Every GTM platform lets you filter companies by industry, size, and location. That works when your target market maps cleanly to a category. But most interesting markets don’t.“Companies building AI tools for developers” — that’s not an industry. “B2B SaaS companies with product-led growth motions” — no checkbox for that. “Mid-market fintechs that process payments” — you could try filtering by “Financial Services” but you’d get banks, insurance companies, and wealth managers alongside the payment processors you actually want.Search by Meaning lets you describe what you’re looking for in your own words and returns companies that match the concept.
You type a description of your target market in plain language. The system converts that description into a mathematical representation of its meaning, then finds companies whose own descriptions are closest in meaning — regardless of whether they use the same words.“AI developer tools” matches companies that describe themselves as building “machine learning SDKs,” “LLM infrastructure,” or “model training platforms.” None of those phrases appear in your query. The system understands they’re the same concept.
Show What's happening under the hood
Your phrase is converted into an embedding vector by a dedicated ML model. Every company in Landbase’s dataset also has a pre-computed embedding of its description and keywords. The system performs an approximate nearest-neighbor search to find the companies closest to your query in meaning-space. Those “seed” companies are then expanded through a pre-computed similarity graph, producing up to 20,000 scored results ranked by how closely they match your description.
Keyword search is literal. If you search for “developer tools,” you get companies whose descriptions contain the exact phrase “developer tools.” You miss companies that say “software development kits,” “engineering productivity platform,” or “code infrastructure.”Semantic search understands meaning. It knows these all describe the same market. The matching happens in a mathematical space where proximity equals similarity — not at the string-matching level.
“autonomous vehicles” matches companies working on “self-driving cars,” “ADAS technology,” and “vehicle automation” — same concept, completely different words
“companies selling to HR teams” matches companies whose descriptions mention “talent acquisition platform,” “employee engagement software,” and “people analytics” — the system understands these sell to HR even though none of them say “HR”
“B2B payments infrastructure” matches payment processors, billing platforms, and accounts payable automation — but not consumer payment apps or general banking software
Semantic search doesn’t replace your firmographic filters — it layers on top. You can say “AI developer tools” AND restrict to:
Series A–C funding
50–500 employees
US and UK headquarters
The semantic signal determines which companies match your concept. The filters constrain the universe. This means you get precise targeting that neither approach could achieve alone.
A single natural language phrase expands to up to 20,000 scored companies. This is production-scale audience building from a description. The scoring is deterministic — same input always produces the same output — and results are pre-ranked by relevance so the best matches are at the top.
