The AI Powered Workflow I Used to Support ROI
Most GTM orgs right now are somewhere on the same curve: experimentation → intentional design → measurable ROI. People are dabbling with AI (a rewritten email here, a summarized call there), but nobody's connected it back to a workflow, an owner, or a number.
The gap usually isn't ambition. It's that most people don't know how to break a workflow down in a way that actually surfaces where AI helps versus where it just adds noise.
So I built a prompt for it.
Start with the workflow, not the tool
Before anyone touches a prompt, I have them do two things:
Name the goal. Individual ("I want to save time on X") or team-level ("drive ARR growth," "increase pipeline" “improve NRR”).
Map the workflow. A workflow is just the sequence of steps you or your team take to complete a task: vendor negotiation, account planning, onboarding, whatever. I ask people to write out 4–6 steps, then rate each candidate workflow 1–3 on frequency, effort, and importance. The highest score is where you start, not the flashiest idea, but the one that's actually costing you the most time or risk today.
That scoring step matters more than people expect. It's the difference between "I used AI to write a cold email" and "I identified the highest-leverage workflow on my team and redesigned it."
The prompt: an AI transformation strategist, not a chatbot
Once someone has a workflow and a goal, I hand them a structured prompt rather than asking them to freestyle. The prompt does five things in order:
Decomposes the workflow: inputs, outputs, decision points, human touchpoints for each step.
Flags AI opportunities by capability type: automation, summarization, classification, generation, prediction, decision support, anchored to whatever tools the team already has in its stack (think ZoomInfo, Salesforce, Claude, ChatGPT, Power Automate). If a step has no real AI application, it says so. Forcing AI into every step is how you end up with theater instead of impact.
Scores effort vs. impact on each opportunity (Low/Medium/High on both axes) and buckets them: Quick Win, Strategic Bet, Low Hanging Fruit, or Deprioritize.
Ranks recommendations, specifying the tool, how it slots into the existing workflow, and what process actually has to change.
Defines success: the metric that proves it's working (hours saved, error rate, cycle time), the quality non-negotiables, the risks, and what 30/60/90 days looks like.
That last piece is the one people skip on their own, and it's the one that actually separates "we tried AI" from "we know it worked." I keep coming back to a simple split: measurable impact (before/after time on a specific task), directional impact (did the underlying metric move), and assumed impact (the vague sense that using AI more must be helping). Most orgs are living entirely in the "assumed" bucket. The goal is to drag as much as possible into "measurable."
Why the structure matters more than the model
The tone instructions in the prompt matter as much as the analytical steps. I told it to sound like a direct, specific advisor, not a generic AI hype machine, and to calibrate its framing to whoever's using it: a manager gets team-level framing, an individual contributor gets personal-workflow framing. Same prompt, same rigor, different altitude.
A link to the full prompt can be found here. Steal it, adapt the tool list to whatever your team actually has access to, and run your own highest-scoring workflow through it. The output won't just be ideas. It'll be a prioritized, measurable plan you can actually report on.
If you use this and find a step that breaks or a bucket that doesn't fit your team, I'd genuinely like to hear about it. That's usually where the next version of the prompt comes from.
