Where AI Actually Helps in Your Role: A Field Guide for Operations, HR, EAs, and Chiefs of Staff
The most common mistake I see people make with AI is starting with the tool.
They sign up for ChatGPT or Claude or Copilot, stare at the blinking cursor, and try to think of something to do. A few half-useful prompts later, they decide AI is overhyped and go back to their day.
That's a tooling mindset. And it's the wrong way in.
If you actually want AI to make a dent in your work, you have to start with your role, not with the model. Here's the framework I use, and how it plays out for some of the most common operational roles I work with.
The framework: audit, categorize, deploy
Step 1: Audit where your time actually goes.
Not where you think it goes. Where it actually goes. For one week, keep a running list of every task that takes you more than 15 minutes. Be honest. Include the boring ones, the ones you do on autopilot, and the ones you avoid.
Most people are surprised. The work you feel busiest doing is often not the work eating most of your hours.
Step 2: Categorize each task into three buckets.
Mechanical: repetitive, rules-based, low judgment. Anyone with the right template could do it.
Pattern-based: requires judgment, but the judgment follows a pattern you've internalized over time.
Pure judgment: ambiguous, novel, requires your specific context, relationships, or taste.
Step 3: Deploy AI accordingly.
AI can fully handle most mechanical work.
AI can accelerate pattern-based work — you give it the pattern, it produces a draft, you refine.
AI augments pure judgment — it's a thinking partner, not a producer.
The single biggest unlock comes from moving the highest-frequency, lowest-judgment tasks into the first bucket and getting them off your plate.
Now let's apply this to four roles.
HR
The mechanical layer is huge here and most teams haven't touched it yet. Job description drafts pulled from existing roles plus a quick context paragraph. First-pass interview question generation tied to the actual competencies you're hiring for. Onboarding checklists built from a role description. Policy language pulled together from your existing handbook and a change request.
Pattern-based is where HR gets a real lift. Synthesizing interviewer feedback into a single decision-ready summary. Drafting performance review narratives from a year of notes. Pulling themes out of engagement survey free-text responses. Comparing your benefits language against three competitors.
Pure judgment stays with you. The actual hiring decision. The sensitive employee conversation. The call on whether a policy needs to change. AI can prep the materials, surface the considerations, and stress-test your reasoning — but the call is yours.
The HR shift: stop drafting, start deciding.
Executive assistants
EAs are sitting on one of the highest-leverage AI use cases in any company. The role is built on context, pattern, and execution speed — three things AI is unusually good at supporting.
Mechanical: meeting briefs that pull together the attendee bios, the last three interactions, the relevant project status, and the open threads — produced in two minutes instead of forty. Travel itineraries cleaned up and re-formatted. Inbox triage drafts. Calendar conflict scenarios with proposed resolutions.
Pattern-based: drafting communications in your executive's voice (after you've fed AI enough samples for it to actually match the voice). Turning a transcript into action items with owners and dates. Building research briefs for an upcoming meeting or decision. Writing the agenda before the meeting and the summary after.
Pure judgment: the relationship calls. The "do we accept this meeting" decisions where political context matters. The moments where you need to read between the lines of an email and make a real-time call.
The EA shift: AI handles the prep, you stay close to the principal.
COO
The COO trap is drowning in cross-functional reporting. You become the human ETL pipeline between every department, summarizing finance for product, product for sales, and sales for the board.
Mechanical: pulling together data from multiple sources into a single weekly view. Standardizing how teams report status. Drafting SOPs for processes that exist in someone's head but not on paper.
Pattern-based: turning a board ask into a structured analysis. Reviewing a vendor contract against your standard terms and flagging deltas. Building a competitive scan from public information. Synthesizing a quarter of metrics into a narrative — what changed, what didn't, what to dig into. Stress-testing a strategic plan by asking AI to argue against it.
Pure judgment: which problem to attack first. Which leader to coach versus replace. When to slow down because the team is at capacity. Which big bet to make with finite resources.
The COO shift: spend less time assembling information, more time interpreting it and acting on it.
Chief of staff
If anyone benefits from a thinking partner that never gets tired, it's a chief of staff. The role demands range — from drafting the principal's all-hands email to synthesizing a strategic memo to running point on the operating cadence.
Mechanical: meeting agendas, action item tracking, recurring status compilations, briefing docs in a standard format.
Pattern-based: drafting memos in the principal's voice. Turning a chaotic strategy session into a one-page synthesis. Running a comparative analysis across three options the executive is weighing. Summarizing the org's work back to itself so people see the bigger picture.
Pure judgment: reading the room. Knowing what to escalate and what to absorb. Sensing where the org's actual energy is versus where the org chart says it should be.
The chief of staff shift: spend less time being the messenger, more time being the multiplier.
The common thread
Across every one of these roles, the pattern is the same: AI is most valuable when it absorbs the assembling, drafting, and synthesizing — and you keep the deciding, relating, and judging.
If you're not sure where to start, pick one task this week that lives squarely in your mechanical bucket and runs more than three times a month. That's your first AI win. Build from there.
The goal isn't to use AI for everything. It's to use AI for the right things — so the parts of your job that actually need a human get the version of you that's well-rested, well-prepared, and thinking clearly.
That's the whole point.
