Question 1

How does Dafin ensure my data is never used for training?

Accepted Answer

Dafin separates data access from model training. Your data is used only for retrieval and reasoning at query time — never to train or fine-tune any model. No client data is sent to external LLM providers, and contractual guarantees back this. Every data flow is logged and auditable for MiFID II, FINMA, GDPR and SOC 2.

Question 2

What makes Dafin different from other AI tools?

Accepted Answer

Three things. Answers are calculated from your verified data, not generated by a language model — eliminating hallucinations. Every answer comes with cited sources and full reasoning, making it audit-ready by design. Your data never leaves your perimeter — no exposure risk, no data leakage, no training on client information. Built for regulated environments from day one.

Question 3

How does Dafin avoid hallucinations?

Accepted Answer

Hallucinations happen when a language model generates content not grounded in real data. Dafin's architecture prevents this by separating reasoning from generation: answers are calculated from your verified data sources, with the LLM used only to interpret the question and format the response. If data is unavailable, Dafin says so rather than inventing an answer.

Question 4

Who in my firm benefits from Dafin?

Accepted Answer

Dafin is used across the front, middle and back office. Relationship managers for portfolio summaries and meeting prep. Portfolio managers for performance attribution and rebalancing. Traders for exposure and limit monitoring. CIOs for house view alignment. Risk and compliance for MiFID suitability, KYC renewals and concentration breaches. Operations for reconciliation and regulatory reporting.

Question 5

What data sources can Dafin connect to?

Accepted Answer

Dafin's open connector model integrates with portfolio management systems (PMS), market data feeds, KYC and AML platforms, internal documents, cloud storage (S3, Azure Blob), analytics APIs, CRM systems, compliance documents, custodian feeds, risk models and order management systems. New connectors can be added in days.

Question 6

How long does implementation take?

Accepted Answer

Most Dafin deployments go live within 3 to 4 months — fast for a regulated wealth platform where comparable rollouts take 12 to 24 months. Because Dafin connects to existing data sources via an open connector model, there is no lengthy data migration or model training phase. Engagements run through discovery, connector setup, security review and user onboarding.

Question 7

Can Dafin be deployed on-premises?

Accepted Answer

Yes. Dafin supports on-premise, private cloud and hybrid deployment. The platform is designed for regulated environments where data residency, network isolation and full infrastructure control are non-negotiable — including Swiss private banks under FINMA, EU institutions under MiFID II and GDPR, and global firms with strict internal data governance.

Question 8

How is pricing structured?

Accepted Answer

Dafin uses an enterprise pricing model tailored to firm size, users, data volume and deployment configuration. There are no per-query fees that penalise adoption — pricing is predictable and aligned with value delivered. A typical engagement starts with a discovery call and personalised demo. Contact contact@dafin.ai to scope your requirements.

AI that fits inside a regulated bank.

Your bankers are already using AI. You just can’t see it.

Pilots die on data risk

Shadow AI fills the gap

The exposure is invisible

More time with clients.

House view, across the book.

Audit-ready, by design.

Plugs into bank infrastructure.

Bring Dafin into your bank.