Operational AI agents · S-PRO

Everyone has Copilot.
Almost nobody gets past personal productivity.

Our solution connects your AI provider with your processes and your data.

Microsoft 365 Copilot

Hi, what can I help you with?

@
Agent templates

The maturity curve

Most companies are stuck in phase 1

Depth and control rise with complexity and risk. We build where it is genuinely viable for financial services today, and architect for what comes next.

Self-service & business config Engineered with S-PRO
Most are here 01

Individual AI

Out of the box. Personal productivity, no setup, no access to your systems.

  • Out-of-the-box productivity
  • Single input, single output
  • No connection to your data
e.g. email, translation, summaries
02

Configured AI

The same tools, configured for your teams, with curated knowledge and the first connectors.

  • Curated knowledge sources
  • Low-code, declarative setup
  • First connectors, single domain
  • Low risk
e.g. policy & procedure search, document Q&A
Where we build 03

Engineered AI

Purpose-built agents on a governed semantic layer, evaluated, tested, and confirmed by a human.

  • Full access to your semantic data
  • Evals & test cases on every agent
  • Multi-step, multi-connector workflows
  • Human-in-the-loop sign-off, full audit trail
e.g. KYC drafting, periodic review, credit memo
Not yet fully viable for financial services 04

Fully agentic

Agents that own a process end to end and decide autonomously, no human in the loop.

  • Autonomous, end-to-end decisioning
  • Mission-critical, near-zero margin of error
  • Accountability & explainability unresolved
We architect for it, and keep a human in the loop until governance catches up.
We help you across AI adoption
How it’s built

Our system-agnostic framework works for your bank

Interface layer
Microsoft 365 CopilotTeams & OutlookWeb & mobileYour own apps · API
Agent layer
KYC & OnboardingDeep research, storyline draft, wealth-gap check
Periodic Review (RAF)Virtual analyst, per-question guardrails, sign-off
Adverse MediaHit triage with verdict + quote, RM routing
Call Report & Cross-BorderPre-call briefing, report draft, audit bundle
Credit ApplicationsDoc extraction, DSCR/ICR, policy & memo draft
+ sub-agent: Transaction Monitoring, open-exploration pattern detection across transactions & alerts
Data layer
Continuous Client File One governed foundation: client data ingested, normalised and versioned once, then reused by every agent (read + write)
Ingestion & normalisationEntity resolutionVersioning & lineageAccess control (CID/PII)
System layer
Core Banking System
  • Transaction tables
  • Portfolio positions
  • Monitoring alerts
  • CRM / contact reports
  • Risk categories
Onboarding / KYC System
  • KYC profiles & fields
  • Client storyline
  • Evidence: IDs, SoW / SoF
  • Credit memo & audit trail
  • Profile history (5-yr)
Document Repository
  • Policies & procedures
  • Investigation / SAR register
  • Counterparty & fund folders
  • Archived reports (PDF)
Email & Communications
  • RM-client correspondence
  • Approval & audit emails
  • Meeting & call transcripts
Screening & External Data
  • Sanctions / PEP / adverse media
  • Market data (prices, filings)
  • Public registers & web research
How we implement it for you

From a free workshop to a foundation you build on

It starts with a free workshop and ends with a foundation you keep building on, not a pile of one-off use cases.

01

Free Copilot workshop

Run with your business teams. We sort your use cases into three buckets: working today, easily configured, and needing further maturity.

OutputA prioritised, bucketed use-case shortlist
02

IT & System assessment

We map your system landscape and recommend the best AI-supported architecture for your stack.

OutputTarget architecture + shared data-foundation plan
03

Process analysis & agent design

Deep-dives with each department, involving the people who do the work today, to design agents around your processes.

OutputProcess maps + signed-off agent designs
04

Implementation

We build a foundational architecture, not use-case-by-use-case, so every new agent builds on the same foundation.

OutputAgents in production on a foundation you keep building on

Engagements with Swiss private banks under NDA, personal references available on request.

Talk to us
Link copied to clipboard