Blog
Notes on running agents in production.
Product updates and field notes from building AI that delivers real work for software companies' customers.
// AI Agents for Customer Success Teams: Work the Exceptions
AI Agents for Customer Success Teams: Work the Exceptions
Most AI agents for customer success teams make CSMs faster. The bigger change is AI owning a job for every account, so the team works the exceptions.
Read// Usage-Based Pricing Examples, Grouped by What the Unit Counts
Usage-Based Pricing Examples, Grouped by What the Unit Counts
Usage-based pricing examples from Twilio, AWS, Snowflake, Zapier and Intercom, grouped by what the unit counts, and what changes when an AI agent drives usage.
Read// MCP vs API: The Difference, and What Your MCP Server Leaves Out
MCP vs API: The Difference, and What Your MCP Server Leaves Out
MCP vs API: an API is a contract for code, MCP lets an AI model find and call tools. Why every MCP server is a subset of the API, and what that leaves out.
Read// Seat-Based Pricing After AI Agents: Count Jobs, Not Users
Seat-Based Pricing After AI Agents: Count Jobs, Not Users
Seat-based pricing counts the people who operate software. When agents do the operating, count what the work is for instead: locations, employees, accounts.
Read// How to Agentify Your SaaS Before Someone Does It for You
How to Agentify Your SaaS Before Someone Does It for You
To agentify your SaaS, give AI the work customers still do by hand in your product and have it act for each one. Six steps, mapped to the moats that still hold.
Read// AI Monetization: Sell a Feature, an Agent, or Finished Work?
AI Monetization: Sell a Feature, an Agent, or Finished Work?
AI monetization starts with what you sell: an AI feature, an agent, or finished work. Each needs a different promise, and pricing only follows from that choice.
Read// AI Agent for SaaS: The Five Things It Has to Get Right
AI Agent for SaaS: The Five Things It Has to Get Right
An AI agent for SaaS is not a chat window on your docs. It must know the user, act with their permissions, use your tools, and own the job after the chat ends.
Read// Usage-Based Pricing vs Outcome Pricing: Who Controls the Meter?
Usage-Based Pricing vs Outcome Pricing: Who Controls the Meter?
Usage-based pricing charges for activity, outcome pricing for results. When an agent does the work, pick by who controls how much activity a job takes.
Read// AI Agent Pricing Models, Compared by Who Pays When It Fails
AI Agent Pricing Models, Compared by Who Pays When It Fails
AI agent pricing models differ less in price than in risk. Compare them by who pays when the agent fails, wastes work, or only looks done.
Read