# Marcus Tillman, Senior AI Infrastructure Engineer at Fieldstack (62 people, Series A SaaS) — read of SearchQ, June 18, 2026

> 9 years backend, last 3 years almost entirely LLM tooling and agent pipelines. Currently fighting a losing battle against our Brave Search integration returning garbage for anything library-specific.

## How I got here

Typed "search api for llm agents not google" into Google last Tuesday at 11pm after spending two hours debugging an agent that kept citing a React Router v5 pattern when we're on v6. Someone on the LangChain Discord had linked this in a thread about tool reliability. Came in cold, no prior brand awareness.

## What I clicked first

The problem framing hit immediately. "Stack Overflow answers from 2019. API docs from deprecated library versions. Your agent confidently integrates against a 3-year-old API signature." That's verbatim what broke my Tuesday. I read that and sat up straighter. The trifecta of stale docs / SEO spam / formatting noise is exactly correct. Whoever wrote this has actually built agents before, or at least talked to someone who has.

## Where I paused

The Jordan Chen quote. "Cold-start latency dropped from 3.2s to 280ms. I got 8 hours of engineering time back per week." I stopped there for a full minute. That's a specific, believable number. Not "10x improvement" or "dramatically faster." 3.2s to 280ms is the kind of precision you get from someone who actually ran the benchmark. I wanted to find Jordan Chen on LinkedIn right then.

Then I scrolled further and found the Wishdeal Factory scoring box. "Honest disclosure: we don't have live customers on this idea yet." I had to re-read that three times. Wait. Jordan Chen isn't a real customer? The product isn't live? I just read a testimonial from someone who may not exist for a product that doesn't exist.

## What I distrusted

The whole bottom half of the page is a different product than the top half. The top is a product page for SearchQ the API. The bottom is a pitch to sell me the *concept* of building SearchQ for $99-$199. These two things cannot coexist on the same URL without one of them poisoning the other. "Jordan Chen, AI Infrastructure @ mid-scale AI startup" is doing a lot of work and the fact that his company has no name, no domain, no LinkedIn profile link now reads as fabricated. Maybe it isn't. But once I see "we don't have live customers on this idea yet," I can't trust any social proof on the same page.

The "1 in 8 meaningful-success odds" and "$-22,000 Year-1 take-home" are interesting honesty signals, but they also tell me the people behind this didn't build a search API. They built a dossier-packaging business and SearchQ is inventory. That's a totally fine business model. But it means I came here looking for a tool to integrate today and I'm actually reading a franchise brochure.

## What would convince me

If SearchQ the API actually existed and was live, I'd need one thing: a public index of domains they crawl, updated. Not a sentence that says "top 10,000 technical sites." Give me the list. Show me that docs.python.org, the Anthropic SDK GitHub, the FastAPI docs, the Pydantic changelog are all in there. That list is worth more than any testimonial because I can check it against my own agent's failure modes.

If this is genuinely just an idea for sale, I want to see the data behind "buyer clarity: 10/10." Who did you talk to? How many developers confirmed this pain? What did they say they'd pay per month before they saw a pricing page?

## What I'd ask in an email reply

1. Is there a working API endpoint I can hit with a test key right now, today? Not a waitlist, not a "coming soon." An actual endpoint that returns results.
2. The Jordan Chen quote -- is that a real person I can contact, or a hypothetical testimonial based on internal testing?
3. If I buy the $99 dossier and decide to build this, what prevents the person who bought it last week from launching the same product two months before me?

## Verdict: on-the-fence

The top of the page is genuinely one of the better problem articulations I've read for this exact pain. Whoever wrote the hero copy understands the developer. But the page is trying to be two things and it ends up being neither -- I can't buy the product because it doesn't exist, and I'm not in the market for a startup idea kit. If there's a live beta, email me. If there isn't, take the Jordan Chen quote down until there is.

---
*Memo by skeptic persona, generated 2026-06-18. Studio breaks own self-grading loop.*