Blog

Notes on running agents in production.

Product updates and field notes from building AI that delivers real work for software companies' customers.

// Product Adoption Starts With Work Done, Not a Feature Tour

Product Adoption Starts With Work Done, Not a Feature Tour

Product adoption guides push tours and checklists. Accounts adopt a feature when real work gets done with it, so make adoption a job the AI owns per account.

Read

// Customer Education for SaaS: Teach Inside the Job, Not Before It

Customer Education for SaaS: Teach Inside the Job, Not Before It

Customer education is usually an academy run before the work. The teaching that sticks happens inside the job, when the AI does a step and shows where it lives.

Read

// AI Agent as a Service: Buyers Want the Job Done, Not an Agent

AI Agent as a Service: Buyers Want the Job Done, Not an Agent

AI agent as a service usually means renting an agent you still set up and supervise. Buyers want the job done, run by the company that owns the product.

Read

// Customer Health Score: Make It the State a Job Holds

Customer Health Score: Make It the State a Job Holds

A customer health score usually ends at a CSM dashboard. It works harder as the state an AI job keeps checking and fixing for every account.

Read

// MCP Authentication for SaaS: Every Call Needs a User

MCP Authentication for SaaS: Every Call Needs a User

MCP authentication guides explain OAuth for one caller. A SaaS MCP server gets five, and every call has to resolve to one user with that user's permissions.

Read

// SaaS Implementation Is a Job, Not a Project Plan

SaaS Implementation Is a Job, Not a Project Plan

SaaS implementation guides plan a project and a cutover. For a vendor with many customers, it works better as a job the AI owns for each account.

Read

// Agentic UX: Decide What the Agent Does, Shows, and Teaches

Agentic UX: Decide What the Agent Does, Shows, and Teaches

Agentic UX is usually framed as one autonomy dial. Shipping it in a real product takes a second: whether the user ends up knowing where the work lives.

Read

// Credit-Based Pricing for AI Products: Five Choices We Made

Credit-Based Pricing for AI Products: Five Choices We Made

Credit-based pricing works for AI that does real work only if you settle five choices the guides treat as options. Here is what we chose, and why.

Read

// What Is an In-App Agent? A Definition, and What It Is Not

What Is an In-App Agent? A Definition, and What It Is Not

An in-app agent lives inside a software product, knows who is signed in, and does work in that user's account. How it differs from widgets and copilots.

Read