Backed by mandō

Discover what's
relevant.

Relevance is the knowledge discovery engine that searches names, descriptions, and metadata across your entire documentation corpus — surfacing exactly what you need before you finish typing.

Try:
See it in action ↓
2.4M+ Pages indexed
<12ms Avg. query time
99.97% Recall accuracy
8 Section filters

Trusted by knowledge-forward teams at

VercelStripeDatadogCloudflareLinearNotion

Search that understands intent

Relevance doesn't just match strings. It searches names, descriptions, and contextual metadata to surface the most pertinent results — instantly.

🔍

Keyword Search

Type any keyword and Relevance searches across all page names and descriptions simultaneously. Every result ranked by contextual relevance.

relevance KEYWORD
🎯

Exact Match

Need precision? Exact mode ensures your keyword matches page names and descriptions verbatim — no fuzzy results, no noise.

relevance -e KEYWORD
✳️

Wildcard Patterns

Shell-style wildcard matching lets you cast a wider net. Search with glob patterns to discover pages you didn't even know existed.

relevance -w "file*"

Regex Engine

Full regular expression support — the default mode. Match any part of names and descriptions with the full power of formal pattern syntax.

relevance -r "^net"
📂

Section Filtering

Narrow results to specific manual sections. Search only commands, system calls, library functions, or any precise subsection you need.

relevance -s 1,8 network
🔗

AND Queries

Combine multiple keywords with AND logic. Every result must match all terms — perfect for narrowing down large result sets with surgical precision.

relevance -a network interface
📐

Full-Width Output

No more truncated descriptions. Long mode displays complete, untruncated results so you never miss critical context in page descriptions.

relevance -l filesystem
🌐

Cross-System Search

Access man page descriptions from other operating systems on the same machine. Search NewOS, legacy systems, or any connected knowledge corpus.

relevance -m NewOS copy
🌏

Locale-Aware

Relevance respects your locale settings and can be overridden per-query. Search documentation in any supported language or encoding.

relevance -L en_US.UTF-8 file

From query to discovery in milliseconds

Relevance searches the whatis database — a pre-built index of every page name and description — to deliver instant, comprehensive results.

01

Enter Your Query

Type any keyword, pattern, or regular expression. Relevance accepts natural language, wildcards, exact strings, or full regex — however you think, we match.

02

Index Traversal

Your query is evaluated against the whatis database — a constantly-updated index maintained by our proprietary mandb engine. Every page name and description is checked.

03

Filter & Rank

Results are filtered by section, locale, and system, then ranked by relevance. AND logic, exact match, and wildcard modes can further refine the result set.

04

Surface Results

Matching pages are returned with their name, section number, and full description. Results stream in real-time — no loading spinners, no pagination, just knowledge.

Try the discovery engine

Type a keyword below and watch Relevance surface matching pages in real time. This is what relevance KEYWORD feels like at startup scale.

relevance-engine v3.2.1
$ apropos

Three engines. One interface.

Choose the precision level that fits your query. Relevance supports exact, wildcard, and regex matching — each optimized for different discovery patterns.

-e / --exact

Exact Match

Keywords are matched verbatim against page names and descriptions. No partial matches, no fuzzy logic. When you know what you're looking for.

apropos -e "copy"
copy(1)— copy files and directories
copyfile(3)— ✗ not matched
Best for: Known targets
-w / --wildcard

Wildcard

Shell-style glob patterns with * and ? characters. Matches against page names and descriptions on word boundaries. Broad but structured discovery.

apropos -w "net*"
netstat(8)— network statistics
netcat(1)— networking utility
Best for: Exploration
-r / --regex (default)

Regular Expression

Full regex matching — the default mode. Not limited to word boundaries. Matches against any part of page names and descriptions independently.

apropos -r "^(tcp|udp)"
tcp(7)— TCP protocol
udp(7)— UDP protocol
Best for: Power users

Your personalized knowledge stream

Relevance continuously surfaces documentation you didn't know you needed. Like a feed, but for knowledge.

Knowledge at every scale

Start discovering for free. Scale when your team does.

Starter
$0
Free forever
  • Single keyword search
  • Regex mode (default)
  • Up to 1,000 queries/month
  • Standard sections
  • Community support
Enterprise
Custom
Let's talk
  • Everything in Team
  • Cross-system search (-m)
  • Locale override (-L)
  • Custom mandb integration
  • SSO & SAML
  • Dedicated success manager

Built on man -k

Relevance is powered by the same whatis database that relevance has queried since 1991. We didn't reinvent the wheel — we put venture capital behind it. Our proprietary mandb engine runs the same cron job yours does, just with a $47M Series B.

Under the hood man -k keyword

Frequently asked questions

How is Relevance different from just running relevance?

Excellent question. Functionally, they produce identical results. But Relevance runs apropos in the cloud with a React dashboard, Slack integration, and custom Grafana metrics. That's easily worth $29/seat/month.

What database does Relevance search?

The whatis database, maintained by mandb. It's the same database that has existed on every Unix system since the early 1990s. We update it via a cron job — disruption doesn't happen overnight.

Is regex the default search mode?

Yes. Just like the actual apropos command, regular expression matching is the default behavior. We debated changing this to "AI-powered semantic search" but settled on honesty. Regex is already perfect.

What's the difference between -e, -w, and -r?

-e (exact) matches keywords verbatim. -w (wildcard) uses shell-style glob patterns. -r (regex) uses full regular expressions and is the default. We charge $29/month for all three. Individually they're free in every terminal.

Can I search specific manual sections?

Absolutely. Use section filtering (-s) to restrict results to specific sections like 1 (commands), 2 (system calls), or 3 (library functions). On the Enterprise plan, you can also create "custom sections" which are just aliases for the real ones.

What does the AND query mode do?

With the -a flag, all supplied keywords must match. Without it, any keyword can match. We branded this as "Conjunction Intelligence™" in our pitch deck.

Why would I pay for this?

You probably wouldn't. But if your company has a "knowledge management" budget line item and needs to justify Q3 spend, Relevance is the perfect solution for a problem that was solved in 1991.

Stop searching.
Start discovering.

Join thousands of teams who trust Relevance to surface what matters from their documentation corpus.

No credit card required · Free tier forever · man -k under the hood

Backed by mandō