Platform

One trust architecture, every recommendation.

The same five-step path runs underneath everything Palladium ships — a revenue lead for a relationship manager, a care-gap nudge for a patient, a resolved dispute for a customer. Here's exactly how it works.

How Concierge handles a single request

Live walkthrough

1
Customer asks a question
on the portal, app, or chat
2
Deterministic core
rules and scoring decide first
3
Plain-language response
the model explains, never decides
4
Sealed, auditable record
replayable on request

How it works

Five steps, every time.

Data stays put

Sensitive fields are masked or de-identified before anything reaches a model.

Rules decide

A deterministic layer scores eligibility, risk, or opportunity first.

AI explains

The model narrates that decision in plain language — it doesn't reason its way to one.

A human acts

A banker, RM, or care coordinator reviews and takes the next step.

Everything's sealed

A replayable record is available for any examiner or auditor on request.

In practice

Automations, not just conversation.

Every one of these runs the same way: a deterministic system decides, Concierge explains or acts within its bounds, and nothing executes without a rule or a human behind it.

Banking · billing question

"Why was I charged a $35 fee on the 14th?"

Core system pulls the transaction record
Concierge explains the fee and the policy behind it
Offers a dispute path if the customer wants one

Resolved in under a minute, no hold time

Banking · high-value transfer

"I need to send $12,000 to my supplier in Frankfurt."

Concierge collects the transfer details
Core system validates limits, entitlements, and cross-border rules
Customer confirms before anything executes

Executed by the core system — never by the model

Banking · revenue discovery

Overnight: 14 accounts flagged in the "contractor" cohort

Scored and ranked automatically against peer cohorts
Routed to the right relationship manager's queue
An explainable pitch attached to each account

The RM starts the day with a worked list, not a raw report

Healthcare · result explanation

"What does my cholesterol result mean?"

Concierge retrieves de-identified result context
Explains it in plain language
Flags anything outside a normal range for clinical review

Nothing abnormal is ever explained unsupervised

Healthcare · scheduling

"Can I reschedule my Thursday appointment?"

Concierge checks live availability
Offers valid slots that fit the patient's constraints
Confirms and updates the record

Rebooked in the same conversation, no phone hold

Healthcare · care-gap discovery

Overnight: 62 patients flagged with no follow-up scheduled

Ranked by clinical urgency
Routed to the care coordinator's queue
A templated, clinically-reviewed outreach draft attached

The coordinator's queue is pre-built every morning

Delivery tiers

Three ways to work with us.

Some teams just need the intelligence. Others want the whole experience handled. Every tier runs on the same deterministic-core architecture — nothing about the trust story changes based on how much you buy.

Bring your own app

Embed

Concierge and Discovery, integrated directly into the digital experience you've already built.

Concierge conversational engine, via API and SDK
Revenue or Care-Gap Discovery feed
Deterministic core and audit trail
Your engineering team owns the integration

Best for — institutions with a mature app and an engineering team ready to integrate.

Talk to us about Embed

Most common starting point

We build the surface

Concierge App

A white-labeled portal and mobile app, with Concierge and Discovery built in from day one.

Everything in Embed
White-labeled portal and mobile app
Design system and onboarding, live in weeks
Your team maintains it post-launch

Best for — institutions without a dedicated app team, or who want to launch fast.

Talk to us about Concierge App

The whole thing, managed

Full Experience

The complete customer or patient experience — app, Concierge, and Discovery — run end-to-end under your brand.

Everything in Concierge App
Ongoing managed service
Continuous model and cohort tuning
Dedicated support

Best for — institutions who want this off their plate entirely.

Talk to us about Full Experience

Security & trust

Automation you can audit. Not AI you have to trust.

Every recommendation Palladium surfaces traces back to a rule, not a guess. The model explains a decision — it never makes one.

Guardrails enforced in code

Not policy documents. The boundaries a reviewer asks about are architectural.

Deterministic by design

The decision happens first. The model's only job is to narrate it.

Built to show your work

A decision an examiner, auditor, or compliance officer can retrace.

Familiar frameworks

Aligned to the regimes your regulator already recognizes.

No autonomous action

A human, or a deterministic rule, is always upstream of the outcome.

Explainable on request

Every output is replayable back to the rule that produced it.

Domain	Representative frameworks	What Palladium maps to it
Data protection & privacy	GDPR · CCPA/CPRA	Consent management, PII redaction before model access, data residency controls.
Banking & 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 & data security	PCI DSS	Segmentation and tokenisation of payment data anywhere it touches the model layer.
Securities & markets	FINRA · MiFID II	Recordkeeping, communications surveillance, best-execution-adjacent audit trail.

See the architecture on your own experience

Walk through a real conversation flow on your own portal or app design — no live customer or patient data required.

Book a Concierge walkthrough →