Seltz

Domain filtering is now live on the Seltz search API: two optional parameters, include_domains and exclude_domains. Allowlist the publishers you trust, exclude the ones you don't, or pass neither and filter downstream off the URL returned on every result. When an agent reasons over the news, the sources it reads decide whether it's working from signal or noise. A claim from a primary wire and the same claim from a press-release aggregator are not the same input, even when the words line up. Here's an example: SpaceX's compute deal with Anthropic. SpaceX filed its S-1 on May 20, and the $1.25-billion-a-month price buried in that filing got echoed across hundreds of aggregator reposts before the June 12 NASDAQ debut. An agent tracking that story in real time should read the wires and skip the reprints. Learn more: https://lnkd.in/eghKEF4d

Seltz is live on Zapier. Real-time news, web search, and people data are now available as actions in any of Zapier's 8,000+ apps. No code. Three workflows to try first: → Brand and competitor monitoring in Slack, within the hour → CRM lead enrichment with this-week's company news → Scheduled topic digests for any team that needs to stay current Free to set up. Works alongside our n8n integration with more platforms on the way. Get started here: https://lnkd.in/eP2Aht_q

Date filtering is live on the Seltz News API. One of our customers wanted results constrained to a narrow time window; sometimes the last 24 hours, sometimes a specific day in the past. Consider an agent tracking Broadcom's earnings call on June 3 next week. It doesn't need analyst previews from last week. It needs the reaction in the 24 hours after the print. Two new optional parameters on the search endpoint, filtered on publication date. If there's a filter you want, drop it in the comments! Learn more: https://lnkd.in/evxuF9jD

Congrats to Jason Grad and the team at Massive for releasing their Web Render API. We believe everyone should have access to top tier research tools. Jason and the team are delivering an API that gathers tough data so everyone LLMs can conduct research more easily. Looking forward to taking this for a spin.

We just shipped a Seltz integration with Orthogonal (YC W26). If you're running agents or RAG and you've been stitching together search APIs, HTML parsers, and caching layers to get usable web context — this collapses a lot of that. One endpoint on Orthogonal, query in, full documents out with source URLs. Not snippets, not links to go fetch yourself. The actual content, structured the way an LLM wants to consume it. Real-time too, so your agent can answer questions about stuff that happened this morning instead of whatever was in the training set. x402 works out of the box if your agents are paying per call in USDC. No keys to rotate, no billing plumbing. $10 in Orthogonal credits to try it (link in the comments)!

Today we're releasing our MCP server. Give your favorite LLM access to real-time US news. Connect Claude, Cursor, or any MCP-compatible tool. Here's a video of it in action. Fresh news should be available everywhere you build! Get your API key here: https://lnkd.in/efeaDYFW Configure the MCP here: https://seltz.run.tools/

Our API for real-time US news is open to the public! Do your agents need real-time access to the web? Try us: https://lnkd.in/efeaDYFW Sign up this week and get $100 in API credits to build with. We spent months building our infrastructure from scratch. It was hard, but worth it. Every day we benchmark ourselves against competitors. Our results on April 11th: - 250 millisecond retrieval latency (7x faster than competitors) - Higher answer accuracy than every provider - No limits on what you can retrieve Our CEO Antonio Mallia wrote about why we built from scratch, started with news first, and where we're headed: https://lnkd.in/e6uGZqMU

Searching the web at inference time = training a model 5× larger. No extra training. No bigger weights. Just better retrieval. The implication is simple: pre-training is leaky. Models don't fully absorb everything they're trained on — and retrieval at query time recovers what's left behind. But here's what the paper doesn't say explicitly: not all search is equal. The multiplier scales with corpus quality. Stale crawls, noisy results, and slow latency all cap your gains. Generic search APIs weren't built for LLMs. That's exactly what we're solving at Seltz — purpose-built web search infrastructure for AI agents and LLMs. Real-time indexing. High-precision retrieval. Designed for inference-time performance. Swipe to see the numbers → 🔗 Paper: https://lnkd.in/dvTi_Ybz 🔗 https://seltz.ai/

Congrats to the depthfirst team on their $80M Series B — just 90 days after their Series A. We asked Claude about the round. It had no idea. Then we asked with Seltz doing the research. Funding round, investors, sources, all there. Same model. Different result. If your agent needs to track your favorite companies' successes in real time, talk to us. Congrats again Daniele Perito, Andrea Michi, Qasim Mithani, onward!

We're hiring at Seltz. We're building the search and data infrastructure that AI applications run on — structured, LLM-ready data at scale. We're a small team moving fast, and we're looking for people who want to own things end to end: → Research → DevRel → Backend → Frontend → AI → Infra → GTM → Product We're fully remote. Early stage means real ownership and real impact on the direction of the product.