# Marcus Yuen, Independent Software Consultant — read of LexiRisk, June 17 2026

> 9 years freelancing for SaaS companies, currently juggling 4 retainer clients, sign 8-12 contracts a year and hate every minute of it. Coach my daughter's U10 soccer team Saturday mornings.

## How I got here

Got a vendor agreement from a new client last week that had a non-compete section I couldn't parse. Googled "non-compete clause freelancer software consulting what does this mean" and ended up in a Reddit thread where someone dropped a link to this. Not an ad. Just a comment. That's about the highest trust vector I have anymore.

## What I clicked first

The comparison table stopped me. Specifically the column that said LexiRisk's limitation is "Risk flagging only" under "Custom negotiation." I appreciated that. Every other tool like this pretends it replaces the lawyer entirely. Calling it a "first read" before you call the lawyer is actually the honest framing, and they said it plainly: "LexiRisk is your first read. When you find a real red flag, you can then call a lawyer knowing exactly what to negotiate." That's the right pitch. I've signed contracts I should have flagged first, then paid $600 to confirm what I already suspected.

## Where I paused

"Negotiation Tips. For each flagged clause, we suggest common counter-terms so you know what to push back on." I stopped on that. That's actually the feature I'd pay for, not the risk score. I don't need a number. I need to know what to say when I push back. But the page doesn't show me what that looks like. Not a single screenshot of a real report. Not a sample flagged clause with the suggested counter-term next to it. That's the hole in the page. I'm supposed to trust this feature exists based on one sentence.

## What I distrusted

Two things. First: "Join 2,000+ founders and freelancers who've used LexiRisk." No quote. No name. No company. No tweet. Just a number sitting there. I've seen that number on probably 30 landing pages. It could mean 2,000 free trials with zero conversions, or it could mean real paying customers. The page doesn't distinguish. Second: "Our model is trained on 10,000+ legal contracts." That sentence is on every AI legal product I've looked at in the past two years. It tells me nothing about how it performs on the specific contract I actually care about, which is a consulting services agreement with a California governing law clause and a pretty aggressive IP assignment.

## What would convince me

One real example. Not "what LexiRisk catches" as a generic bullet list. I want to see a real clause, the AI's flag, the plain English translation, and the suggested counter-term, side by side. Ideally from an NDA or a consulting services agreement since that's my world. Show me one page of actual output. If the report looks like a wall of GPT-generated legalese I'll close the tab. If it looks like something a smart paralegal wrote for a non-lawyer, I'll upload a contract right now.

Also: what jurisdiction does the model know best? California freelancers face different clause norms than New York ones. Does it know that? I have no idea.

## What I'd ask in an email reply

1. The negotiation tips feature: can you show me a real example of what a suggested counter-term looks like for, say, an aggressive IP assignment clause? Not a description of it. The actual output.

2. How does it handle contracts where the risky part is what's *missing*, not what's buried? Like a services agreement with no limitation of liability clause at all. Does it flag absence of protection, or only presence of bad terms?

3. The 94% accuracy claim: what's the test set? Is it public contracts, standard NDAs, venture deals? Or is it a mix that happens to skew toward easy cases?

## Verdict: on-the-fence

The framing is right and the price is genuinely low enough that I'd try it on a throwaway contract. But the page has no proof of output and the social proof is thin. I'm not skeptical of the idea, I'm skeptical that this specific implementation delivers on the negotiation tips claim, which is the one thing that would actually change my workflow.

---
*Memo by skeptic persona, generated 2026-06-17. Studio breaks own self-grading loop.*