AI agents are everywhere, but not all are worth building

AI agents are everywhere, but not all are worth building

Apr 14, 2026
Lenny's Newsletter AI SprinklerAS Gtm_strategy

The Gist

  • Hamza Farooq and Jaya Rajwani spent 50+ hours on a guide to AI agent strategy
  • Most teams have 5-10 agent initiatives but struggle to prioritize
  • The guide breaks down three types of agents and common pitfalls
Key Quotes

Prioritization breaks not because teams are bad at planning but because they’re comparing apples, oranges, and jet engines on the same spreadsheet.

Categorization isn’t just a technical exercise. It’s the foundation for smart prioritization.

Key Insights
  • AI agents can be categorized into three architectural types: deterministic automation, reasoning and acting agents, and multi-agent networks.
  • Most agent opportunities (60-70%) belong to Category 1 (deterministic automation), which are fastest to launch and deliver measurable ROI quickly.
  • Teams often misprioritize agent initiatives by treating architecturally different products as if they’re in the same category, leading to ineffective planning.
  • Category 2 (reasoning and acting agents) is suitable for 25-30% of agent opportunities, where dynamic decision-making is required.
  • Multi-agent networks (Category 3) are typically reserved for later stages when multiple teams must coordinate across domains.
  • Organizations often overengineer Category 1 problems with Category 2 frameworks, adding unnecessary complexity and cost.
Actionable Takeaways
  • Categorize your agent ideas into the three architectural types (deterministic automation, reasoning and acting agents, multi-agent networks) before prioritizing them.
  • Start with Category 1 (deterministic automation) projects for quick wins, especially if you have limited AI engineering capacity or need fast ROI.
  • Use success metrics tailored to each category (e.g., workflow completion rate for Category 1, task completion rate for Category 2) to evaluate performance.
  • Avoid overengineering Category 1 problems with Category 2 frameworks, as this adds unnecessary complexity and cost.
Data Points
  • 60-70% (Percentage of agent opportunities that fit into Category 1 (deterministic automation).)
  • 25-30% (Percentage of agent opportunities that fit into Category 2 (reasoning and acting agents).)
  • 52% to 87% (Workflow completion rate improvement over 8 weeks for a Category 1 email support agent.)
  • $18K/month (Savings achieved by automating 3,000 support emails/month with a Category 1 agent.)
  • 71% to 86% (Task completion rate improvement over 4 months for a Category 2 voice + image shopping assistant.)
  • $0.12 to $0.08 (Cost per session reduction for the same Category 2 shopping assistant.)

RevBots.ai View:

Before jumping on the AI agent bandwagon, ensure your initiatives align with broader GTM strategy and avoid common traps.