litlyzer.com

What works

There's a real product here with actual depth, not a landing page pretending. Full-PDF RAG, method extraction, knowledge graphs, collaborative writing, the arXiv/PubMed agent. And the pricing is refreshingly honest: $5 one-time, credits never expire, no subscription trap. That alone puts you ahead of half the "AI research tool" crowd.

The biggest problem

I cannot tell who this is for or what it's best at. The homepage throws 20 features across four unrelated audiences at me: PhD students doing systematic lit reviews, materials engineers running polymer selection, medical researchers analyzing clinical trials, and IP teams doing patent landscapes. A "Fuel Feasibility Engine" simulating tensile retention across 16 fuels sits three lines away from "summarize any paper." Those are different products for different people, and cramming them into one page called an "AI Engineering Research Operating System" means nobody lands here feeling it was built for them. It reads as an AI wrapper that bolted on everything rather than a tool that nails one job. The trust signals also work against you. "Trusted by top researchers & R&D teams at leading institutions" with zero logos. Three testimonials with initials-only avatars from IIT Madras, NTU and TU Munich that I have no way to verify. For a product whose selling point is research integrity and anti-hallucination, unverifiable social proof is the worst possible look.

Top suggestion

Pick one user for the homepage. The "I used to spend a weekend on lit reviews, now it's 3 hours" hook for PhD students is your clearest and most universal pitch. Lead with that, show it running on one recognizable paper end to end, and move the other 18 features to a /features page. Give the materials and fuel-compatibility engine its own separate landing page for R&D engineers, because that's a genuinely different sale and it's strong enough to stand alone.

What works

Real depth for a real pain. Grounded RAG (FAISS + BM25), method/dataset extraction, and the domain "Expert Labs" (materials, chem, petro) show you actually understand engineering research, not just generic PDF chat. Credit top-ups with no subscription and no expiry lower the barrier to try.

The biggest problem

The page tries to be everything, so it lands as nothing. Hero calls it an "AI Research Operating System," then throws 15+ capabilities and 4 different audiences (PhD, materials R&D, medical, corporate IP) at me in one scroll. As a PhD student who already uses ChatGPT and Semantic Scholar, I cannot tell the one thing LitLyzer does better than what I have, so I bounce. And for a tool selling "research integrity," bundling an "AI detector + academic humanizer" (a tool to make AI text pass as human) reads as the opposite of integrity and quietly kills trust.

Top suggestion

Pick one user and one killer job for the hero (e.g. "Systematic lit reviews for PhD students, days to hours") and prove it with a real before/after output, not a feature grid. Move the other disciplines to sub-pages. And either reframe or drop the "humanizer" next to your integrity claim, they contradict each other.