AI Sales & Lead Qualification Agent
Turns free-text project enquiries into scored, classified, CRM-ready leads with an AI-drafted follow-up — and a live sales dashboard.
Built by Anya · AI Engineering Associate
The system, up close
Real captures from the running system — the demo, screenshots and architecture come first, then the deep dive.
Demo
Watch an enquiry become a scored, classified, CRM-ready lead with a drafted follow-up.
The demo shows the complete lead-qualification pipeline end to end.
Screenshots
Why this needed to be built
Sales teams drown in unstructured inbound enquiries and lose hot leads to slow, manual triage — someone has to read each form, judge fit, and decide who to chase first.
Without a consistent scoring rubric, leads get ranked by gut feel, follow-ups are generic, and there is no real-time visibility into pipeline quality for the team.
What we built, and how it changes the workflow
We built an n8n-orchestrated agent that scores and classifies every inbound enquiry the moment it lands, drafts a personalized follow-up, and feeds a live Looker Studio dashboard — no custom backend required.
A Bolt.new enquiry form posts to an n8n webhook. n8n logs the raw submission to Google Sheets, then a Google Gemini agent extracts pain and need points, computes a 0–100 lead score across four weighted criteria, classifies Hot/Warm/Cold, marks Qualified/Disqualified/Needs Review, explains its reasoning, and drafts a follow-up email.
The structured record is written back to Sheets, routed by classification, and emailed via Gmail. Looker Studio reads the same sheet live, giving the team lead volume, average score, classification mix and service-category demand with zero manual reporting.
Core design principle
- Explain every score instead of a black box — missing info is flagged, never guessed
- A fixed JSON schema constrains the model's output so downstream routing stays deterministic
- A single Google Sheet is the system of record for both storage and live reporting
What it can do
A focused set of capabilities that define the system — no marketable filler, only what actually runs.
Enquiry intake form
Public React/Tailwind form that posts structured JSON to the n8n webhook.
AI lead scoring (0–100)
Four weighted criteria — scope clarity, budget alignment, urgency and contact viability — summed into one score.
Hot / Warm / Cold classification
Tiers paired with Qualified / Disqualified / Needs Review status for routing.
Reasoning & missing-info flags
The agent explains each decision and lists what's missing rather than inventing values.
AI-drafted follow-up email
Personalized Gmail follow-up routed by classification.
Live Looker Studio dashboard
Real-time lead volume, average score, classification mix and service-demand charts.
The end-to-end workflow
The system processing path, in order.
- 1
Submit enquiry
A prospect fills the form with project details, budget and timeline.
- 2
Webhook capture
n8n receives the JSON and logs the raw submission to Google Sheets.
- 3
AI scoring
Gemini extracts pain/need points, scores and classifies the lead, and drafts an email.
- 4
Write CRM record
The structured output is written back to the LeadScore sheet.
- 5
Route & email
n8n switches on classification and sends the AI-drafted follow-up via Gmail.
- 6
Live dashboard
Looker Studio reflects the new row instantly for the sales team.
Stack & tech choices
- Intake
- Bolt.new, React, Tailwind CSS, Lucide
- Orchestration
- n8n (webhook, code, switch, respond-to-webhook)
- AI reasoning
- Google Gemini (gemini-3.5-flash-lite), LangChain Agent, Structured Output Parser
- Storage & messaging
- Google Sheets (LeadIntakeDB, LeadScore), Gmail (OAuth2)
- Reporting
- Looker Studio
Tested, verified and honestly scoped
Validation & tests
Validated against a reproducible test set covering Hot/Warm/Cold pass cases plus deliberate failure cases (missing fields, malformed input, spammy submissions).
- Scoring rubric spans scope, budget, urgency and contact viability (max 100)
- Classification tiers: Hot 80–100, Warm 50–79, Cold 0–49
- Failure cases intentionally surface missing-information flags instead of guessing
Security & hardening
- Webhook has no built-in auth — a shared-secret header is the documented public-deployment hardening
- Lead PII is stored in plain text in Sheets and sent to Gemini — flagged for data-handling review
- A consent checkbox is required before any email follow-up
Honest limitations
AI Sales & Lead Qualification Agent is an internally developed engineering prototype. We state limitations rather than overclaim production readiness.
- Relies on Google Sheets as the system of record, not a production CRM
- Per-tier email behaviour is uniform in the MVP (no Slack alert for Hot leads yet)
- No native webhook authentication without added hardening
Deep-dive references
Where this can go next
Realistic next steps we've scoped — not aspirational marketing copy.
- Tier-specific email behaviour (auto-email Hot, hold Cold for review, Slack alert for Hot)
- Native CRM integration (HubSpot/Salesforce) instead of Google Sheets
- Webhook authentication and spam protection by default
- Multi-language enquiry support
Have a similar workflow?
Zenera Labs can build a custom version for your business.
This is one of Anya's internally developed systems. Tell us about your requirement and we'll scope a tailored build around it.
Other systems we've built internally
AI Contract Analysis & Risk Detection
Faster document review with traceable evidence — upload a contract and get risk findings with source-verified citations.
naman · Engineering Portfolio
View case study
SentinelAI — Security Log Anomaly Detection
Risk-scored alerts from security events — a full-stack Mini-SIEM with ML detection, attack-chain correlation and explainable alerts.
pragna · Engineering Portfolio
View case study
AI Data Analyst & Business Intelligence Agent
Evidence-grounded business intelligence from spreadsheets — upload CSV/XLSX, get analysis, visualizations and explained answers.
priyanka · Engineering Portfolio
View case study