AI Lead Qualification Agent for B2B SaaS

Problem

A 12-person B2B SaaS team was spending ~3 hours daily manually qualifying inbound demo requests. The SDR would look up each lead in LinkedIn, check company size and funding, write a short summary, then route to the right AE. Same steps, every time.

The bottleneck: 40% of demos were no-shows or clearly out of ICP. The team was losing sales time to ops work.

Solution

Built a fully automated qualification pipeline:

1. Inbound trigger — New form submission hits an n8n webhook
2. Enrichment — Clearbit and Apollo APIs fetch company size, funding, tech stack, and LinkedIn profile
3. AI scoring — GPT-4o receives a structured prompt with enrichment data and ICP criteria, returns a score (0–100) + reasoning + recommended next step
4. Routing — Based on score: hot leads (70+) → immediate Slack DM to AE + Airtable update; warm leads (40–70) → added to nurture sequence; cold leads (<40) → politely rejected with resource links
5. Summary card — Auto-generated lead card in Airtable with enrichment data, AI summary, and one-click "Accept" button for the AE

Result

−87% time spent on manual lead triage +34% demo-to-close rate (better ICP fit) Full ROI in 11 days

The SDR team went from 3 hours/day of copy-paste research to reviewing a pre-scored Airtable queue that takes 20 minutes.

What made it work

The key insight: instead of replacing human judgment, the agent prepares it. The AE still decides — they just get a full briefing before the call instead of winging it.

GPT-4o's structured output (json_mode) made parsing reliable. Zero hallucination issues in 3 months of production use.

Stack

• n8n (self-hosted on Railway) — orchestration
• GPT-4o — scoring and summary generation
• Clearbit + Apollo — enrichment
• Airtable — CRM-lite and ops dashboard
• Slack — hot lead alerts