Agents won't save broken CS

We've been talking a lot about agents in post-sales. Andrew Carothers wrote a guest essay for us on what that era actually changes, and what it doesn't.

His point: the agents work. That was never in doubt. The problem is what you feed them. Put an agent on siloed data, unwritten processes, and unclear ownership, and it runs the same confusion, only faster.

The teams getting something real out of this do three unglamorous things first:

• Pick the number - renewals, expansion, churn. Wire the agent to a metric that matters, not the easiest workflow to automate.

• Fix what's underneath - connected data, a process you could teach a new hire in an afternoon, and a number you're already tracking.

• Rebalance human attention - agents cover the book with enough context that people only show up where a person changes the outcome.

Capacity used to be the bottleneck: how many accounts one CSM could actually cover well. Agents help with that. What's left is defining what a good interaction looks like clearly enough that an agent can run it without you.

→ Read the full essay

If any of this is useful, reply - or grab time if you'd rather talk live.

Chandrika