Skip to main content
Submit a recording. Get back 6-pillar AI scoring, coaching feedback, action plans, and rep-level intelligence — typically in 30 to 90 seconds.

Quickstart

Submit your first analysis in 5 minutes — REST or MCP, your pick.

Install the MCP server

Drop one config block into Claude Desktop, Cursor, or Windsurf. 63 conversational tools.

​
See it in 30 seconds

curl -X POST https://parlay-api-dev-o7nogixtqq-uc.a.run.app/v1/analyses \
  -H "Authorization: Bearer pk_sandbox_..." \
  -H "Idempotency-Key: $(uuidgen)" \
  -d '{
    "org_id": "acme",
    "rep_id": "alex-smith",
    "recording_url": "mock://perfect-pitch"
  }'
Returns a queued job. Poll GET /v1/analyses/:id until status: "completed" (or use webhooks). The response includes 6-pillar scoring, per-pillar feedback, action plan, KPIs (filler words, words-per-minute, questions asked), and the full transcript.

​
What you can build

Coaching automation

Submit every rep call. Auto-tag persona + methodology. Generate weekly synthesis with action plan. Pipe into Slack, email, or your CRM.

Win/loss analysis

Tag every call with sold / lost / canceled. Correlate scoring patterns to deal outcomes. Generate org-level insights monthly.

AI-drafted playbooks

Upload top-performer call notes and methodology docs. Parlay synthesizes a publishable playbook in about 20 seconds. Push it to every future analysis.

In-app coach

Embed Scout — Parlay’s call-aware coach — in your app. Multi-turn chat against any completed analysis. Welcome questions tuned to the rep’s weakest pillars.

​
How it works

1

Submit

POST /v1/analyses with a recording URL (HTTPS audio file or pre-signed S3/GCS URL). Returns immediately with a job id.
2

Process

Parlay transcribes (Deepgram), then analyzes (Gemini 2.5 Flash) using your org’s playbook + custom prompts. Typical latency: 30 to 90 seconds for a 5-minute call.
3

Notify

Webhook fires when complete (analysis.completed). Or poll GET /v1/analyses/:id if you prefer.
4

Query

Pull scores, transcripts, feedback. Aggregate at the rep, org, or pillar level. Chat with Scout for deeper insight.

​
What you get back

A completed analysis includes structured data ready to ingest into your dashboards.

6-pillar scoring

Clarity, influence, objection handling, discovery, delivery, close — each scored 0 to 100.

Per-pillar feedback

Positive observations + specific improvement areas, citing moments in the call.

KPIs

Words-per-minute, filler-word count, questions asked, talk-listen ratio, total duration.

Action plan

Double-down on strengths, general improvement, quoted principle to remember.

Full transcript

Speaker-diarized transcript with timestamps, ready for replay or further analysis.

Rep-level rollups

Persona, methodology, coaching synthesis. Computed across many calls.

​
Pricing model

You’re billed per AI operation, not per seat. Mock-fixture URLs (mock://perfect-pitch, etc.) cost $0 — use them freely while integrating.
OperationApprox. cost
Analyze a 5-min real recording0.02to0.02 to 0.04
Assign rep persona (over N calls)0.01to0.01 to 0.05
Generate playbook draft0.01to0.01 to 0.10
Scout chat turnunder $0.01
All read endpointsfree
Sandbox keys (pk_sandbox_*) bill at $0 against test data — no real charges until you flip to a live key (pk_live_*).

​
Get started

Quickstart

5-minute integration walkthrough

Authentication

Bearer keys, sandbox vs live, idempotency

MCP install

Connect Claude, Cursor, Windsurf, Zed

First analysis

Mock fixture → real audio → webhooks

API reference

Every endpoint with try-it-now panels

Errors

Stable error codes and remediation