Relay
ProductHow it worksDocs
Book a demo

Relay — a fictional CRM product

Inbound that routes itself.

Relay qualifies, scores, and routes every inbound lead — so your team only sees the ones worth their time.

Inbound from acme.io · Mid-market SaaS

00

COMPANY FIT0 / 30
INDUSTRY FIT0 / 25
INTENT CLARITY0 / 25
BUDGET SIGNAL0 / 20
TOTAL00 / 100

Classified 2026-10-10 14:32 MYT · Ref LF-20261010-0087

Try the live demo

Pick a demo below · See the score breakdown in admin

What Relay does with one form fill

  • 01AI enrichmentEvery submission is classified — company size, industry, intent, budget signal — seconds after it arrives.
  • 02Transparent scoringFixed rubric, fixed weights. The score is deterministic given the extraction — no black-box 78/100.
  • 03Smart routingHot leads surface to sales instantly, warm ones queue for review, spam archives itself without contact.

Get a demo — this form is live

Signal

AI classifies, code scores.

The AI returns discrete categories. A pure function maps them to points with fixed weights. Every score is deterministic given the extraction.

Audit

Sub-scores you can audit.

Company fit, industry fit, intent clarity, budget signal — each weighted, each visible in the admin.

Route

Routing that reflects intent.

70+ auto-reply. 30–69 queued for review. Below 30 archived. No black box.

Trace

Every decision logged.

Timestamped, reviewed, and traceable from lead to action.

How it works

01

Lead submits

A two-minute form. They get a reference code instantly — no waiting on AI.

02

AI classifies

Groq extracts discrete categories, the rubric maps them to points, routing decides the lane.

03

Your team acts

Sales sees hot leads with summaries and scores. Everything else is queued or archived.

The pipeline

Five stages. One second.

01 / INGEST

The form arrives.

Name, email, company, and message are captured. The row is inserted with status "pending" — the visitor sees a reference code in under 300ms.

name: "Maya Chen"

email: "maya@acme.io"

company: "Acme"

message: "Need a demo…"

02 / EXTRACT

The AI reads the message.

Groq classifies company size, industry, intent, and budget signal. Structured output — no free-text guessing.

name: "Maya Chen"

email: "maya@acme.io"

company: "Acme"

message: "Need a demo…"

company_size: "mid"

industry: "saas"

intent: "demo_request"

budget_signal: "implied"

03 / SCORE

The rubric scores it.

Four weighted sub-scores. Fixed math. No black box — the total is deterministic given the extraction.

name: "Maya Chen"

email: "maya@acme.io"

company: "Acme"

message: "Need a demo…"

company_size: "mid"

industry: "saas"

intent: "demo_request"

budget_signal: "implied"

company_fit
22/30
industry_fit
25/25
intent_clarity
20/25
budget_signal
11/20

04 / ROUTE

The pipeline decides.

70 or higher: auto-reply and notify sales. 30-69: queue for human review. Below 30: archive.

name: "Maya Chen"

email: "maya@acme.io"

company: "Acme"

message: "Need a demo…"

company_size: "mid"

industry: "saas"

intent: "demo_request"

budget_signal: "implied"

company_fit
22/30
industry_fit
25/25
intent_clarity
20/25
budget_signal
11/20
auto_replyqueuedarchive

05 / ACT

Your team moves.

Sales sees a summary and a score. Every decision is logged. The right lead gets the right response.

name: "Maya Chen"

email: "maya@acme.io"

company: "Acme"

message: "Need a demo…"

company_size: "mid"

industry: "saas"

intent: "demo_request"

budget_signal: "implied"

company_fit
22/30
industry_fit
25/25
intent_clarity
20/25
budget_signal
11/20
auto_replyqueuedarchive

14:32:01 → classified

14:32:01 → scored 78

14:32:01 → routed queued

14:32:03 → notified sales

RelayRelay is a fictional product built as a portfolio piece. Demo submissions are stored and scored for illustration.