Putting it together · for operations and project managers

Building AI workflows in operations and project management

Much of operations work repeats. The weekly report, the monthly vendor review, and the follow-up after every steering meeting follow the same pattern each time. Writing a fresh prompt for each one produces a different format every week and a different set of errors.

Putting it together means breaking recurring work into stages and deciding what AI does at each stage. It also means fixing where a person checks the output. Done well, the team runs the work the same way each cycle and can hand it to someone else when needed.

Where it comes up

  • An automated weekly operations report

    Export the same data each week. A saved prompt drafts the report in a fixed template. You recalculate the headline figures and write the commentary yourself. The format stays the same from week to week.

  • Meeting follow-up as a routine

    Transcript to summary, summary checked against notes, action items moved into the tracker, reminders scheduled. Each step has one job and one person responsible for it.

  • Keeping procedures current

    When a system or policy changes, a saved prompt compares the change against each affected procedure and lists the steps to revise. The process owner reviews and approves each revision.

What it looks like at each band

  1. E1 · Foundational

    Applies AI to single steps and carries results between them by hand. Starts each recurring report from a new prompt. Output format and quality vary from cycle to cycle.

  2. E2 · Proficient

    Runs recurring work as a saved sequence of prompts and templates. Reviews output at a fixed point before it moves on. The same report looks the same each week.

  3. E3 · Distinguished

    Defines where each AI step ends and where a person reviews. Matches the tool to the stage, such as a spreadsheet for calculations and a language model for the narrative. Documents the workflow so a colleague can run it during leave.

  4. E4 · Exceptional

    Builds the team’s reporting and process system and tracks how it performs. Knows at which step errors appear, which steps reduce effort, and which need closer review. Removes steps that no longer help. Other teams adopt these workflows because they produce reliable results.

Common mistakes

  • Asking the model to do the arithmetic

    Language models can make calculation errors. Do the calculations in the spreadsheet and give the model the results to describe.

  • Automating the send step

    A workflow that sends a report or updates a tracker with no human review will eventually distribute an error to everyone on the list.

  • A workflow only one person understands

    If the prompts and steps live in one person’s chat history, the process stops when that person is away. Write the steps down where the team can find them.

A worked example

Producing a monthly fulfillment report from order and shipping data for the operations director.

A weaker attempt
Each month, paste the raw order export into a conversation with the model and request a fulfillment report.

The model calculates figures itself, the layout changes every month, and nobody checks the numbers at a fixed point. Month-to-month comparisons become unreliable.

A stronger attempt
Step 1: a saved spreadsheet calculates on-time rate, average days to ship, and returns by site. Step 2: a saved prompt drafts the report from those figures in the fixed template. Step 3: you compare every figure in the draft with the spreadsheet. Step 4: you write the explanation for any site that changed by more than the agreed threshold. Step 5: send.

The spreadsheet does the arithmetic and the model writes the narrative. A person checks the figures and explains the changes. The report keeps the same layout each month, and a colleague can produce it during the owner’s leave.

The full study guide for this skill, for every role, is Putting it together. The other five skills for operations and project managers are on the AI skills for operations and project managers page. A one-page summary is in the AI workflow checklist.

See where you stand

The free check grades putting it together and the other five skills in about fifteen minutes. Nobody sees the grade but you.

Then build the skills and prove it. The three steps for operations and project managers list the courses and templates that fit the work.