Nimble Reviews & Product Details

I was skeptical about the “self-learning” claim for the Web Search Agents, but after about a dozen runs on our use case (pharma competitive intel), the precision noticeably improved. I saw fewer irrelevant results and better source selection overall. It also seems to remember what good output looks like for us. Review collected by and hosted on G2.com.

No public changelog page that’s easy to find. Features appear and you learn about them from the newsletter or by accident. For developers that’s frustrating. Review collected by and hosted on G2.com.

Reliability. We run crawls nightly and I can’t remember the last time one failed for platform reasons. Error messages are clear when a target blocks you. Rate limits are generous on our tier. The discovery piece matters too, since the search agents surface sources we didn’t know to crawl. Review collected by and hosted on G2.com.

The web dashboard feels like an afterthought. Filters reset when you navigate, the job history view is slow with large numbers of runs, and exporting logs requires API calls. For a product this good the UI should be better. Review collected by and hosted on G2.com.

Data shows up in our Snowflake tables on schedule, schema'd, deduped. We didn't build or maintain any of that plumbing. Our analysts don't even know Nimble exists, they just query tables that are always fresh. Which is kind of the point of paying for this stuff. Review collected by and hosted on G2.com.

Backfills on historical data cost extra and aren't self serve. Wanted 2 years of history on a category and it required a services engagement. Review collected by and hosted on G2.com.

We evaluated a range of providers, including Bright Data, Oxylabs, Apify, Firecrawl, and Zyte. For AI agent workloads in particular, Nimble is ahead. Retrieval is optimized for token efficiency, the agent layer reduces the amount of orchestration code we need to write, and MCP support made the integration straightforward. Review collected by and hosted on G2.com.

Firecrawl is free to start, while Nimble isn’t, so we end up losing some internal hackers who just want to try something quickly. The 5,000 free pages do help, but the signup flow still adds friction compared with keyless open source options. Review collected by and hosted on G2.com.

Search, extract, crawl, map, agents. Each is good alone but together they compose into real pipelines. Our flow: map the target, crawl new sections, extract to schema, monitor for changes. All one vendor, one SDK. Review collected by and hosted on G2.com.

The SDK and the raw REST docs occasionally drift. Found a parameter in REST that wasn't in the Python SDK yet. Reported it and they fixed it within a release or two. Review collected by and hosted on G2.com.

We do regulatory research across government sites in multiple countries. The agents navigate, render JS, and pull structured results from sites that defeated our internal scrapers. Audit logs let us show exactly what was collected and when, which our compliance team loves. Review collected by and hosted on G2.com.

Scheduling options for recurring agent runs are a bit rigid. Cron works but a visual scheduler with dependencies between jobs would be better for non-engineers on the team. Review collected by and hosted on G2.com.

We monitor 3 competitors' full catalogs. Map discovers new URLs, crawl pulls the pages, extract structures them. The whole loop is one API family which keeps our codebase small. Data lands in GCS and our analysts query it in BigQuery. Review collected by and hosted on G2.com.

Crawl jobs on very large sites (1M+ pages) occasionally need babysitting. Had two stall out and support restarted them. Throughput SLAs on the enterprise plan should cover this but we're on a lower tier. Review collected by and hosted on G2.com.

Everything about the API is designed for LLM consumption. Markdown output, token-conscious retrieval, structured schemas you define. We compared with Firecrawl which is nice for quick stuff, but Nimble's governance, SLAs, and monitoring make it the enterprise answer. Review collected by and hosted on G2.com.

Honestly the signup to first successful call flow could be smoother. API key provisioning took longer than expected and the onboarding emails assumed a sales call we hadn't had yet. Review collected by and hosted on G2.com.

Daily structured feeds have run without a miss for 6 months. Validation rules catch source-side changes before bad data lands downstream. Trustworthy in the way data infrastructure needs to be. New source discovery goes through the search agents before anything becomes a feed. Review collected by and hosted on G2.com.

Job scheduling is cron-based. Fine for engineers, hostile for the analysts who want to adjust timing themselves. A simple scheduling UI on top would democratize it. Review collected by and hosted on G2.com.

We replaced a SERP API plus a separate scraper with Nimble search + extract. One vendor, one bill, one auth. The search results include rich content so we rarely need a follow-up fetch. Latency on the standard search is fine for our batch workloads. Review collected by and hosted on G2.com.

If you need real-time sub-second responses for user-facing features, test carefully. Deep search is thorough but not fast. Fine for agents, not for autocomplete-style UX. Review collected by and hosted on G2.com.