DeepLens is a self-hosted research engine that runs one question across academic, financial, legal, technical and social sources at once — then hands your team a single synthesised answer.
See it in action
Watch Maya — founder of a mid-market instruments company — pull competitor moves, market signals and compliance updates from a dozen sources at once. No tabs, no portals, no analyst on retainer. This plays live in your browser.
Gross margin down 3.1 pts; a new tiered price list for the mid-market line is disclosed in the filing.
412 points · 233 comments — a thread benchmarking Cobalt's new pricing against Meridian and Vantage.
Docketed order requiring revised data-retention disclosures — the compliance change rivals now have to match.
One question, five sources, one synthesised answer — the research a whole team used to do across a dozen tabs. Try it on your own competitors below.
Watch it work
Type a question, watch the source lanes light up, and read the results stream back in seconds. These are live recreations of the real engine, animated in your browser — no video, no GIF.
Type a question the way you'd say it. Intent detection reads it and fans out across exactly the sources that fit — no query language, no per-source portals.
Randomized trial: 20% reduction in major adverse cardiovascular events over 3-year follow-up in a high-risk cohort.
Phase 3 · 17,600 participants · primary endpoint: time to first MACE. Status: completed.
Every search returns a cross-source summary that reconciles what the lanes agree and disagree on — so your team reads a conclusion, with each claim traceable to its source.
OpenSanctions flags a 32% beneficial owner on the OFAC SDN list; ICIJ ties the same entity to a Cyprus shell; no matching SEC filing on record.
Stop opening portals one by one. A single search hits code hosts, the vulnerability database and community lanes at once — results in seconds, not an afternoon.
CVSS 10.0 critical. Remote code execution via crafted log messages; patched in 2.17.0.
Fully automated scanner for the Log4Shell vulnerability across nested payloads. 3.4k stars.
Watch how it works
Two short, narrated walkthroughs of the live site — a 90-second overview, and a real signed-in search fanning out across every lane at once. No fluff.
Example results
Pick a category to see a genuine DeepLens run — a synthesised AI answer over real multi-source results, free and open. Click any card to read the full result. Nothing here is staged.
Example query · artificial intelligence · 5 sources at once · free & open, no sign-in
Recent discussions highlight both enthusiasm and skepticism around AI: prominent figures like John Carmack are pursuing artificial general intelligence, while commentators such as Ted Chiang push back on claims of AI consciousness. Real-world deployment challenges are also evident—Ford had to rehire engineers after AI-driven automation produced flawed designs, and a federal judge canceled a trial after discovering both opposing legal teams had used generative AI tools improperly. Meanwhile, foundational questions persist, with ongoing technical discourse on machine learning definitions and implementation (e.g., in C#), alongside broader thought leadership on AI's trajectory and global competitiveness.
Try it yourself
This is the actual multi-source engine the platform runs — not a canned demo. Type a question and DeepLens opens the full engine on it, fanning out across every live source lane in parallel and handing back a single synthesised answer.
The sources
Most teams have never heard of half of these. Here are the 55 live sources behind one DeepLens query — a dozen business-critical lanes explained in plain English, then the full catalogue of niche and specialist sources. Availability is enforced live: any source we can't serve is hidden automatically.
Every public-company filing the SEC holds — 10-Ks, 10-Qs, 8-Ks and insider-ownership disclosures — searchable in full text.
Three of the largest scholarly indexes — Semantic Scholar, OpenAlex and arXiv — merged into one citation-aware search.
The US National Library of Medicine's index of 36M+ biomedical papers, reviews and clinical studies.
Search repositories, source files and issues across the world's largest code host in one query.
The US government's authoritative feed of software vulnerabilities, each scored for severity and exploitability.
Millions of US court opinions, dockets and oral arguments from federal and state courts.
A consolidated feed of global sanctions lists, watchlists and politically-exposed persons (PEPs).
Full-text search of granted US patents and applications via PatentsView.
The global registry of clinical studies — phase, enrollment, endpoints and recruitment status.
The FDA's open data on drugs, medical devices, recalls and adverse-event reports.
One search across the npm, PyPI and crates.io registries — versions, metadata and maintainers.
A meta web search (SearXNG) alongside Hacker News stories and their comment threads.
Investigative, government, academic, financial and fediverse lanes, grouped by category. Every tile below runs a real search on the live engine — and new lanes land regularly.
Roadmap
Beyond the live sources, these are lanes we light up per customer — open APIs, licensed feeds, or private indexes over your own data.
Open APIs we wire up quickly, licensed feeds against your own key or contract, and private indexes built on your infrastructure — where nothing leaves it. One conversation scopes what your team needs.
Who it's for
Compare what academic literature, news archives and government records say on one screen — with citations you can open in a click.
Cross-reference sanctions lists, offshore leaks, court records and corporate filings in a single sweep instead of a dozen portals.
Search code hosts, package registries and the vulnerability database together, then auto-extract the pages that matter.
Track the conversation across a dozen social and community networks alongside filings and news — the signal, in one place.
Chase a name across the fediverse, court dockets, leak archives and sanctions data — sourced, linked and ready to cite.
Validate a market in an afternoon: papers, competitors, community sentiment and pricing signals from one query.
Book a call and we'll walk your team through DeepLens on your own questions — and scope the private sources worth indexing for you.
Before you go
Sign in and run a real multi-source search on the live engine — your own question, every source at once, with a synthesised AI answer. Free to try.
Free to try · sign in to run your own question
DeepLens Premium gives your team unlimited multi-source searches, AI synthesis on every query, and the private source lanes we index over your own data.
Upgrade to Premium →Open the full engine