To help an enterprise buyer approve your first pilot, reduce the work required to start, limit the exposure involved, and define a narrow test of value. Be willing to discuss a smaller initial commitment while you earn evidence from customers who actually use the product.

That was the practical advice Dave Lambert gave during TGN Investor Office Hours on October 16, 2026. Lambert, Managing Director at Right Side Capital Management, an early-stage investor focused on capital-efficient technology companies, responded to a founder whose enterprise prospects had shown interest but hesitated to work with a startup.

The founder described an AI product for insurance-claims workflows. Its performance claims were the founder’s account, not independently verified results. Lambert’s answer focused on a question that applies across enterprise software: what must the customer commit before they can discover whether your product helps?

Find the work hidden inside “let’s try it”

A pilot can sound small to a founder and substantial to a buyer. Lambert highlighted two barriers: the resources needed to get a test running and the exposure created by sharing sensitive business data.

Before proposing another demonstration, ask the prospect to describe what starting a pilot would require. Which people would need to participate? What systems would change? What information would your team receive? Who would approve those steps?

These are discovery questions, not an assumption that every enterprise follows the same process. A promising contact may want the product while colleagues see an unfamiliar vendor, implementation work, or an uncertain result. Learning what creates hesitation gives you something specific to address.

Write down the buyer’s explanation in their language. “We need three teams to connect this” calls for a different response from “we cannot share that dataset.” Treat each as a design constraint for the first test.

Narrow the pilot before expanding the promise

Lambert suggested exploring a sandbox, an internal trial with less sensitive information, or a test using a separately held set of historical claims. The shared idea was to demonstrate value without requiring the startup to touch live business data immediately.

Historical data is not automatically safe to use. Its age does not establish permission, sensitivity, or suitability. Ask the customer which information and environment they can approve for the proposed evaluation. A sandbox also needs a clear purpose and agreed boundaries; the name alone does not resolve the customer’s concerns.

Then narrow the work. Choose one workflow, one group of users, or one question the pilot should answer. Avoid making the initial evaluation depend on adopting the complete product vision.

For example, a hypothetical document-workflow pilot might ask whether reviewers can locate a defined set of information more efficiently in a customer-approved test environment. It would not need to prove that the software can replace an entire department. A smaller test is useful when it resolves an important uncertainty the buyer actually has.

Agree on what would count as useful evidence

Lambert’s central commercial point was to prove enough value that the customer has a reason to work with an early company. Turning that into an evaluation plan requires deciding what you will observe before the test begins.

As an editorial planning tool, write a short pilot brief that answers these questions:

• What specific task will the customer try?

• What will the customer provide, and what will your team do?

• What result will the customer review, and how will they assess it?

• When will you meet to decide what happens next?

A hypothetical success criterion might be that a customer reviewer completes a specified task within an agreed time while meeting an agreed quality threshold. The threshold should come from the customer’s needs and an appropriate baseline, rather than a number selected to make your product look impressive.

For an AI product, speed alone may leave the buyer’s main concern unresolved. Ask what quality checks matter and who will review the output. Record limitations and unexpected results alongside improvements. A credible pilot creates information for a decision; it does not guarantee a purchase.

Make the initial commitment discussable

Lambert also challenged the instinct to begin with a rigid annual commitment. He described how an annual contract with automatic renewal can create another concern for a buyer who has not yet established whether a startup’s product works for them. He suggested that a shorter, more flexible arrangement may help a company get its first paying users and reference customers.

That is a prompt for a commercial conversation, not a standard agreement to copy. Ask which commitment the buyer can make now, what they need to learn before expanding, and how the initial scope fits your own delivery capacity. Consider the cost of supporting the test as well as the opportunity to learn.

The aim is to make the next decision achievable. A smaller starting arrangement should still be clear about the work both teams expect to do and the point at which they will review the result.

Turn the first customer into evidence for the next conversation

Lambert emphasized getting customers live and using the product. An interested prospect and an active customer provide different evidence. After a pilot, document what the customer tried, what changed, what remains unresolved, and whether they continued using or paying for the product.

If you want to use their name, comments, or results in future sales conversations, ask what they are willing to share. Keep the description within what the evaluation actually established.

Lambert also expressed optimism that interest in AI is making enterprises more willing to engage startups. That was his assessment of the market, not a guarantee of shorter sales cycles. His advice still returned to the initial trust problem: make the first useful test easier to start.

Before your next enterprise conversation, revise one pilot proposal around the buyer’s effort, approved information, evidence needed, and initial commitment. The useful question is concrete: what can this customer responsibly try next, and what would that help them decide?