← Products
lead generation & outreach

Acquisition OS

Automated discovery, verification, scoring and outreach — from lead to qualified pipeline without manual follow-ups.

92 leads • A40 B11 C41 • recurring every 2 days

What It Is

Acquisition OS is an operating system for customer acquisition. It solves a specific problem: discovering, verifying, and reaching qualified prospects requires manual research, verification across multiple sources, scoring decisions, and approval workflows—all before a single outreach email is sent.

The system automates discovery and verification. It runs recurring research cycles (every 2 days), cross-references leads against multiple sources to eliminate duplicates, scores prospects based on brand strength and digital weakness, enriches each lead with AI-generated outreach angles, and routes candidates for human review before campaign deployment.

Acquisition OS is for founder-led companies, agencies, and service businesses that acquire customers through direct outreach. It eliminates the 80% of acquisition work that is research and verification, leaving humans to focus on strategy and relationship-building.

Lead Corpus

Acquisition OS maintains a persistent lead database. Every 2 days, it runs a discovery sweep across configured geographies, deduplicates against the existing database using fuzzy matching, and ranks all candidates (new and existing) by lead score.

The current production database contains 92 leads across three tiers:

Tier A (High quality)40 leads
Tier B (Medium quality)11 leads
Tier C (Exploratory)41 leads
Total92 leads

Generation Pipeline

Every production cycle follows this verified pipeline, running every 2 days at 09:30 UTC:

1. Discovery Sweep — Tavily + DuckDuckGo with rotating queries
2. Deduplication — Fuzzy match against SQLite lead database
3. Deep Research — Multi-source cross-check per business
4. Website Probe — Classification (none | placeholder | basic | good | excellent)
5. Scoring + Tiering — Lead score calculated, tier assigned (A/B/C)
6. AI Enrichment — GPT-4o-mini generates problem/opportunity/outreach angle
7. Persistence — Data stored in SQLite with metadata (first_discovered, previous_score, status)
8. Sheets Export — CSV + XLSX for manual review and CRM import
9. PDF Report — Formatted lead report for stakeholder review
10. Batch Record — Run metadata logged (timestamp, counts, paths, delivery status)

If any step fails, the entire run is logged as failed and no incomplete report is generated. Failure alerts are sent via Telegram.

Lead Qualification & Scoring

Scoring Model

Brand Strength (0–50):

• Rating × 4 (normalized)

• Log-scaled review count

• Signals: premium location, mall presence, multi-branch

Digital Weakness (0–50):

• No site: +46

• Aggregator-only: ~42

• Placeholder site: +41

• Basic site: ~28

• Good site: ~14

• Excellent site: ~5

Tier Assignments

Tier A

Lead Score ≥ 65

High-intent, strong fit, ready for outreach

Tier B

Lead Score ≥ 40 and < 65

Medium-fit, exploratory outreach

Tier C

Lead Score < 40

Lower priority, research-needed candidates

Score Calculation: lead_score = 0.55 × brand_strength + 0.45 × digital_weakness ± AI adjustment (±10)

Automatic Exclusions: Corporate chains and franchises are automatically excluded to focus on independent operators.

Segmentation & Filtering

Acquisition OS supports segmentation across multiple facets for targeted campaign deployment:

Geography

  • • Navi Mumbai
  • • Mumbai (Central, South, West)
  • • Pune
  • • Configurable regional zones

Business Category

  • • Cafes & restaurants
  • • Retail & boutiques
  • • Services & salons
  • • Gaming & entertainment
  • • Gyms & fitness

Export Format: Leads are exported as CSV and XLSX with all segmentation facets as filterable columns. Campaign managers can slice by tier, geography, category, lead score, or enrichment data.

Campaign Workflow

Acquisition OS automates candidate selection but requires human approval before outreach deployment. The complete workflow is:

1. DiscoveryAutomated sweep runs every 2 days, finds new candidates in configured geographies
2. QualificationDeep research, website analysis, lead scoring, tier assignment
3. EnrichmentAI generates outreach angle, problem statement, opportunity assessment
4. SelectionSystem identifies candidates matching campaign criteria (tier, geography, category)
5. Review & ApprovalHUMAN GATE: Campaign manager reviews selected leads, adjusts filters if needed, approves outreach roster
6. Outreach PrepExport approved roster as CSV/XLSX for CRM import or email platform integration
7. DeploySend cold emails via approved channels with personalized outreach angles

Key constraint: Acquisition OS does not automatically send emails. It qualifies and routes candidates. Human review happens before any outreach is deployed.

Architecture

System Components

Web Application

Vercel (Next.js) — manages UI, campaign config, approval queue

Lead Engine

Python service — handles discovery, research, scoring (runs independently)

Data Persistence

SQLite database — maintains lead corpus, run history, tier assignments

AI Integration

GPT-4o-mini via OpenAI API — enrichment only, scores adjusted per urgency

Research APIs

Tavily (primary) + DuckDuckGo (fallback) — web search and business discovery

Data Flow & Integration

Production Mode

LEAD_ENGINE_MODE=remote

Vercel app → HTTP request → Lead Engine → Database

Execution

Task Scheduler runs every 2 days at 09:30 UTC

Outputs

PDF reports, XLSX exports, CSV mirrors, batch metadata

Architecture principle: The Lead Engine runs independently from the web application. Vercel does not make direct API calls to research services or LLMs. All research logic, scoring, and enrichment happens within the isolated Python service.

This separation prevents rate-limiting issues, keeps credentials secure, and allows the system to handle long-running research tasks without web request timeouts.

Key Capabilities

Custom scrapers for discovery across configured geographies

Lead scoring and tiering (A/B/C based on real scoring model)

Verification (email status, website classification)

AI-driven outreach (personalized angles per prospect)

Multi-facet segmentation (geography, category, tier, score)

Human approval gate before any outreach deployment