Landbase
Toolkit · AI Qualification
← All tools
T6 · Tool

AI Qualification

Visits a candidate's website (or runs a targeted search), reads the live content, and answers a qualification question with Yes / No / Unknown plus a cited evidence quote.

Examples

How customers use it

01
Distinguishing similar-looking companies
Spell out an ICP like "freight forwarders, not 3PLs" — a distinction no column captures cleanly. AI Qualification visits each candidate's website, reads how they describe themselves, and returns Yes / No / Unknown with the evidence quote that proves it.
02
Verifying compliance and certifications
Require "SOC 2 certified providers" or "HIPAA-handling clinics." AI Qualification searches each candidate's site for cert language, trust badges, and compliance pages — and returns the cited evidence for the audit trail.
03
Detecting product-fit signals on the site
Restrict to companies running a specific competitor or feature live. AI Qualification visits the candidate, reads product pages, integrations lists, and customer logos, and returns Yes / No with the quote that proves it.
04
Catching hiring and expansion intent
Specify "actively hiring SDRs" or "expanding into EMEA." AI Qualification reads job listings, press releases, executive bios, and recent posts — and returns the answer with evidence, the same way a sharp SDR would qualify a prospect.
Pairs with
T5Enrich T1Semantic Search T4Match T8Advanced Dataset Generator
Found in
TAM Mapping Always-On Lists Complex Queries Seed Expansion
Under the hood+
How it's used

For signals that resist filter logic. Emulates how a sharp SDR/BDR would qualify a prospect: looking for specific keywords on the site, affiliated people, product details, customer logos, hiring posture — but at scale across an entire list.

Why it matters

The justification quote is the audit trail. A deliverable is defensible at the record level — a salesperson can show the customer exactly which line on the prospect's website proves the fit.

The Web Research Agent
Entry URL
or search query
→
Visit Entry Page
render + extract
→
Navigate Salient Links
LLM picks targets
→
Crawl & Gather
read page content
→
Form Answer
Yes / No / Unknown + quote
The LLM stays in the loop at every step. When the answer isn't yet complete, the agent loops back to "Navigate" and follows another path.
Stage by stage
  • Entry point. Either a specific URL or a search engine query. Either way, the agent navigates from there — it isn't a generic site crawler.
  • Salient-link navigation. The LLM scans the page, picks the most relevant links given the research question, and follows them. Goes deeper into "About," "Services," "Customers," "Compliance" pages — whatever matches the question.
  • Crawl & gather. Reads the content from each followed page. Loops back to navigate further when the answer isn't yet confident.
  • Form answer. Returns a structured boolean (Yes / No / Unknown) plus an evidence quote from the source. The quote is the audit trail.
Operational controls
  • Three invocation modes — existing-field eval, general web research, directed company_url research — auto-inferred from the prompt.
  • Operator-reviewed prompts — every prompt is read verbatim before a scale run fires.
  • Per-tier precision sampling — stratified spot-checks with named thresholds (T1 ≥ 85%, T2 ≥ 70%, T3 ≥ 55%) before a list is promoted.