DUNDERBLIXEM
Menu

Most AI spending speeds up the wrong work.

We help find what's actually holding your business back, then equip your people to fix it.

Bring us to your retreat Get the free playbook

Video placeholder

Dylan on camera.

The problem

Everyone's investing in AI, but few are actually seeing a return.

Too much AI deployment is aimed at things that aren't the real bottleneck for the business.

That's a recipe for tools that don't get used, or worse, unhappy employees who either actively try to sabotage the project or just quit.

See it happen

A few things worth noting

  1. 1
    Leads might really be your constraint.We're not against using AI for lead gen. If your team has room for more work, more leads is exactly the right move. But more leads only means more money if the rest of the business can turn them into finished work.
  2. 2
    You can put AI in more than one place.Of course. Order matters, though. Fix what's really holding things back, and the constraint moves somewhere else. Then you go after the next one.
  3. 3
    This is a simple model.Real businesses have busy seasons, different kinds of jobs, people out sick, and sometimes more than one problem at once. The numbers are rough industry averages for a small HVAC shop: three techs at about four jobs a day, a $400 average ticket, and about $104 a lead from paid ads. Your business will differ. The point is to find where your work is actually stuck before you spend.

More leads won't fix a full schedule.

Here's a typical small HVAC shop. Each dot is a job. Turn on AI wherever you think it'll help, then watch what happens to the work and the people doing it.

Put AI on:
New leads–per week
Jobs finished–per week
Wait for a tech–days
Customers lost–per week
AI cost$0per week
Return on AI$0per week, vs. today

Today

On a bigger screen, there's an interactive model here: turn on AI in different parts of a service business and watch what happens to the work.

If you speed up the wrong thing, you get busier but not more profitable.

What we do differently

01

Constraint first, tools second.

Don't get sold on a specific deployment unless you know it's what's actually limiting your business. We help you find where investments will really pay off.

02

The answers come from your people.

They already know where work gets stuck. We know how to get it out of them.

03

People-first, on purpose.

We don't take projects that are only about cutting payroll. AI should make your people better, not get rid of them.

Our method

Pattern Interrupt

Our job is to help you and your people generate fresh insights and execute on them as effectively as possible.

It's inevitable: the people in a business get stuck viewing it a certain way, and blind spots develop.

A pattern interrupt is purposefully generating a fresh perspective. That leads to more effective problem solving and execution. That's what our methods are built to provide, whether you're doing it yourself with your team or we're organizing an entire retreat around it.

Aim

See what's really going on.

  1. GoalWhat are we actually trying to achieve? And what happens if it works?
  2. GroundWhat's the actual daily work? Watch it and check the data.
  3. WallGet everything that's going wrong out of heads and onto the wall.
  4. Root causesMap how it all connects and find the root causes.

Strike

Act on it, with your people.

  1. RoadmapBuild an action plan that maps what needs to be done to go from your current reality to the way you want things to be.
  2. PeopleEngage the people who will do the work, both internally and externally as needed.
  3. DeployMake the changes, starting with what matters most. Delete steps before you automate them.
  4. CheckCheck results continuously, starting at day 30 and day 90.
  5. AdaptAdjust as you learn, then go after the next problem.
More on the pattern interrupt

How to work with us

We meet people where they're at. There's a range of ways we can help you, from doing it on your own to building a whole retreat around it.

Start here · Free

On your own

Are you or someone on your team excited about doing this and want to lead it yourself? Here's material that can help.

  • The playbook
  • 1-month boot camp, doing it together (?)
Start on your own

We lead it with your team

2-week sprint

Two weeks working with your team toward one specific outcome we design together.

  • Virtual sessions throughout
  • Recommended At least one day on site
  • A day-by-day schedule

From [PRICE]

See the schedule

Most effective

Built around your retreat

An engagement with the retreat at the center.

  1. Before the retreatat least 1 month
  2. At the retreaton site
  3. After the retreatat least 3 months

Starting at [PRICE]

Plan your retreat with us

This works at any size. An individual, a team of one, a small team, or a whole company. If you're small but still want to work with us, we're happy to.

* We also speak at events. If you need help motivating your people, getting them to think differently, and giving them ideas they can take action on, we'd be happy to talk at your event.

Free playbook

How to avoid wasted AI spending.

Find what's really holding your business back and check every AI dollar against it. Do it with your team in an afternoon.

Inside

  1. 01Why most AI spending misses.
  2. 02Find what's really holding you back, in an afternoon.
  3. 03Check your spending against it.
  4. 04Equip your people.
  5. 05What to measure at day 30 and day 90.

Questions people actually ask

Draft answers (?).

