What We Look For

We Build AI That Does Work.

Useful work for a specific customer. Value you can measure.

Specific Customer.

Know exactly who we're building for.

Specific Task.

Perform real work or improve a workflow.

Measurable Value.

Demonstrate why the improvement matters.

Niche Is a Feature.

Become indispensable to a clearly defined customer. Start with a narrow market and a specific problem.

Read moreShow less

We prefer narrow markets.

From broad to specific

“AI for businesses” isn’t particularly interesting.

AI that eliminates a repetitive estimating task for commercial roofing contractors?

Now we have something to investigate.

Specificity gives us an advantage.

When we know exactly who the customer is, we can understand what their day looks like, what consumes their time, what frustrates them, what those problems cost, who makes the buying decision, and how to reach them.

We would rather deeply understand 20,000 potential customers than vaguely understand 20 million.

We don't start by asking how big the market can become. We start by asking:

The Product Must Do Work.

Complete a task or make an existing workflow meaningfully easier. The closer it gets to finishing the work, the better.

Read moreShow less

This is one of our strongest filters. A LaunchWorks product should perform a task or make an existing task meaningfully easier.

It should do something.

Complete a task.
Review documents.
Move information between systems.
Produce a necessary business output.
Find something someone would otherwise have to find manually.
Reduce errors.
Automate a workflow.
Help one employee accomplish significantly more.

We aren't interested in AI simply because the technology is impressive.

We Look for Time.

Measurable time savings create a clear business case.

Read moreShow less

Time is one of the easiest forms of value to understand.

The before and after

Imagine someone currently spends: 4 HOURS completing a recurring business task.

The product reduces it to: 20 MINUTES

Measure the work today

We want to understand how a task is performed today, who performs it, how often it happens, how long it takes, what that person's time costs, and how much of that work the product can eliminate. When we can measure the difference between before and after, we can begin measuring the economic value the product creates.

Illustrative example

4 hours 20 minutes

Before

4 hours

After

20 minutes

Painkillers Over Novelties.

Solve problems customers already spend time, money, or labor addressing.

Read moreShow less

The existence of AI doesn't create demand. Real problems do. We look for workflows people already need to complete and problems they're already spending time, money, labor, or attention solving.

A question we ask:

What happens if the customer doesn't use this product?

A warning sign

If the answer is:

“Not much.”

That's a warning sign.

But if the answer is:

“Someone still has to spend six hours doing this manually every Friday.”

Now we're interested.

Boring Can Be Beautiful.

Invoices. Estimates. Contracts. Compliance. Ordinary work can create extraordinary value.

Read moreShow less

Some of the best AI businesses may look remarkably boring from the outside.

Reviewing invoices.
Preparing estimates.
Reconciling records.
Processing applications.
Creating compliance documentation.
Reviewing contracts.
Building reports.
Scheduling field employees.
Analyzing inspection records.
Updating systems.
Finding discrepancies.

These aren't necessarily exciting AI demonstrations. They're work. And businesses spend enormous amounts of money paying people to perform work.

AI Should Create Leverage.

Move people from doing to reviewing. AI performs the work; humans verify, approve, and handle exceptions.

Read moreShow less

We're particularly interested in products where AI fundamentally changes the economics of performing a task.

Not:

“AI helps someone do this slightly faster.”

But:

“This used to require three hours of skilled labor. Now it requires fifteen minutes of review.”

That's the shift we're looking for:

Doing → Reviewing

AI performs more of the work. The human verifies, approves, corrects, and handles exceptions.

Distribution Matters From Day One.

A useful product needs a practical path to buyers. Can we identify—and name—the first 100 customers?

Read moreShow less

A great product without a practical path to customers isn't necessarily a great LaunchWorks opportunity. That's another reason we like niche markets. Specific industries often have concentrated groups of buyers, trade associations, conferences, professional networks, publications, existing software ecosystems, and other ways of reaching the people who actually have the problem.

One of our tests is remarkably simple:

Can we identify the first 100 potential customers?

Better yet:

Can we name them?

Start narrow. Earn the right to expand.Read our approachShow less

We don't need enormous markets on Day One. We would rather dominate a small problem first.

One customer type.
One workflow.
One painful task.
One exceptional solution.

Then expand.

A product that owns one workflow can move into the workflow next to it. A product built for one role can expand across an organization. A product built for one industry can eventually enter adjacent industries. But expansion should be earned through customer understanding.

The LaunchWorks Test

Ten questions we ask before an equity partnership.

View the ten questionsHide the ten questions

Before we enter an equity partnership, we want clear answers to a few fundamental questions:

CUSTOMER: Who specifically uses this?
TASK: What specific work does the product perform or improve?
FREQUENCY: How often does that task occur?
CURRENT PROCESS: How is the work performed today?
TIME: How much time does the current process consume?
VALUE: What does solving it save or create?
BUYER: Who has the authority and incentive to pay?
DISTRIBUTION: Can we identify and reach the first 100 customers?
AI ADVANTAGE: Does AI materially change the economics of the task?
EXPANSION: If we win this narrow problem, what can we logically solve next?

Niche. Useful. Measurable.

Built something that fits? We'd like to see it.

Pitch Your Product →