Confluent · 2026
Confluent’s GTM AI program.
I built Confluent’s GTM AI program from zero and hired the team. The job was to turn use cases across Sales, Customer Success, RevOps, and Marketing into supported production workflows.
$12Mdocumented annual valueHow to read this result +
Annual value attributed to the AI program at Confluent, including time saved. This is different from cash savings.
122agents in productionHow to read this result +
122 agents were in production, within a total of 190+ agents and automations across the program.
$1.1MAI tooling spendHow to read this result +
Annual AI tooling spend. Team costs and other operating expenses are separate.
- 01Choose the workA use case, an accountable owner, and a reason to build.
- 02Build with contextConnect the right data and define what the agent may access.
- 03Review before launchTest with the business and set the human-review requirements.
- 04Run and measureMonitor the workflow, adoption, and value after launch.
Team · Access rules · Builder training · Value measurement
The starting point
The experiments needed a way into production.
There was no established GTM AI program, common ownership model, or shared production path. Building an agent was one part of the job. Choosing the work, defining access, testing it with the business, and supporting it after launch needed owners too.
My responsibilities
Team, delivery, and support.
I hired the team and led delivery across Sales, Customer Success, RevOps, and Marketing. I set up intake, data-access reviews, user testing, and production support. I also ran training for employees building agents outside my team.
Inside the program
Two projects I worked on.
The result
Program results.
The portfolio reached 190+ agents and automations across environments, including 20 multi-agent workforces, across three platforms. The program delivered $12M in documented annual value with $1.1M in AI tooling spend.