Value Stream Mapping for AI: Find the Work Worth Automating

Value stream mapping is how you identify where you create value — and then find new ways to create it.

Done well with AI in mind, it produces two kinds of outcomes, and it's work you do together with everyone involved in the process.

The exercise means sitting with all the people in the value stream — for example, from sales through invoicing, through implementation, through post-implementation, customer service and support — every person involved in that life cycle.

You look at all the things they do and all the systems they use .

This grounds your AI search in reality instead of starting from a demo you saw.

When you map everything a team does, most of it is fine.

There may be a little friction, but if something costs one person thirty minutes a week, that's fine — leave it alone.

The red-pen process is about finding the few things worth circling with a big red pen.

These are the steps where: You copy and paste out of one system into a spreadsheet You send it to someone You wait for someone to call somebody You fill in the spreadsheet All of that should be a process.

So step one is fixing and optimizing the problems in your existing process — the obvious, high-friction handoffs you circled — before you reach for anything more ambitious.

The first pass fixes broken steps.

The second pass is where new value appears.

As the speaker puts it, you take the microscope and dial it out one level to look at the thing as a whole.

When you can see your company as a whole, you can develop capabilities that translate into different ways you deliver your service to your customer .

You're able to see something you didn't see before — something that's hard to spot from inside any single step.

Two specific openings show up here: Small use cases that were too expensive to run — you can now do them.

Mundane, repetitive tasks — hand them to AI, and keep humans for what they do best, like customer support and driving upsells.

One effective approach is to let your own people try use cases first.

It does a good job of weeding out the good and bad ideas.

Rather than deploying a team of engineers to invent use cases, the better-spent dollar is to have 100 people come up with things .

That list whittles down to the six or seven pretty good ideas — and from those, you turn one or two into the real home runs.

The winners are the proof-of-concepts worth taking the last mile.

There's a catch: not every use case is right.

There's no one-size-fits-all with AI .

AI is really good at some things and awful at others, so you have to use the right tool for the job.

A voice agent isn't right if every request is highly specific and technical, but it fits when the requests are mostly FAQs.

Once you start building, you end up having to become a product manager pretty quickly.

In pre-AI days, the time from an idea through planning, development and testing was long enough that a product manager could catch their breath and really think.

With AI-driven development, the time between something you've thought about and something you actually see on the screen might be six minutes .

That speed rewards having a point of view: what's the strategy for this product, what's it supposed to do, who does it serve?

People trained in that answer the questions the AI comes back with quickly.

Without that grounding, it's easy to go down a rabbit hole or a path that won't scale — which is why the winning POCs eventually go to professionals to take them from prototype to production .