Back to Blog
Guide Technique

Building a Complete Sales AI Agent in O137: Step-by-Step Guide

1/30/2026•4 min read

Create an AI agent that automatically qualifies leads, proposes personalized offers, and hands off to sales reps? Possible in 45 minutes with O137. This concrete guide shows every step from "zero" to production-ready workflow connected to your CRM.

Context: The Perfect AI SDR Agent Objective: Autonomous agent handling 80% inbound leads:

Detects real intent

Qualifies via CRM data

Proposes perfect offer

Smart follow-ups

Hands off to right rep

Input: CRM lead, website visit, inbound email Output: Qualified opportunity + assigned task

Step 1: Define Role and Triggers Clear agent roles text O137 SDR Agent = qualification + nurturing + handoff ❌ DOES NOT: price negotiation, closing, technical support ✅ DOES: intent detection, scoring, first offer, first sequence 5 concrete triggers New CRM lead (form, outbound, inbound)

Pricing/demo visit (session >3min)

Unanswered email (>72h)

Asset download (case study, pricing)

LinkedIn mention (comment, like)

Step 2: Agent 1 - Enrichment & ICP Scoring O137 connections to configure (5min) text CRM: Salesforce/HubSpot (lead data) Data: Clearbit/Apollo (firmographics) Site: GA4 (pages viewed, time) Language: auto-detect (FR/EN/DE...) Enrichment logic text

  1. Retrieves lead data (CRM)
  2. Enriches (Clearbit → revenue, tech stack, employees)
  3. ICP Score (0-100):
    • Priority sector? +25pts
    • Revenue >€10M? +20pts
    • Aligned tech stack? +15pts
    • Target region? +10pts O137 prompt example:

text "Analyze this lead against our B2B SaaS ICP:

  • Sectors: [list]
  • Min revenue: €10M
  • Tech: [stack] Give score 0-100 + 3 strong/weak signals" Step 3: Agent 2 - Intent Detection 4 commercial intents to classify text URGENT (40pts): pricing, demo, "how much", "integration" COMPARISON (30pts): "vs [competitor]", "open source" DISCOVERY (20pts): blog, case study, "solution" DISGUISED SUPPORT (10pts): "problem with", "bug" Inputs analyzed text
  • Pages viewed (pricing=40pts, /demo=35pts)
  • Message text (NLP intent)
  • Traffic source (Google=20, LinkedIn=25)
  • Timing (Monday 10 AM=bonus) O137 intent prompt:

text "Classify this commercial intent (URGENT/COMPARISON/DISCOVERY/SUPPORT) based on these signals: [pages viewed] [message] [source] Explain reasoning in 2 lines." Step 4: Agent 3 - Complete Qualification Final score (ICP + Intent) text Total Score = ICP(50%) + Intent(30%) + Behavior(20%) Handoff threshold: 75/100 → senior SDR Nurture threshold: 50-74 → auto sequence Parking: <50 CRM context retrieval text O137 → CRM API: GET /leads/{id} → history, notes, closed deals GET /account → expansion potential? GET /tasks → ongoing actions Step 5: Agent 4 - Personalized Offer Recommendation logic text IF ICP>80 AND URGENT: → Starter + demo call 48h IF ICP>80 AND COMPARISON: → Enterprise + ROI calculator IF medium ICP: → Discovery call + industry case study Content generation O137 generates:

Personalized email (subject + body)

Call brief (3 key points to cover)

Asset to attach (industry case study)

Next action (call/demo/RFP)

Offer prompt:

text "Create email + call brief for [persona] [industry] [intent] Offer: [suggested pricing] Tone: [expert consultant, not salesy] Language: [auto-detected]" Step 6: Follow-up Sequence Orchestration 5 automated emails (O137 → email API) text Email 1 (H+2h): intro + immediate value Email 2 (D+2): case study + social proof Email 3 (D+5): objection handling Email 4 (D+10): last chance + urgency Email 5 (D+15): qualified parking Dynamic conditions text IF email 1 opened → skip email 2 IF pricing clicked → call brief + urgent task IF reply → stop sequence + SDR handoff Step 7: Intelligent Sales Handoff Assignment rules text Score >85: Senior SDR territory Score 70-85: Junior SDR Score 50-70: auto nurture (emails 1-5) <50: 30-day parking What rep receives text ✅ Auto-created CRM task ✅ AI brief (2min): context, signals, suggested offer ✅ Agent history: decisions, sent emails ✅ Score + confidence (75% → 92%) Step 8: O137 Production Deployment (5min checklist) text ✅ [ ] Visual builder: 5 chained agents ✅ [ ] APIs connected: CRM + Clearbit + email ✅ [ ] Models: GPT4o-mini (qualif) / Claude (offer) ✅ [ ] Security: PII detection + RBAC ✅ [ ] Tests: 20 real leads (sandbox) ✅ [ ] Monitoring: logs + costs + alerts ✅ [ ] Deploy: cloud staging → prod Step 9: Measurement and Continuous Optimization KPIs to track from Day 1 text Volume: leads processed/day Quality: % >75 score → SDR Conversion: MQL→SQL vs historical Cost: €/qualified lead (<€1 target) Time: first contact → SDR task delay O137 auto-improvement text

  1. CRM results fed back (win/lost/no-show)
  2. A/B test 2 prompt/offer versions
  3. Monthly ICP scoring retraining
  4. Model routing (cost/performance) Ready-to-Clone O137 Template Total duration: 45min for first workflow Scale: 1000+ leads/month no human intervention ROI: 3 virtual SDRs from month 2

text 🚀 Clone this O137 template:

  1. Import JSON workflow (attached file)
  2. Configure your APIs (3min)
  3. Test 10 real leads
  4. Go live