# Marcus Trejo, Head of Integrations & Partnerships at FieldSync — read of Agent Marketplace, June 20, 2026

> 8 years getting SaaS tools wired into other SaaS tools. Currently managing 3 developers and a partner program for a 35-person field-service scheduling company. My 4-year-old wakes me up at 5:45 regardless of what I want, so I read a lot of product pages before 7am.

## How I got here

Someone dropped this into the #tools channel of a Slack group I'm in for SaaS operators. The message was just "anyone tried this?" with no context. That kind of nudge is actually more credible to me than a LinkedIn ad because there's no incentive behind it. I clicked during the commute. Thirty minutes on the 101.

## What I clicked first

"Make your SaaS discoverable to AI agents" is the premise, and I didn't immediately bounce because I've been sitting on a real problem: we built a clean REST API two years ago and I have no idea whether any AI agent orchestration layer is actually routing work to us or if we're just invisible to that whole ecosystem. So the premise landed.

What I actually read twice was this: "No vague marketing copy - agents need signal." I appreciated the irony of that sentence sitting in the middle of a fairly vague marketing page, but the underlying point is correct and it's something I've been trying to explain to our marketing team for months.

## Where I paused

The scoring section. Specifically the honest disclosure: "we don't have live customers on this idea yet. We shipped the strategy package; you ship the customer conversations."

That stopped me because I have never seen a product page say that. Either this is genuinely transparent or it is a very clever way to pre-handle the objection. I genuinely could not decide which. I read it three times. The numbers they surface for this idea -- "$-12,864 Year-1 take-home (Fermi)" and "1 in 9 Meaningful-success odds" -- are wild to put on a page you're trying to sell from. Which is either honest or performance art about honesty. Still not sure.

## What I distrusted

"Registry of 500+ active agent teams and frameworks." That number with no definition of "active" is doing a lot of work. Active how? Posted code once? Have a live integration that called an API last week? There's a massive difference and the page doesn't come near it.

Also "verified checkmark" -- verified against what standard, by whom, refreshed how often? The Integration Verifier section says "we test discovery, auth, and billing flows" but I want to know what a failing test looks like, what percentage of products fail it on first submission, and what the turnaround on getting listed is. None of that is here.

The framing of the whole thing is a bit recursive too. I'm looking at a page about getting your SaaS discovered by agents. The page scored itself 62/100 and assigned itself negative year-one income. If the product worked the way it claims, wouldn't a company selling this product be the best possible demonstration case? Where is their own agent-ready listing?

## What would convince me

One specific agent framework -- Langchain, AutoGen, CrewAI, pick one -- that I can see actually routing a real task to a vendor they discovered through this registry. A screenshot of a Claude or GPT session where an agent pulls up a tool from the directory and executes an OAuth handshake is worth more to me than all the copy on this page combined. Even a redacted one.

The "1 in 9 odds" is honest but it makes me want to know what the 1 in 9 looks like. What did the one operator who succeeded actually do differently? That story, even hypothetical, would move me more than Fermi estimates.

## What I'd ask in an email reply

1. Can you show me one agent framework where I can verify, right now, that a listed vendor's OAuth flow was actually initiated by an agent task? Not a demo, a real log.
2. What happens after I pay the $99 and adopt the build? Is there a person I talk to, or am I just getting files?
3. You say "500+ active agent teams and frameworks" -- what's your definition of active, and how did you count them?

## Verdict: on-the-fence

The honesty about having no customers is unusual enough that I'm still thinking about it. But I could not explain to my CEO what our product actually gets after paying $99, and the core claim about agent discoverability has zero demonstrated evidence behind it. One real working example and I reply within the day.

---
*Memo by skeptic persona, generated 2026-06-20. Studio breaks own self-grading loop.*