Research inquiries

When do teams trust an AI agent with a task?

Probeworks studies what happens after an AI agent goes into production: the failures that surface, how teams catch them, and how much human review it takes to keep the agent trustworthy.

We learn this by asking the people who run these agents. If you received a research inquiry from us, this page explains what it is and how we treat your reply.

This is research, not a sales process. Replying doesn’t sign you up for anything, and we won’t follow your reply with a pitch.

If you received an inquiry from us

Who sent it

Inquiries come from hello@probeworks.ai and are signed Reed. Occasionally we send the same inquiry as a direct message from @probeworks on X.

What it asks

One or two questions about a recent situation with your AI agent, such as a confident answer that turned out to be wrong or a handoff that came too late, and the work it created for your team.

What we need from you

A short written reply is enough. We don’t ask for calls, meetings, attachments or forms.

If you’d rather not

If we don’t hear back, we send at most two short follow-ups and then stop. Reply “stop” at any time and we won’t contact you again.

How we treat your answer

Unattributed by default. We won’t name you or your company, or quote you, unless we’ve asked first and you’ve agreed.
General examples are enough. Please don’t send customer data, conversation logs, credentials, internal prompts or confidential incident details. We never ask for them.
We don’t promise anonymity. We know who replied, and a very specific incident could point back to you. So we describe patterns across teams rather than individual cases, and we leave out details that would identify you.
Private. We keep replies privately, and we don’t sell them or add you to a marketing list.

What the research is for

Your answers shape how Probeworks tests AI agents: which failures to look for, what counts as a successful outcome, and how to measure the human time an agent still needs.

Probeworks also develops evaluation reports for teams deciding whether to deploy an agent. The research is kept separate from that work. Taking part doesn’t mean you’ll be offered anything, and we’ll only tell you about the reports if you ask.

Anything we publish from this research describes patterns across teams, not individual replies, unless you’ve agreed to be named.

Contact

Questions about an inquiry you received: hello@probeworks.ai.

About this page

This page has no forms or accounts. It records an anonymous, cookie-free count of visits and which link brought you here; it stores no identifier and does not track you across sites. It is hosted by Cloudflare, which may process request and security data. If you email Probeworks, your email address and message are received so we can respond.