Hire an AI Operations Lead
Give one person the dedicated job of automating everyone else's repetitive work with AI
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 4
- Confidence
- 92%
People fighting daily fires never have time to stop and automate their own tasks, even when they know AI could. Every solved this by creating a dedicated 'head of AI operations' whose only job is to spot repetitive work across the team and build the prompts and workflows to automate it — so the people doing the work don't have to interrupt it to build automations. Shipper argues every company should have this role.
Origin
Dan Shipper describes the role held by Katie Parrot at Every; he notes others (Rachel Woods, Kora's own hiring) are converging on the same idea.
Core principles
- 01Someone embedded in daily work rarely has slack to build the automation that would save them
- 02Separating the automation-builder from the work-doer makes automation actually happen
- 03The right hire wants to tinker and build, ideally with process orientation and craft knowledge of the work
- 04You are effectively building small internal applications, so adoption is the real success metric
How to run it
- 1
Create the dedicated role
Appoint someone whose explicit job is improving how the team uses AI, separate from the line work itself.
Pro tip At minimum hire someone who 'just wants to tinker and build stuff'; process orientation and craft understanding are strong bonuses.
- 2
Run a recurring capture session
Meet regularly (Shipper sits with his lead weekly) and every time you catch yourself doing something repetitively, add it to a to-do list for automation.
- 3
Build prompts and workflows for the whole team
The lead builds automations for everyone, not just the founder — Every started with the editorial operation (style-guide copy-edit prompts).
Pro tip Translate a prompt into a form engineers can use — e.g. a Claude Code command that copy-edits the whole codebase and opens a PR for review.
- 4
Drive adoption deliberately
Treat the output as internal applications; success depends on people actually using them, which requires behavioral nudging.
Watch out You must actively get people to use the tools ('Did you put this through the prompt yet?'). Easier in an already AI-first culture; a real challenge where people resist.
In the wild
Editor-in-chief Kate spent hours a day on small copy edits for house style. The AI operations function encoded the style guide into an Opus prompt; engineer Nateesh turned it into a Claude Code command that checks the whole codebase for copy edits and opens a GitHub pull request for Kate to approve.
→ Engineers effectively produce marketing copy in Every's voice, and Kate reviews PRs instead of hand-editing everything.
Common mistakes
Expecting busy operators to automate themselves
People in the daily grind default to the way they already know because experimenting feels risky and slow; without a dedicated owner, the automation never gets built.
Is it for you?
Best for
Small-to-mid teams with genuine repetitive knowledge work and an AI-curious culture who want systematic, not ad-hoc, automation
Not ideal for
Organizations where staff broadly resist AI — the adoption problem will swamp the build, and needs solving first
From the transcript
“we have an AI, a head of AI operations”
“she's just constantly like building prompts and building workflows”
“Highly recommend having an AI AI operations lead”
“built a cloud code command that just uses that prompt and checks through the entire codebase”
“suddenly your engineering team is writing marketing copy in the style you want”
From the episode
The AI-native startup: 5 products, 7-figure revenue, 100% AI-written code
Dan Shipper (co-founder/CEO of Every)