Hands-on and document-based reviews of individual AI tools, with clear evidence, limitations, pricing, and alternatives.
Posts in this category are hands-on evaluations written for practitioners comparing tools — what each product is good at, where it falls short, and how it fits next to alternatives already in the directory. See our editorial policy for how we review and rank.
By Leo Liang
Anonymous Felo search answered a three-part Argentina fact check in about 21 seconds, caught the Aconcagua-in-Chile trap, and rewrote the answer in Spanish with citations kept. Signup stopped on an image captcha. Final rating: 4 out of 5.
By Leo Liang
I wrote a two-page darkroom handbook, hid a trap question in it and asked ChatPDF's free tier four questions. It got all four right and refused to invent a price, but one cross-page answer cited the wrong page for half its maths.
By Leo Liang
Cleanup.pictures let me wipe SAMPLE WATERMARK, FOR SALE — DEMO, and REMOVE ME off a watch photo in about six seconds of server time — then handed back a 720×523 free download with the HD gate spelled out up front.
By Leo Liang
I watched Gamma polish slides 5/7 on a Sunday night in La Rioja. Twenty seconds later I had a seven-card ToolifyNav pitch I could edit, present, and export to PDF on the free tier.
By Leo Liang
Card-free Humata signup, 18s PDF ingest, ~2s answers that matched the paper, and citation clicks that highlight the source. Free caps and Team-only OCR keep it at 4/5.
By Leo Liang
Free Clipdrop cutout on a watch photo in a few seconds — sharp mask, watermark-free PNG after SD downscale. Relight crashed, no credit meter, Turnstile mid-job. Final rating: 3 out of 5.
By Leo Liang
Free Grammarly web Docs: 56-word paste, score 55→61 after weather→whether and a weird its→it accept. No card. Loud Try Pro for $0. Final rating: 4 out of 5.
By Leo Liang
I highlighted a Transformer abstract span on Explainpaper’s sample paper, got a plain-language explain in about five seconds, then a useful follow-up — Free, no credit wall. Final rating: 4 out of 5.
By Leo Liang
I tried ChatCSV with email-only login and a 5-row CSV. The magic link never arrived, demo upload failed, and two queries produced no new visible result. Final rating: 2 out of 5.
By Leo Liang
I ran a falsifiable intermittent-fasting question through Consensus: email signup, ~13s Pro synthesis, Consensus Meter N=20, Study Snapshot, then Free vs Pro pricing. Final rating: 4 out of 5.
Weekly digest of new AI tools, reviews, and early access to featured listings.