Turn discovery calls into
proposals ready to sign.
Proposal Agent is a desktop app that drafts branded, signature-ready proposals from your discovery calls, built on the Problem, Plan, Payoff methodology. It knows your business, learns from your wins, and plugs into your OpenClaw and Hermes agents.
runs on your machine · learns from your wins · never auto-sends
- Extract✓
- Generate✓
- Critique✓
- Render✓
Invoice Automation — Acme Corp
Read.ai webhook fires the moment the call ends.
The call ends hot.
The proposal goes cold.
Every proposal should move a deal forward, not slow it down. But most teams still rebuild them by hand: re-listening to calls, digging through transcripts, copying pricing into three places, and reformatting the SOW while the prospect’s interest quietly fades.
Proposal Agent is valuable because it knows your business: services, SKUs, pricing, proof, and real terms. It learns from every deal you win and connects to your notetakers, so calls flow in automatically. Every proposal follows your branded Problem, Plan, Payoff methodology, grounded in APMP (Association of Proposal Management Professionals) best practices: the client’s situation in their own words, a concrete plan, and the payoff before the price, with each claim traced to the call. Add it to your team, and your OpenClaw and Hermes agents can use it too. You review, approve, and send.
Connects to what you already use.
Four meeting recorders for inbound transcripts.
Six e-signature providers for the close.
Connect your own accounts. Add more via the same provider abstraction.
Secure. Reliable. Built for your workflow.
Signature verification, idempotent webhooks, and a human-in-the-loop gate on every send.
Runs on any model. Learns on yours.
An LLM drafts every proposal, and you’re never locked in. Connect an open-source model or use Anthropic, OpenAI, or Google Gemini with your own key, or run open-source and self-hosted models. Then it learns: every proposal you approve trains it on your voice, so each draft sounds more like the ones that win.
OpenAI
Anthropic
Gemini
Mistral
Ollama
Your agents draft and file. You always review, approve, and send.
Questions teams ask.
How does Proposal Agent build a proposal?
It ingests your discovery-call transcript from Read.ai, Fathom, Granola, or Fireflies, extracts the scope, outcomes, and pricing signals, then drafts a branded proposal and SOW on a proven methodology: a clear problem, the plan, the payoff, and single-source three-tier pricing. Every claim traces back to what was said on the call. You review and approve before anything goes out.
Can my other AI agents use it?
Yes. Proposal Agent runs a local server that any agent can call: OpenClaw, Hermes, or any MCP client. Your agents can ask it to draft a proposal from a client’s calls, look up a status, or record a win or loss. What they cannot do is send: a human always approves and sends. Agents draft and file, you close.
Which AI models can I use?
Any of them. Connect Anthropic, OpenAI, or Google Gemini with your own API key, or run open-source and self-hosted models like Mistral, Llama, or Ollama. You’re never locked to a single provider.
Can it learn from the proposals I actually send?
Yes. Every proposal you approve or edit trains the system on your structure, tone, and pricing, so drafts get closer to the ones that close. The more you use it, the sharper it gets.
Will it ever send a proposal without me?
Never. A human approves every proposal. There is no auto-send. You review, adjust, and hit send yourself, even when an agent drafted it.
How does pricing stay consistent across the document?
Pricing lives in one place. Every mention, from the executive summary to the investment table to the terms, pulls from that single source, so changing one number updates the entire proposal.
What do I need to get started?
Download the app for Mac or Windows, connect a meeting recorder and the model of your choice, pick a template, and run your next call. Your first draft lands minutes after you hang up. Everything runs on your machine.