AI Agent · Now Live

The sales prep
that already happened.

Closr is an AI agent for sales reps. It researches tomorrow's prospects, writes your pre-call briefs, and shows its reasoning — all before you open the app at 8 am. For setters, closers, and field reps who want the work done, not a chatbot to talk to.

Book a 10-min demo See how it works
Example brief — generated this morning
Hutchinson Builders DISCOVERY
10:00 AM
**Company Intel** — Hutchinson is Australia's largest private builder, ~1500 staff, commercial + residential. Procurement is the DM; last intel card flagged a Parramatta DA approval (15 days old). Likely pain: coordinating waste across 40+ active sites. **Recommended Opener** — "Mark, I noticed the Parramatta approval came through — how are you handling waste provisioning across the new sites while the existing ones are still drawing from the same pool?" **Discovery Questions (NEPQ)** 1. What's the biggest logistical bottleneck on your current sites this quarter? 2. How has your current provider handled the volume scaling? 3. What would it look like if you had one consolidated view across all sites? **Qualification Signals** — HIGH INTENT. Push for a same-week site walk.
Show reasoning 2 tool calls · openrouter-auto
The Morning Huddle

At 7 am your agent is already working.

While you're still in bed, Closr pulls tomorrow's schedule, researches every prospect, checks their history for signals, and writes a brief for each call. By the time you open the app at 8, every conversation is ready.

07:00

Agent wakes up.

Reads your calendar for the day. Pulls every prospect record. Queries past win, loss, and objection cards. Fires web research where intel is stale.

07:15

Writes each brief.

One brief per call. Company intel. NEPQ discovery questions. Two likely objections with reframes. A recommended opener grounded in the freshest signal. Stored with full reasoning trace.

07:30

Notifies you.

One Telegram: "4 calls today. Top priority: Hutchinson at 10am — they had a DA approval last week." You read the briefs over coffee. Walk in ready.

Why Closr isn't another chatbot

Four things a chat UI can't do.

Autonomous by default.

The agent runs on a cron, not a user click. It researches, writes, and stores briefs without a human prompt. You wake up to finished work, not an empty message box.

Real tool use, not replies.

Six live tools: fetch prospects, update records, pull call history, search the web, insert cards, schedule lookups. The agent writes to the CRM directly when you say "Hutchinson changed accountants" — not after five clarifying questions.

Visible reasoning.

Every brief shows its work. Which tools were called. Which records were read. Which web queries fired. How long each step took. If the agent got it wrong, you can see exactly where — not guess.

Voice-to-structured-data.

After a call, speak a paragraph: "Hutchinson is not ready til Q3, budget's $8K, Mark signs off but finance has veto." Closr transcribes with Whisper, extracts three structured cards (activity, intel, task), shows them for approval, saves to CRM. No form filling.

Under the hood

Built on the same stack the world's best agents run on.

No proprietary black box. Every component is auditable, provider-neutral, and production-tested.

Model layer
Claude Sonnet GPT-4.1 Gemini 2.5 via OpenRouter
Router
LiteLLM auto-routes to whichever model handles tool calling best per request, with local-model fallback
Agent loop
Vercel AI SDK v4 streaming tool calls, schema-validated args, max 5 reasoning steps per request
Data
Supabase (self-hosted) Postgres + pgvector, tenant isolation, row-level security
Voice
OpenAI Whisper iOS Safari compatible, structured output extraction for cards
Frontend
React 19 + Vite PWA offline-first, mobile-first, installable on iOS and Android

Six tools, fully auditable

Every tool the agent can call is a named function with a Zod-validated schema. Every call is logged. Every result is summarised in the brief's reasoning trace.

fetchProspects · fetchSchedule · fetchProspectHistory
updateProspect · searchWeb · insertCard

Runs on your own data

Prospects, schedule, cards — all live in your Supabase instance. No third party sees your CRM. Model providers (Claude, OpenAI) see only the prompts you choose to send, routed through LiteLLM's logged proxy.

Pricing

Straight pricing.
No seat minimums.

First 14 days free. Cancel any time. Prices in AUD.

Rep
$39 /mo
Individual setter or closer
  • → Morning Huddle daily
  • → Chat co-pilot + tool use
  • → Voice-to-cards
  • → Up to 30 calls/day
  • → Mobile PWA + offline
Most popular
Team
$99 /mo + $29/seat
Managers with 5-20 reps
  • → Everything in Rep
  • → Manager dashboard
  • → Setter→Closer handoff view
  • → Team performance analytics
  • → CRM sync (Salesforce, HubSpot)
Agency
Custom
Setter agencies running multiple clients
  • → Everything in Team
  • → White-label branding
  • → Multi-tenant (per client)
  • → Custom scoring per industry
  • → Managed onboarding

See it do its first Morning Huddle.

Ten-minute demo. We'll walk you through a live huddle, the chat co-pilot, and the voice-to-card flow. You decide if it belongs in your team's workflow.

Book a demo Open the app

Built with Claude · OpenRouter · Supabase · Vercel AI SDK · By HitCreate