Reading a map doesn't move you. At some point you pick a direction and walk.

Had coffee last week with a founder who'd read the whole series. 20-person B2B company, knows his numbers cold. He had a workflow in mind - lead qualification, CRM update, draft follow-up in under 5 minutes. Current speed-to-lead: around 4 hours. He'd taken notes on every issue.

He said: "I've done everything but the thing."

Yeah. That's where most people land at the end of this.

Nine issues of field notes doesn't get you an agent. You have to pick your position and take the next step for that position. There are three positions. The honest next step is different for each.

Position 1: You know what you want and your team can build it

You have a specific workflow in mind. The mapping work is done. You have actual internal technical capacity - not "we could probably figure it out" but a real person who can own the build and maintain it after launch.

Go back to Issues 4 and 8. Use the playbook framework from Issue 4 to evaluate whether your workflow fits an established pattern or needs custom work. Use the 12-step sequence from Issue 8 to structure your pilot. Keep Issue 5 close for when the first production edge cases appear.

Before you build anything: answer the three prerequisite questions from Issue 7. Do you know where your data lives and who owns it? Do you have the system connections you need? Do you have an ownership plan for after launch? If any answer is fuzzy - that's your actual starting point. Not platform selection.

Position 2: You have a workflow in mind but not the internal capacity to execute it

You know what you want. You can describe the trigger, the outcome, the systems it needs to touch. But you don't have - or don't want to commit - the internal time to build and maintain it. The economics from Issue 6 made clear that DIY isn't cheaper once you count your own opportunity cost.

You want people who've done this before, who design it right the first time, and who are still responsible for it 6 months from now.

This is what the mapping call is for. Book a 30-minute working session with me - not a sales pitch. We map your current growth workflows against what agent automation would actually produce, give you an honest read on cost and whether it makes sense, and you leave with a clear picture of the opportunity whether or not you ever work with us.

It's the right move when the workflow is complex enough that DIY carries real risk, when it's novel enough that no off-the-shelf tool covers it, when data sensitivity means you want someone who's designed for those requirements before - or simply when expert design and ongoing ownership are worth more to you than doing it yourself.

The next step is at the bottom of this email.

Position 3: You're not ready for either

Reasonable place to be.

Maybe the mapping work from Issue 2 surfaced questions you haven't answered yet. Maybe the economics from Issue 6 gave you pause about whether the ROI case is actually clear. Maybe the workflow you have in mind is still an idea rather than a proven, repeatable process.

Right move here: don't buy anything. Run one workflow manually with AI assistance for 30 days and measure what it produces.

Pick the workflow where late or inconsistent execution costs you the most. Use the tools you already have - Claude or ChatGPT for the drafting and synthesis steps. Track how many times the workflow fires, what it produces, and what falls through anyway.

That 30 days of data is the most useful thing you can generate right now. It tells you whether the workflow is worth automating. What the edge cases look like. Whether the leverage is real enough to justify the investment. It also gives any future implementation partner a concrete picture of what you're actually trying to solve - which makes everything downstream faster and more accurate.

One workflow. Manual, with AI assistance. 30 days. Measure it.

The compounding argument for doing something

There's a version of this decision that does nothing. You read the series, you understand the territory, you file it under "things we should probably do at some point."

That's the most expensive option - not because AI agents are urgently necessary this quarter, but because the businesses building this capability now are accumulating something that compounds.

Second agent workflow is easier than the first. Third is easier than the second. Data governance decisions, permissions framework, monitoring infrastructure, team's understanding of what agents can and can't do - built once, leveraged repeatedly. Businesses building that foundation now will be operating with a real structural advantage by the time their competitors start.

Most businesses that end up with effective agent systems didn't start with a platform selection or a consulting engagement. They started with one workflow they decided to take seriously.

Start there.

One thing to do this week

Name one workflow. Write down the trigger, the current outcome, and what a better outcome would look like. Decide which of the three positions above fits your situation - then take the next step for that position.

That's the whole task.

Want me to look at yours?

Bring one workflow to a 30-minute call with me. I'll tell you exactly where an agent would fit in your business, what it would take to build, and whether it's even worth it for you. No pitch - just an honest read from someone who does this for a living.

Book a mapping call

What Comes Next