Where do we start?

Start with what you're trying to achieve, not with a tool. Get specific: what would be meaningfully better six months from now, and what number would show it? If nobody can name that number, it's too early to buy anything.

Next, look at how the work actually gets done today. It's often different from how leaders think it gets done, so ask the people doing it. Write down everything that's going wrong, then look for the few root causes behind most of it.

Only then ask where AI fits. Sometimes it's the answer. Often the first fix is simpler: a deleted step, a clearer handoff, or an owner for something nobody owned. Pick one workflow, make the change, measure it for a month, and let that result decide what comes next.

Do we have to fix our data first?

It depends, and it's worth getting right.

You don't need perfect data across the whole business before you start. Trying to clean everything first is one of the most common ways AI projects stall for a year. What you need is trustworthy data for the first thing you're trying to improve. A small, clean, well-understood set of data usually beats a huge messy one.

Trustworthy means more than tidy. A few questions to ask about the data your first change will touch:

  • Does it mean what you think it means? "Closed" might mean won in one system and gave up in another. Definitions that drift between teams quietly break everything built on top of them.
  • Which version is the real one? Ask three people where the authoritative customer list lives. If you get three answers, that's the first problem to solve.
  • Does it reflect what actually happens? If half the team works around the system, the system records the workaround, not the work.
  • Is it complete and current? Gaps and stale records look fine until something acts on them.
  • How does it come in? A lot of so-called AI errors are really intake problems: free-text boxes, missing required fields, and three different date formats.
  • Who owns it? Every set of data an AI system depends on needs a named person who keeps it right.

AI makes data problems louder, not quieter. A person notices when a number looks off. An automated system acts on it, at scale, with confidence. So check the data your first change depends on, fix how it comes in, write down what each field means, and keep a person reviewing results until you trust them.

Getting your data in order is part of the work, not a gate you have to pass before you can begin.

What happens to our people?

Start with them. People who help shape a change tend to use it. People who have a change done to them tend to work around it, sabotage it, or leave.

Much of the resistance to AI isn't really about the technology. It's about relevance, identity, and job security, and it deserves an honest conversation, not a pep talk. In practice that means:

  • Talk to the people who do the work before deciding anything.
  • Be straight about what the change means for their jobs.
  • Decide ahead of time what happens to the time it frees up.
  • Find the people who are already excited, and let them lead and teach others.

If a plan only works by cutting people, we're not the right fit. We don't take projects that are only about replacing people or cutting payroll.

Can't we just do this ourselves?

Often, yes, especially if you have someone who cares about this and has the time to lead it. That's why the playbook is free.

What's hard to do alone is see your own blind spots. Inside a business, everyone's view is shaped by the daily firefight, and rank changes what people are willing to say in the room. An outside facilitator's job isn't to hand you answers. It's to create the conditions for honest input and fresh perspective, and to keep the process moving when it would otherwise stall.

If you try it yourself and get stuck, that's a good moment to bring someone in.

What happens if it works?

Ask this before you start, not after. Three questions:

  • Who runs it day to day?
  • Who improves it when it breaks or the business changes?
  • What does your team do with the time it gives back?

Hours saved only count if they go somewhere: more customers served, faster turnaround, work that kept getting dropped, or simply less overtime and stress. If nobody decides, the time tends to disappear into more busywork, and the return never shows up.

Plan for how the work itself changes, too: what people stop doing, what they start doing, and how you'll support them through it.

Do you build the AI too?

Sometimes. A lot of AI deployment can and should be led by your own team, especially once the right people are equipped. Some of it needs a little outside help.

Bigger or more involved deployments, especially ones that touch sensitive data, compliance, or core systems, may require pulling in additional implementation experts. We can help with that if and when it's needed, including finding the right people and working alongside them.

How do we vet an AI consultant, including you?

A few questions go a long way:

  • Can they show you something they built that's running today, and can you talk to that customer?
  • What will they ask your people, and can they explain their recommendations out loud, not just in a document?
  • What exactly will their discovery or audit produce, and how does it get you closer to a result?
  • How will success be measured, and who owns the result after they leave?
  • Will you own everything they build, including the prompts, settings, and code, or will you need them for every change?
  • What happened on a project that didn't go well?

Be wary of quick audits that conclude you need exactly what they sell, and of roadmaps that came straight out of a chatbot. Ask us every one of these questions too.

Is this only for big companies?

No. It works at any size: an individual, a team of one, a small team, or a whole company.

Smaller teams often move faster, because there are fewer handoffs and less politics. The size of the engagement changes. The thinking doesn't.

Let's find what's really holding you back.

Bring us in