B2B SaaS · AI outbound/Case study
ColdBind
The all-in-one autonomous AI outbound sales engine.
Live at coldbind.com- Sprint duration
- 3–4 weeks
- Sprint type
- SaaS
- Core modules
- 8
- Auth
- Built from scratch
The ColdBind landing page: autonomous AI SDR platform
01
Overview
ColdBind is an autonomous outbound sales engine: find leads with AI, enrich them, run multi-step cold email sequences, and turn replies into a deal pipeline. The founder came with deep domain knowledge of outbound and a full product vision. We ran a SaaS Launch Sprint: architecture first, then the platform module by module across 3–4 weeks, with the founder reviewing each milestone.
02
The challenge
Outbound is a compliance-sensitive domain wearing a product hat. Deliverability depends on mailbox rotation, send caps, warmup windows, and CAN-SPAM requirements. The auth model needed full control rather than an off-the-shelf provider, so sessions, credentials, and workspace isolation were built from scratch on Passport.js. And the AI layer had to do real work: ICP parsing, lead verification and fit scoring, sequence drafting, reply-intent classification, and deal recommendations, all on metered credits.
03
What we built
We built a React and Astro frontend on a Nest.js API with PostgreSQL and Redis, containerized with Docker for consistent environments. Custom Passport.js authentication handles credentials, sessions, and workspace membership. The AI layer runs on the Vercel AI SDK against Gemini and DeepSeek: the prospector turns natural-language ICP prompts into verified lead lists with fit scores, the agent executes workspace actions, and sequence drafting adapts to each step. Campaigns manage sending accounts, daily caps, sending windows, follow-up chains, and a pre-launch compliance checklist. Replies flow into a founder CRM with AI deal intelligence, and Resend handles transactional email.
04
The outcome
ColdBind is live at coldbind.com with a four-tier pricing model from a permanent free plan to an agency scale tier. The platform covers the full outbound loop: prospect, enrich, sequence, send, reply, and close, with the AI agent able to operate every module. Multi-step drip campaigns, deliverability guardrails, and the CRM shipped inside a single sprint window.
Demo
Watch the walkthrough: from ICP prompt to launched campaign
Screens
Inside the product
Shipped
What was built
- AI lead prospector from natural-language ICP prompts
- Outreach AI agent executing workspace actions
- Multi-step drip campaigns with sending windows and caps
- Mailbox rotation with deliverability guardrails
- Pre-launch CAN-SPAM compliance checklist
- Reply classification and unified inbox signal views
- Founder CRM with AI deal intelligence and pipeline board
- Custom authentication built on Passport.js
- Four-tier billing with metered AI credits
"The sprint plan was honest about what ships when, and every milestone landed. We got the whole outbound loop, auth included, in one window."
Next case study
BIP OS