NewsCatcher

Recall-first web search API for AI & enterprise (YC S22)
86.8% recall | 2B+ web pages | 100+ countries

Built for when missing events isn't an option.

[NewsCatcher @newscatcherapi]

Full dataset: https://t.co/E7L8A02ObV

[NewsCatcher @newscatcherapi]

Is the U.S. office market recovering or repricing? Look at the deals.

83 commercial office transactions by real estate investment firms in the past 30 days — pulled from 8,773 pages, with property, deal size, currency, investor, and date already structured.

A CRE analyst's week of work in one query.

Query of the Week ↓ in the comments.

[NewsCatcher @newscatcherapi]

Tutorial → https://t.co/qL6USPvJiU

[NewsCatcher @newscatcherapi]

Enterprise risk software tracks known risks well.
But most risks start outside your organisation: in supplier news, court filings, breach reports, and local press.

A web search API adds the external layer risk teams are missing.

New post: how to build an early warning risk intelligence system with CatchAll → link in the comments.

[NewsCatcher @newscatcherapi]

We wrote a practical breakdown of both approaches — real trade-offs, hidden costs, and a decision framework for your team.

👇 https://t.co/4Hzx9GJOJy

[NewsCatcher @newscatcherapi]

Building a custom scraper seems like the cheap option.
Then you add proxies, CAPTCHA handling, headless browsers, parser fixes every time a site changes its layout.

The real cost shows up 6 months in🧵

[NewsCatcher @newscatcherapi]

Custom scrapers make sense when:
→ You need logins or session handling
→ Workflows are small + sites rarely change
→ You need full control over timing and logic

Managed APIs win when:
→ Scale grows and infra becomes the job
→ You need clean JSON for AI/analytics pipelines
→ You want to ship fast without debugging crawlers

[NewsCatcher @newscatcherapi]

The full dataset: https://t.co/meYFdaMBFU

[NewsCatcher @newscatcherapi]

♠️ Tracking new gambling and casino licenses worldwide is a compliance nightmare.

We ran the query anyway: 15,982 pages → 39 license approvals from the last 30 days, with the licensee, jurisdiction, regulatory authority, license type, and approval date pulled into clean fields.
Compliance, gaming investors, and payments/KYC teams — this is the kind of dataset that usually takes a small research team a week.

Grab the full dataset via the link in the comments. ↓

[NewsCatcher @newscatcherapi]

3/3
10 signals worth monitoring programmatically:

→ Funding rounds & deal sourcing
→ M&A announcements
→ Executive hires & departures
→ Layoffs & restructuring
→ Product launches
→ Regulatory developments
→ Partnerships
→ Competitor activity
→ Distress indicators
→ Emerging industry trends

All appear on the open web weeks before structured databases catch up.

Full breakdown + how to build the pipeline:
https://t.co/Gxi2RuDehM

[NewsCatcher @newscatcherapi]

By the time a funding round hits Crunchbase, the window to act has usually closed.

We wrote a blog on how Investment teams are building web search APIs into their stack to close the information gap 🧵 1/3

[NewsCatcher @newscatcherapi]

2/3 Financial databases are built around disclosure events: filings, earnings, official press releases.

But the early signals? They live on regional news sites, niche trade blogs, and company IR pages. Scattered, fragmented, no aggregator catches all of it.

[NewsCatcher @newscatcherapi]

The full dataset: https://t.co/OzGVIFAwXK

[NewsCatcher @newscatcherapi]

🤌 If you build with agents, you need to see this.

86 multi-agent framework releases in 30 days, pulled into clean columns: framework, version, org, date, key features.

The agent updates, structured.
Query of the Week ↓ link in the comment.

[NewsCatcher @newscatcherapi]

Our Substack:
https://t.co/CkvjrMH9tf

[NewsCatcher @newscatcherapi]

🤫 Pssst... do you want to be faster than your competitors? Then read us on @Substack!

May updates from NewsCatcher: reusable webhooks that route results anywhere, project containers to organize work by client or team, and a free query validator that catches issues before they cost you.

And *secret new feature* we're working on, that allows you to process the primary documents — PDFs, filings, legislative votes — before journalists (and your competitors) do.

→ Read the full May issue on Substack – link in the comments.

[NewsCatcher @newscatcherapi]

The full tutorial (working Python code, prompts, architecture) is live.
The retrieval layer is where most agents quietly fail. This one doesn't.

🔗 https://t.co/ruqTMGh5JV

[NewsCatcher @newscatcherapi]

How “deep” does your deep research agent actually go?It's probably just summarizing the top 10 search results.

Here's how to actually fix that 🧵

[NewsCatcher @newscatcherapi]

We built it.
One open-ended question → orchestrator breaks it into focused sub-tasks → sub-agents each run a CatchAll search → cited markdown report comes out the other end.