AI sales coaching · call evaluation
Sales Coach listens to the call you already recorded, scores it against Sandler, talk-tracks, and your rubrics, and tells the rep what to fix next — not a generic “be more consultative” summary. Open source (MIT). Self-host free, or run hosted here.
The problem
A mid-market team records dozens of conversations a week. A manager can coach a few. The rest get a thumbs-up in Gong or nothing. Feedback that arrives days later, untied to stage and persona, does not change the next call.
Product
Same core whether you clone the repo or buy hosted. No fake feature matrix.
Upload a sales call, get a transcript, scrub audio against the text. Coaching points land on timestamps you can replay.
Run evals across providers you already pay for. Compare outputs. Bring your own API keys on self-host; we manage keys on hosted.
Score against Sandler methodology, custom rubrics, and talk-track coverage — not a single opaque “AI score.”
Feedback shaped by who is on the call and where they are in the funnel. New SDR and senior AE should not get the same note.
Managers get concrete talk tracks for the next 1:1 — what to rehearse, what to drop, what the prospect actually said.
Standalone local / Docker with almost no setup, or Clerk-backed multi-tenant for teams. Cloudflare Workers ready.
Open source vs hosted
The MIT repo is the product. Hosted prices buy managed infra, updates, seat admin, and support — not a secret better model we withhold from GitHub.
You run it. You own the data path. You wire LLM keys.
We run it. You coach. Email us to start a plan.
Pricing
USD, billed monthly. Hosted plans start with an email — no fake checkout that goes nowhere. Margins are intentional: you are paying for managed infra and support, not a different codebase.
DIY
MIT. Self-host. Bring your own API keys.
Small team
Managed cloud for a pod that needs coaching without ops.
Most teams
Room for a full sales org plus custom coaching content.
Company-wide
Unlimited seats (fair use), SSO, onboarding, SLA.
Self-host is free forever. Hosted prices buy managed infra, updates, and support — not a different product core. If your eng team already runs Docker and LLM keys, clone the repo. If you would rather coach than babysit workers, email us.
FAQ
No. Same coaching engine. Hosted is ops, seats, quotas, and support. We are not shipping a secret “pro” model only to paying customers.
Because running reliable transcription, multi-provider evals, storage, and support is expensive — and because the alternative is you spending eng time on the same stack. Self-host stays $0. Hosted is for teams that want the outcome without the ops.
Multi-provider by design. On self-host you bring your own keys (OpenAI, Anthropic, and others the app wires). On hosted we manage provider access within your plan’s evaluation quota.
Yes. Hosted Team and Enterprise include custom rubrics and talk-tracks. Self-host: configure them yourself in the app. Hosted Coach starts with standard rubrics; ask if you need a custom set early.
No for standalone / zero-config self-host. Yes if you want multi-tenant orgs — Clerk is the auth path the project already supports. Enterprise hosted includes SSO / Clerk orgs.
On our managed cloud for the duration needed to transcribe and evaluate, under our operational controls. Self-host keeps audio and transcripts wherever you deploy. Details for a specific plan: email hello@refreshqueue.com.
Email hello@refreshqueue.com with the plan in the subject (Hosted Coach, Hosted Team, or Enterprise). We confirm seats, eval volume, and onboarding. No self-serve card form yet — on purpose while we are early.