G2

Alessio B.

Web Developer

Small-Business (50 or fewer emp.)

8/26/2026

"Free page fetching that made my own AI news tool viable"

4.5/5

What do you like best about TinyFish?

I built my own AI news-monitoring and research tool, and Fetch is the piece that made it viable. I hand it a URL and get back clean Markdown instead of a wall of nav bars, cookie banners and ads, which cuts my token cost on every LLM call downstream. Search and Fetch being free is genuinely rare — everyone else meters exactly this. Integration was trivial: one API key, plain REST, no SDK to adopt. Failed URLs not counting against quota is a small thing that matters a lot when you're pulling from news sites that move or drop content.

Over my last 100 fetches the success rate is 98%, with an average latency of 2.7 seconds. Most pages come back in under two seconds, which matters when I'm processing a batch of feed links in one go.

Onboarding was about as short as it gets: sign up, copy the key, first successful fetch inside five minutes, no credit card in the way. And the output is built for the job — Fetch returns content already shaped for an LLM to read, so the AI layer of my own tool gets a clean input instead of raw HTML it has to fight through.

The dashboard deserves a mention too. Success rate, average latency, extracted content size and credit usage are all on the landing page for each API — I went looking for my failure rate and had it in one click, without digging through logs. For a free tier that's more observability than I expected. Review collected by and hosted on G2.com.

What do you dislike about TinyFish?

Fetch is unreliable on paywalled and aggregator pages, and unpredictably so. My success rate is 98% over the last 100 fetches, but the failures aren't random: a paywalled article on The Athletic, and a press-review page on an Italian sports aggregator. What makes it awkward to design around is that four other articles from the same publication went through fine in the same window. So I can't route by domain — I have to handle failure per URL and decide at runtime whether a source is usable. Clearer signalling on why a fetch failed would help more than a higher success rate would.

Second, the free tier ends exactly where the platform gets interesting. Search and Fetch cost nothing, Agent and Browser need a funded wallet. There's no middle step, so one stubborn source pushes you from free to pay-per-step.

Related to that, I'm running production tooling on a free tier with nothing stated behind it — no SLA, no published commitment that Search and Fetch stay free. I'd rather see a policy than assume it holds. Review collected by and hosted on G2.com.

What problems is TinyFish solving and how is that benefiting you?

I pull transfer-market news links from RSS feeds on a set of selected sources — Italian, German and English sports sites — and then use Fetch to turn each article into clean content my LLM can actually reason over. Before, the choice was either hand-rolled scraping I had to maintain per site, or dumping raw HTML into the model and paying tokens for nav bars and cookie banners. Fetch removed both problems in one call, and it renders in a real browser so JS-heavy pages come through intact rather than half-empty.

The second use is research. I use Search plus Fetch instead of Perplexity for digging into a topic — same result, but inside my own pipeline, where I control the prompt, the sources and the output format, and with no API cost on the retrieval layer. That's the part that changed how I work: the retrieval layer stopped being a budget line I had to think about, so I run it on every article in the feed instead of picking which ones are worth the spend. Review collected by and hosted on G2.com.