know where ai belongs before you build it
Most AI roadmaps start with a survey, a workshop and a strong opinion. macro starts with what actually happened: which workflows repeat, how often, who touches them and where they stall. Then it ranks them, using the same measures for every workflow.
the pilot with no baseline
If you cannot say what the work cost before, you cannot say what the agent saved.
Employee surveys capture what people remember, not what happens most often. Task analysis looks at one desk at a time and misses the handoffs. Executive intuition cannot say where the problem lives. None of them can see a loop that runs across three departments, because nobody who answers can see it either.
macro evaluates automation opportunities using real operational patterns instead. It watches how work moves through the tools you already use, finds the workflows that repeat, and measures each one the same way. A request in sales sits next to a close in finance on one page. When the board asks why this agent first, the answer is a number.
what macro measures, and why it matters
| dimension | what macro measures | why it matters |
|---|---|---|
| frequency | how often the workflow occurs | repetition is the case for automation |
| people and systems | which teams, people and systems are involved | shows who an agent would touch |
| delays | where handoffs stall and steps repeat | the cost sits in the waiting |
| impact | estimated time, cost and efficiency effect | tells you what it is worth |
| feasibility | automation feasibility and implementation complexity | tells you what can actually be built |
| recommendation | the agent to build and its human checkpoints | turns a finding into a plan |
three questions the board will ask
what work is repeating?
macro shows recurring workflows across teams, systems and departments. Not tasks but whole loops, from the first email to the last approval. The sales-to-finance request shows up as one loop with a count next to it, not as six people's separate to-do lists.
where will ai have the greatest impact?
macro compares opportunities on frequency, effort, cost, delay and automation feasibility. A workflow that runs weekly across three teams outranks one that runs monthly at one desk.
what should we implement first?
macro prioritizes the AI agents most likely to deliver measurable value. Each comes with the teams affected, the systems touched and the checkpoint where a person still decides. You start at the first line and work down.
a sample opportunity map
Illustrative only. The workflows are typical of a mid-size enterprise, and the hours, rankings and agents are examples, not measurements from a real customer.
| workflow | teams | frequency | estimated hours/week | feasibility | recommended agent | human checkpoint |
|---|---|---|---|---|---|---|
| customer request to signed contract | sales · legal · finance | ~12 per week | ~9 | high | drafts the contract from crm data and routes it for review | legal approves terms before anything is sent |
| new vendor onboarding | procurement · finance · it | ~4 per week | ~6 | medium | collects documents and checks required fields | finance approves payment terms |
| support escalation to engineering | support · engineering · product | ~20 per week | ~8 | high | summarizes the ticket history and files the issue with context | an engineer confirms priority |
| new-hire access requests | hr · it · security | ~6 per week | ~4 | high | provisions standard tools from the role | security approves anything outside the standard set |
| monthly close reconciliation | finance · operations | monthly | ~5 | medium | matches entries across systems and flags gaps | the controller reviews every exception |
from intuition to evidence
a survey asking teams what feels slow
a task inventory taken one desk at a time
a pilot chosen before the baseline existed
an agent with no one responsible for its output
a count of how often each workflow actually ran
the whole loop, from the first email to the last approval
a ranking built from frequency, effort, cost, delay and feasibility
a recommended agent with its human checkpoints written in
bring evidence to the next budget meeting
Connect the tools you already use and get a ranked, evidence-backed map of where AI belongs in your company, with the checkpoints already drawn in.