task Distiller

Goal

Implement OAuth 2.0 login flow with GitHub and Google providers before the beta launch.

Acceptance criteria

Both providers redirect correctly
JWT stored in httpOnly cookie
Session persists across page reloads
Error states handled gracefully
Claude
Cursor
Codex

Context Packs

2 active

NextJS API Gateway v2

ACTIVE
Next.js 14 App RouterPrisma ORM + PostgresEdge middleware authREST API conventions

React Frontend v3

shadcn/ui componentsTailwind CSS v4React QueryZustand state

Project Context Packs

Teach your AI tools once. They remember forever.

Define your stack, conventions, and constraints one time. Every distilled spec is automatically enriched. No more pasting READMEs into every prompt. Set it once, ship with context every time.

AI Tool Exports

Claude (XML), Cursor (Markdown), Codex (system prompt). One click, the right format every time.

Team Task Feed

Distilled tasks visible team-wide. Everyone ships from the same spec, with the same context.

AI Transcription

High-accuracy transcription across 50+ languages. Every word searchable and distillable.

Meeting Coach

Talk-to-listen ratio, filler words, pace. Personalized coaching per participant.

Model Context Protocol

Your meetings, in every AI tool that speaks MCP.

Reverbic ships with a native MCP server. Plug it into Claude Desktop, Cursor, or Codex and your AI can search transcripts, pull action items, and cite decisions. All without leaving the IDE.

Reverbic → Claude / Cursor

Outbound MCP server

Mint a key, drop it into your tool's MCP config, and ask Claude to pull last week's decisions or your open action items. No copy-paste, no exports.

{
  "mcpServers": {
    "reverbic": {
      "command": "npx",
      "args": [
        "-y", "mcp-remote",
        "https://reverbic.ai/api/mcp",
        "--header", "Authorization:Bearer rvb_..."
      ]
    }
  }
}
  • · list_meetings, get_meeting, search_meetings
  • · list_action_items, list_decisions
  • · Per-key scopes, instant revoke

Your stack → Reverbic

Inbound MCP integrations

Register the MCP servers your team already runs: Linear, Notion, GitHub, internal docs. Reverbic uses them while distilling so every spec lands with the right context attached.

linear-mcp.acme.comOK
notion-mcp.acme.comOK
github-mcp.acme.comOFF
  • · Bring any MCP-compliant server
  • · Org-scoped, admin-managed
  • · Smoke-tested on save
Atmos · Live thought-board

Talk through the build. Atmos opens the pull request.

As you speak, your ideas auto-cluster on a live board. Shape them into functions, tables, and endpoints — then Atmos reads your repo's conventions and existing code and ships a real, reviewable PR. Grounded in your codebase, not a guess.

LIVE BOARD
convertUnits
function
· mm → in
· lbs → kg
leads
table
· id · uuid PK
· source · varchar
POST /quote
endpoint
· req: { sku }
· res: { price }
rate limiting
idea
· per-org
· TBD
ONE CONFIRM
atmos/unit-converterOPEN PR
+src/lib/units.tscalls existing round()
~src/lib/leads.tsappend, no dup
+migrations/031_leads.sqlsnake_case
01

Capture

Recording starts, transcription streams in. A board spins up automatically.

02

Cluster

Phrases group themselves into a calm, live mind-map — no model lag.

03

Curate

Loose talk settles into typed items with fields. Your edits are locked in.

04

Commit

Generate a reuse-aware, convention-matched PR. Review, then ship to GitHub.

Other tools transcribe.
Reverbic contextualizes.

Every conversation becomes a structured, project-context-aware task your AI coding agent can execute immediately.

Other tools
Reverbic
Output
Transcript + summary
Structured task spec: goal, requirements, acceptance criteria
Context
None
Project Context Pack applied automatically
AI tools
Copy-paste manually each time
One-click export in the right format
MCP
Locked-in dashboard
Native MCP: query meetings from Claude or Cursor directly
Team
Share the recording link
Share the executable spec
Outcome
Someone still writes the ticket
Conversation → Claude command in 30 seconds

1-click

export to Claude, Cursor, Codex

30s

meeting → task spec

50+

languages supported

$0

free forever for solo devs

Start building
with context.

Turn every conversation into an executable task. Solo devs free forever.

AES-256 at rest·TLS 1.3 in transit·Org-scoped access controls