Enter one public clue
Start with an authorized domain, web address, company, email, file or another supported target type.
How it works
Tigess prepares a bounded plan, collects authorized public sources and keeps every saved result traceable through review and delivery.
Investigation lifecycle
Each step preserves scope, source context and the reason a data source ran—or did not.
Start with an authorized domain, web address, company, email, file or another supported target type.
Review automatic classification, correct ambiguity and choose how broad the research should be.
See compatible data sources, gaps, safety boundaries and the estimated credit range before confirming.
A bounded research run gathers public signals and keeps the original source references connected to normalized results.
Compare direct observations, candidates, contradictions, confidence and useful negative results.
Deliver reviewed findings or use workspace-scoped developer access without losing source identity.
Coverage
Coverage grows cumulatively within explicit compatibility, source, depth, time, safety and credit ceilings.
A focused core for validation and immediate context under the tightest limits.
Broader passive discovery for questions that need more context across compatible public sources.
Deeper passive and configured-provider research for complex questions and wider associations.
Maximum authorized breadth, including separately approved active data sources when every additional gate passes.
quick · 78 registry-eligible modulesbalanced · 151 registry-eligible modulesdeep_research · 195 registry-eligible modulesadvanced_investigation · 202 registry-eligible modulesPrepared plan
Previewing is planning-only. Execution begins after explicit confirmation and an account-credit reservation.
Enter one authorized public clue. Tigess detects its type and prepares a compatible investigation plan.
Previewing a plan does not contact data sources, start a research run or consume credits.
Larger research runs progress through bounded background work while status stays visible.
A missing provider configuration does not crash the entire run or silently change the rest of the plan.
Reservation happens before work; settlement follows the data sources actually run.
Not found, not applicable, not assessed, unavailable and timed out remain distinguishable.
The worker executes the persisted engine-authored plan under module, provider, fan-out, runtime and credit ceilings. Durable invocation ledgers preserve idempotency and one settlement per invocation.
Start with a target
Create an account to plan your first evidence-backed investigation, or talk to us about team and workflow requirements.