Unlock AI Sales: The Power of Managed Agents

Summary

Managed agents represent a significant opportunity for selling AI solutions by pre-packaging agent workflows with skills, memory, and sub-agents into deployable AI products. These agents can be configured with specific system prompts and given access to tools and memory, allowing them to run long-running tasks and be deployed anywhere via an API. For instance, a churn recovery agent can automatically follow up with clients based on Stripe subscription changes. The real potential lies in selling these pre-built solutions to clients, who can integrate them into custom apps, AI SaaS products, or even chat applications like Slack, which can increase AI adoption within businesses. Managed agents also offer more deterministic automation through the use of testable and improvable skills, reducing the need for traditional deterministic platforms and allowing experts to monetize their skills without revealing their proprietary IP. Memory stores and the 'dream' feature enable continuous learning, paving the way for AI operating systems and self-learning AI SaaS products like advanced customer support agents. Building these agents is most efficient using Cloud Code, which can programmatically interact with the managed agent API, though the console offers an overview of features like environments, credential vaults, sessions, and memory stores. Deployment can be handled through custom dashboards, automation platforms like n8n or Zapier for scheduled or event-triggered actions, or integrations with chat and productivity tools. Costs for managed agents include standard API fees plus an additional 8 cents per hour of operation, making them more expensive than local Cloud Code or Co-work usage.