In Palladium, rules decide
Eligibility, amounts, limits and escalation are decided by versioned rules in code, written from your policy. The model never decides.
Architecture, request path, guardrails, audit trail, integrations, deployment and controls on one page. Written for CIOs, CISOs, architects and model-risk teams reviewing Palladium before a pilot.
Deterministic AI is a design where explicit rules make the decision and always return the same output for the same input. A language model is used only to understand the request and explain the result.
Eligibility, amounts, limits and escalation are decided by versioned rules in code, written from your policy. The model never decides.
Each step is sealed in a record that can be replayed later, so you can show exactly which rule produced which outcome.
The difference is what the model is allowed to do.
| Question | Rules in code (Palladium) | Generative decisioning |
|---|---|---|
| Who decides | Rules in code, written from your policy | The model, guided by a prompt and retrieved context |
| Same input, same output | Always | Not guaranteed. Outputs can vary run to run. |
| How you prove a decision | Replay the rule, inputs and outcome step by step | Read the model's rationale and its cited sources |
| Changing a policy | Edit and version the rule. The change is explicit. | Change prompts or retrain, then re-test behavior |
| Changing the model | Affects the wording of explanations only | Can change decisions and needs re-validation |
| Model validation scope | Narrow: classify and explain | Broad: the model is the decision-maker |
Screening, the Rulebook, the Concierge and the Ledger, always in that order. The model sits inside the Concierge only.
Talks to the customer in any channel and routes each request to the right specialist agent: identity and KYC, disputes, payments, lending, fraud and risk, or collections. It decides nothing on its own.
Agents DecidesDecides everything, including dollar thresholds, eligibility and escalation, in code. It talks to no one, and the model never sees this layer.
The Rulebook ProvesSeals what the Rulebook decided and what the Concierge said. Records are replayable and exportable, and nothing written to the Ledger ever changes.
Trust & securityRecord #TR-8841: a customer disputes an $84.50 charge in chat.
| Time | What happens | Layer |
|---|---|---|
| 9:41:02 | Request received, identity verified, PII masked. The model classifies it as an unauthorized-transaction dispute. | Screening, Concierge |
| 9:41:04 | Rule reg-e-provisional-credit checks eligibility: within the 10-business-day window, provisional credit eligible. | Rulebook |
| 9:41:09 | The model writes a plain-language response. Output screening checks it matches the decision and leaks no PII. | Concierge |
| 9:41:12 | Provisional credit of $84.50 issued through the payment rail. | Specialist agent |
| 9:41:40 | The full trace is sealed as one record, replayable on request. | Ledger |
Each policy is a versioned rule with explicit conditions, actions, limits and an escalation path. Natural-language policy enforcement is layered on top, never instead.
# policy/fee-waiver-eligibility.yml condition: first_nsf_in_window(days: 365) action: waive_fee(max_amount: 35.00) requires_human: false on_breach: escalate(context: full)
Enforced in code. The model never sees this decision.
Palladium screens the model's words, and it also limits what the model is allowed to do in the first place.
The Ledger is Palladium's AI audit trail. Records are queryable by user, date and event type, and export as a digitally signed PDF with the full trace.
Connectors are pre-built and bi-directional, with granular tool-calling permissions on every action.
Three ways to run Palladium. The rules, agents and Ledger are the same in each.
| Option | Managed concierge | Co-managed rollout | Private deployment |
|---|---|---|---|
| Where it runs | Hosted by Sphere, with strict data isolation | Hosted by Sphere | Your VPC on AWS, Azure or Google Cloud |
| Who writes the rules | Sphere, from your policies | Your team | Your team |
| Who operates it | Sphere | Sphere | Your operating team |
| Models | Model-agnostic | Model-agnostic | Private models, or bring your own |
Palladium is model-agnostic and not locked to one cloud. Your customer data is never used to train models, by default and by contract. See pricing and engagement options.
Conventional controls plus a security layer for the risks that only AI brings.
Inbound messages are screened before the model. Outbound responses are screened before the customer sees them.
Runs on both passes, so personal data does not reach the model.
Role-based access, granular tool-calling permissions and bank-grade encryption.
Data residency and consent management set for where you operate.
Escalation carries the full context: identity state, request progress and everything already verified.
Independently audited controls over security, availability and confidentiality. Palladium is also HIPAA compliant.
Your model-risk and compliance teams can see which rule decided, what the model said, and what changed and when.
The controls Palladium enforces, mapped to the frameworks your reviewers work from.
| Domain | Representative frameworks | What Palladium maps to it |
|---|---|---|
| US model risk | SR 26-2 (April 17, 2026) · OCC Bulletin 2026-13 · FDIC FIL-15-2026 | Versioned, testable rules; a narrow, documented model role; vendor documentation for third-party model review. |
| US consumer banking | Reg E · ECOA/Reg B · UDAAP · interagency TPRM guidance | Requirements held in the Rulebook as versioned rules, with every decision sealed in the Ledger. |
| Organizational security | SOC 2 Type II | Independently audited controls over security, availability and confidentiality. |
| Data protection and privacy | GLBA · GDPR · CCPA/CPRA | Consent management, PII redaction before model access, data residency controls. |
| Banking and tech risk | DORA · FFIEC · PRA/FCA · MAS TRM · OSFI · APRA | Model risk management, operational resilience, third-party risk documentation. |
| AI-specific regulation | EU AI Act | Risk classification, explainability, human-oversight requirements for high-risk use. |
| Payments and data security | PCI DSS | Segmentation and tokenization of payment data anywhere it touches the model layer. |
| Securities and markets | FINRA · MiFID II | Recordkeeping, communications surveillance, best-execution-adjacent audit trail. |
This describes the controls Palladium enforces. Regulatory mapping is not a certification by, or endorsement from, any regulator. SR 26-2 and AI
The decision is deterministic. The conversation is not. The Rulebook decides eligibility, amounts and escalation in code, so the same facts always produce the same outcome. The model only classifies the request and explains the result, and output screening checks the explanation matches the decision.
Palladium is model-agnostic. You can use private model options, or bring your own model in a private deployment. Because the model never decides, changing models changes the wording of explanations, not the decisions.
Every prompt, tool call, decision and reasoning step, timestamped and sealed in the Ledger. Records are queryable by user, date and event type, and export as a digitally signed PDF.
In a private deployment, Palladium runs in your VPC on AWS, Azure or Google Cloud with private model options. In every deployment, PII is redacted before the model and your data is never used to train models.
Circuit breakers sit on every external system, and the platform targets 99.9%+ uptime. When a decision cannot be completed, the request goes to a person with its full context instead of a guess.
Decisions come from versioned, testable rules, and the model's role is limited to classifying and explaining. That keeps the scope of model validation narrow and documented, and the due-diligence kit gives your model-risk team vendor documentation to review.
Bring your security questionnaire. A senior architect walks your team through the data flow, the controls and a live sealed record.