Financial services
Run it as a container in your own tenant: the network controls, key management and logging your reviewers have already approved govern it.
Who runs it
M&A research at one end, high-volume customer answering at the other, and internal knowledge in the middle — from Tokyo to Montevideo. Each runs on the same governed platform; what changes is what it reads and where it answers.
Trading house · M&A
Retail & commercial banking
M&A advisory
Universal banking · State bank
Pension fund administration
Fintech · Digital payments
Story 1 of 6: Itochu
Also running NeuralSeek in financial services
What a security review checks
These are not a finance-specific tier — they ship in every deployment, including the managed one you would evaluate on. They are listed here because in a bank they are the four that get asked first.
13 detector categories and five enforcement actions — mask, flag, hide, delete, or pass through with a warning — configurable per intent.
Every question, answer, score and configuration change logged, timestamped, attributable, and exportable to S3, Splunk or Datadog.
Separate keys per tenant. Your prompts, documents and outputs never train shared models and never co-mingle.
Your data stays yours. NeuralSeek does not train models on your prompts, documents or interaction logs.
The governance surface
Each one is the detail behind a claim above, written for the person who has to sign rather than the person who has to buy.
Behind them sits the packaging question: what is in the container, which marketplace agreements it arrives under, and what your platform team is actually taking on operationally. The systems this connects to, and the audit destinations the log exports to, are set out by family on the connectors page.
Guardrails
Drop the container into an environment your auditors have already signed off and it inherits that boundary: your network controls, your key management, your logging, your incident process. That is why the honest answer to “is it certified?” is mostly a question about your tenant — and why the one thing attested at the vendor level is stated separately from everything that is not.
Attested
SOC 2 Type II
Attested at the vendor level and audited annually. The report is available under NDA.
The frameworks below are not certifications NeuralSeek holds. They are requirements it is built to meet inside your environment, where your own controls and your own audit apply.
Control mappings, not compliance claims. Validate your posture against your own obligations, with your own counsel — the certifications page sets out where the line between the two sits.
The first meeting is about whether it works. These three are the ones that decide whether it ships.
Every answer carries the passages it was grounded in and a confidence score, and the run behind it is replayable: which agent ran, which model was called at which step, what each step returned, and how long it took. Configuration is versioned with a visible history, so a question about behaviour six months ago is answered from the record rather than from memory. All of it exports to the collector you already run rather than living in a vendor console.
Then it does not. NeuralSeek is a container you run in your own account, your own region or your own datacentre, and retrieval, orchestration, logging and the audit trail run there with it. The one boundary crossing left is the model call, and that is a choice: bring your own provider keys, or point it at an open-weight model on hardware you own and nothing leaves at inference time either. Fully air-gapped is a standard topology, not a special engagement.
That is the normal starting point, and the model is not what NeuralSeek replaces. The model is one configurable layer inside the container; what sits around it is the retrieval, the enforcement and the record, which is the part hand-built projects tend to leave until an audit asks for it. Swapping the provider later is a configuration change rather than a migration, and the applications built on top do not move when it happens.
Talk to a NeuralSeek expert about how it fits your stack, where your data has to live, and the governance your auditors already expect.
