Bespot (Gatekeeper)

Bespot (Gatekeeper) — The licensable byproduct is a cross-enterprise corpus of AGENT BEHAVIOUR verdicts, not merely crawler hits. Gatekeeper evaluates every incoming session in real time with no sampling, matching against behaviour profiles as well as known agent, crawler, scraper, headless-browser and automation-framework signatures, then returns a verdict carrying agent name, category and classification details - or an explicit 'no known automated agent detected' rather than a guess. The distinctive scope is that it watches agents ACTING, not just crawling: the product targets automated tra…

Bespot (Gatekeeper) offers (Alternative, Sentiment, Reference) — Potentially the largest event corpus in this cluster, because detection is stated as full-coverage with no sampling across enterprise customers' entire inbound session traffic - so volume scales with customer traffic rather than with bot hits alone, plausibly millions of classified sessions monthly. Retention period and whether full session features are stored or only the verdict are both undisclosed and are the deciding questions for any licence..

Behavioural training data for agent detection - session-level fingerprints of headless browsers and automation frameworks, which is richer than user-agent matching and directly useful for fraud, bot-mitigation and trust-and-safety model training; AI-agent interaction modelling, since form-filling, chat and API-driving behaviour is the missing dataset for anyone building or studying agent traffic; agent-consumption cost attribution for teams running public AI endpoints; LLM-training-harvest detection, i.e. labelling which corpora and documentation sets are being swept for training data; and negative-class data for human-versus-agent classifiers, since the honest-unknown verdicts expose the current frontier of detectability.

Coverage spans Europe, Other; Application Software, Data Processing & Outsourced Services; alternative, sentiment, reference; equities.