How to build a B2B SaaS solo

The playbook

1. 1

   Pick an ICP narrow enough to automate

   Broad ICPs kill solo B2B because every variation requires human judgment. A narrow ICP (e.g. 'Series A-B DTC brands running Shopify with 10-50 employees doing $5M-$50M revenue') lets your AI team run a lean, repeatable funnel. Before anything else: write a 200-word ICP description and 10 concrete company names that fit. If you can't name 10, the ICP is too abstract. Medvi's ICP was 'overweight adults with insurance willing to try GLP-1'; narrow, scalable, automatable.

   AI CMOAI Head of ResearchLinkedIn Sales NavigatorClay for enrichment
2. 2

   Ship a product that solves one painful thing well

   Solo B2B can't win feature wars against 50-person product teams. Win by being the cleanest solution to one specific, valuable problem for your ICP. Your AI CTO helps scope v1 to 2-3 core capabilities, ship in 4-8 weeks, then hold the line against feature sprawl. Every feature request goes through an AI Head of Product filter: does it serve the ICP's core pain, or is it a distraction? The discipline of 'no' is the hardest part of solo B2B — your AI team can surface the logic but the founder must commit.

   AI CTOAI Head of ProductLinearFeature request triage
3. 3

   Inbound + outbound hybrid, run by the AI sales team

   Inbound: AI CMO ships 20-30 SEO pages/month targeting the ICP's buying-intent keywords (see the Solo SEO Playbook). Outbound: AI Sales Rep runs a weekly cohort of 100-300 personalized emails to ICP-fit accounts, using AI Head of Research for pre-call intel. Founder reviews 10-20 flagged high-intent leads per week and books the demos personally. Rule: every outbound email is pre-approved in template form by the founder, and AI SDR only personalizes within guardrails. Never send AI-generated cold email without founder-set voice.

   AI CMOAI Sales RepAI Head of ResearchClayApollo or Instantly
4. 4

   Founder runs the demos, AI runs everything around them

   The founder shows up for 2-5 live demos per week. The AI team handles everything around them: scheduling (AI assistant), pre-call brief (AI researcher pulls company + person context 24h before), post-call summary + CRM update (AI CRM agent), follow-up email within 60 minutes (AI sales rep), proposal within 48 hours (AI sales rep with founder review). The founder's job in the demo is product expertise, commercial empathy, and trust. Everything else is automated.

   AI Sales RepAI CRM AgentCalendly or similarProposal template engine
5. 5

   Onboarding and customer success as a product, not a person

   Every new customer gets a templatized onboarding flow: welcome sequence, kickoff call with the founder, 30/60/90 day AI-driven check-ins, health score monitoring, and proactive outreach when health drops. The AI Customer Success Manager runs all of this; the founder only joins calls at kickoff, 60-day QBR, and renewal. Support tickets route through AI first responder with founder-escalation on edge cases. Target: <5% of customer interactions require founder presence after the first 30 days.

   AI Customer Success ManagerAI Support AgentHealth score dashboardHubSpot or Attio
6. 6

   Enterprise readiness: SOC 2, MSA, security review, solo

   Enterprise buyers gate on security + legal, not team size. Your AI COO runs SOC 2 via Vanta or Drata (90-120 days to Type I, 6-12 months to Type II). Your AI Contract Review agent handles inbound MSA redlines with a pre-approved fallback playbook. Security questionnaires (Whistic, SecurityScorecard) get answered by the AI Security Agent with founder-approval on gray areas. This is where solo B2B wins or loses — 80% of founders skip it; the 20% who ship compliance unlock $50K-$500K ACV deals.

   AI COOAI Contract ReviewerAI Security AgentVanta / DrataWhistic
7. 7

   Support at scale: AI first-line, founder for escalations only

   Your AI Support Agent handles 80-90% of tickets via Intercom/Front/Zendesk with your knowledge base as ground truth. Founder gets paged only for: (1) tickets the AI flagged as unclear after 2 attempts, (2) churn-risk signals, (3) enterprise-tier customers. Target SLA: first response <5 min (AI), resolution <24h for 90% of tickets. The AI agent writes every new resolution back to the knowledge base so the next similar ticket is faster. Support quality often goes up vs. a human-only team because response time collapses.

   AI Support AgentIntercom / Front / ZendeskKnowledge base (Notion or Helpjuice)