DeepLens: One Search Box Over 40+ Sources
Research rarely lives in one place. Checking a company or a topic can mean academic databases, SEC filings, patent and court records, sanctions lists, news, code hosts and community forums, each with its own search box. DeepLens puts them behind one. Its engine page lists 40+ sources in 13 categories, including Academic (Semantic Scholar, OpenAlex and arXiv, CrossRef, PubMed, bioRxiv), Financial (SEC EDGAR, CoinGecko), Government & Legal (USPTO patents, US Congress, CourtListener, GovInfo), OSINT (OpenSanctions, ICIJ Offshore Leaks), News & Media (web search, DuckDuckGo, Hacker News, Wikipedia, the Wayback Machine), Technical & Code (GitHub), Security (NVD), Health (openFDA, ClinicalTrials.gov) and Social & Community (Stack Overflow, Mastodon, Lemmy and others). A sidebar picks which sources a search uses, with controls for results per source and sort order. The page states the pricing in credits: a standard search costs 1 credit, and a Deep Research mode runs a multi-step pipeline into a longer cited report, billed by length. For this case study the homepage and the engine page were checked on the live site; no search was run, so it makes no claim about the quality, ranking or speed of results.
Key Metrics
Source coverage
Before
One search box per source
After
40+ sources in 13 categories listed on one engine page, checked on the live site
Control
Before
Fixed search scope per tool
After
Pick sources per search, choose results per source and sort order
The Challenge
Research does not live in one place. Understanding a company, a market or a technology means checking papers, filings, patents, court records, news and community discussion, each in a separate tool with its own query syntax and result format. A researcher spends as much time switching between those tools as reading what they return. What is missing is one front door: a single search box over many sources, with control over which sources each question goes to.
Why This Approach?
Many sources, one interface
40+ sources in 13 categories behind a single search box
Keeping the sources in one interface removes the tab-switching between research tools; the categories keep a long list navigable.
Alternatives considered: One tool per source, a general web search only
Who chooses the sources
A sidebar where the user selects exactly which sources a search uses
Different questions need different sources. Letting the user pick (with select all and clear) keeps each search to the sources that fit.
Alternatives considered: Always search everything, or a fixed set
How usage is priced
Credits, as stated on the engine page: 1 credit per standard search; Deep Research billed by length
A per-search unit makes the cost of each question visible before it is asked.
Alternatives considered: Unlimited flat plans
Build Process
Source catalogue
40+ sources in 13 categories: Academic, Financial, Government & Legal, OSINT & Intelligence, News & Media, Technical & Code, Security & Cyber, Health & Safety, Science & Environment, Package Registries, Jobs, Entertainment, Social & Community.
Engine page
One search box, a sidebar to pick sources, results per source (5, 10, 20 or 30), sort by relevance or newest, and suggested queries.
Standard search and Deep Research
As the page states: a standard search costs 1 credit; Deep Research runs a multi-step pipeline (search, extract, follow links, synthesize) into a longer cited report, billed at 1 credit per 4 pages, minimum 1. Neither was run for this case study.
Challenges & Solutions
Impact
Papers, filings, patents, court records, news and forums each need their own search.
Solution
One search box over 40+ sources in 13 categories.
Result
The engine page lists all 40+ sources in one sidebar, checked on the live site.
Results & Metrics
Source coverage
ImprovedBefore
One search box per source
After
40+ sources in 13 categories listed on one engine page, checked on the live site
Control
ImprovedBefore
Fixed search scope per tool
After
Pick sources per search, choose results per source and sort order
System Overview
The architecture below is the technical foundation behind every business outcome on this page — built for reliability, low running cost, and the speed your customers feel.
Public web app. The engine page lists 40+ sources in 13 categories, lets the user select which sources a search uses, and offers results-per-source and sort controls; on load it reads a source-health endpoint. The page states its pricing in credits (1 per standard search; Deep Research billed by length). Checked on the live site: the homepage, the engine page, the source list and its controls. Not exercised: running a search or Deep Research, sign-in and subscription, so no result quality, ranking or speed is claimed here. The backend stack was not re-verified for this revision and is not named.Technologies Used
Public research web app over 40+ selectable sources (backend not named)