Every output is compliance-checked before delivery
Every customer-facing output is run through a structured compliance review before it leaves us. The review enforces a fixed rule set: no 'advice' or 'recommendation' language, no claim of regulatory authorisation we don't hold, no phrasing that misrepresents what we are. Anything that fails the check routes to human review, not the customer. PensionHunter is research-only, we identify pension entitlements and provide an Information Request Authority (IRA) that lets you authorise the information request yourself. We do not provide financial advice. The review layer is what enforces that boundary on every output.
Every action is logged
Every agent invocation, every confidence rating decision, every refund decision, every customer email sent, all of it is logged. The audit log is what makes the operation defensible to regulators and verifiable for our customers. If a question is ever raised about how a finding was produced or why a refund was triggered, the audit log holds the answer. The audit log is built from case one, not retrofitted later.
Our regulatory information comes from verified official sources, not AI training data
AI models trained on the internet contain regulatory errors, outdated rules, misattributed sources, plausible-sounding mistakes. PensionHunter's regulatory knowledge base is populated from official government and regulator sources across the 41 countries we cover, and verified by our team before any content becomes available to the agents. Every regulatory statement in a PensionHunter Pension Identification Report traces back to an authoritative source with a verification timestamp. If a rule changes, the knowledge base is updated and re-verified before customers see the new version.
We tell you how confident we are in every finding
Every finding in a PensionHunter Pension Identification Report is rated HIGH, MEDIUM, or LOW confidence. HIGH confidence findings appear in your report directly. MEDIUM confidence findings appear with transparent flagged language explaining the specific source of uncertainty, variant employer spellings, post-acquisition administrator transitions, scheme type uncertainty. LOW confidence findings are suppressed entirely. We would rather tell you we found nothing than send you chasing a wrong lead. Confidence ratings are not marketing language. They drive the report you receive and the refund decision.
Refunds are decided by the agents, not negotiated with you
If a PensionHunter Pension Identification Report delivers nothing material, neither a regulatory entitlement determination nor an identified scheme administrator, the refund triggers automatically before final report delivery. No customer discussion, no negotiation, no human discretion. Refund eligibility is determined by structured agent output, which means the decision is consistent, defensible, and auditable. Refunds for the Enhanced Pension Identification Report and the Global Pension Identification Report work the same way: if the research identifies nothing material, no administrator and no entitlement position you did not already know, the refund triggers automatically before your pack is delivered. We built it this way because a refund decided by an agent on the basis of structured confidence data is fairer than a refund negotiated case by case.
Humans handle the judgment calls
AI handles the volume. Humans handle the judgment calls. Disputes, edge cases, regulatory updates, the conversations that need empathy rather than automation. Every customer interaction has a human path: questions go to a real inbox monitored by a real person, complaints have a documented process, escalations route to people not bots. Our research team triages every pension provider response in our first 50 cases before any agent automation takes over that work. As we scale, the agents take more of the volume, but the human escalation path stays open.
Where PensionHunter stands today
PensionHunter is currently building out its agent pipeline ahead of paid launch. Our FCA Innovation Pathways application is under active assessment by the Financial Conduct Authority. We are not regulated by the FCA, research-only services do not require FCA authorisation, but we have engaged with the regulator voluntarily because the standards that good regulation enforces are the standards we want to hold ourselves to. The operational discipline described on this page is what we run, today, and what we will continue to run as we scale.