the token

Bespoke, at the pace of intent.

Rebuild your business AI‑native.

Most enterprises are AI‑first: agents and tools layered onto foundations built before any of it existed. We are AI‑native — the foundation itself, redrawn for the company in front of us.

Example target operating model · Retail & Distribution

An example operating model for a retail and distribution business. Nineteen operating functions arranged as a constellation: customer-facing functions sit upper-left, operations spread across the centre, foundations sit to the right and bottom. Hairline edges connect functions that share day-to-day working relationships. AI participates at three levels: plain nodes are human-led; nodes with a hollow violet ring are AI-assisted; nodes with a solid violet pip are agentic. Marketplace Storefront E-com CRM Wholesale Service Merchandising Pricing Brand Planning Marketing Inventory Fulfilment Data Finance IT HR Compliance Governance
Human-led
AI-assisted
Agentic
Signal
Zone · 01 Customer & Channels
Zone · 02 Operations
Zone · 03 Foundations
Hover a function to inspect

Each function maps to one of three levels of AI involvement — see the legend.

§ 01 · Thesis

Building with AI, not just using it.

Most enterprises today are using AI rather than building with it — tools licensed, copilots deployed, vendor APIs called, third-party chatbots adopted department by department. Each one a layer between the company and the model that does the work. The agents are generic. The data leaves the perimeter. The dependency compounds.

AI-native is a different posture. The operating layer is your own. The agents are designed for the work you actually do. The model can sit inside your perimeter if the regulator or your competitive instinct says so. No middleman between the work and the intelligence doing it. HBS, Nov 2025.

What is usually sold as AI transformation is AI-first with a roadmap attached. The substrate work is harder, slower from the outside, and the only kind that compounds.

Agents designed for the work in front of us, integrated with the data that runs through it, deployed inside your perimeter or on the infrastructure of your choosing. The system is yours; the methodology travels with us; the rigour holds where vendor demos do not. We engage one client at a time, deeply.

§ 02 · What we do 02 / 10

Three engagement types. One narrative.

Across all three — Advisory

A sustained engagement, when that’s the right shape.

Some situations call for a strategic conversation rather than a delivery — board-level questions, CTO-shaped questions, AI questions that need to be answered before the work can be scoped. The shape is bespoke. The cadence, the scope, and the outcome are agreed before anything starts. Advisory →

§ 03 · Agent-Native Operations 03 / 10

The business that runs itself, supervised.

An operating layer that turns a function chart into a working fleet of agents. Roles, decision rights, handoffs, audit trails — explicit, visible, editable by the people who own them. The work that used to live across seven meetings lives in one surface.

See the Target Operating Model flagship

Stylised view. Your real surface is built for your operating model.

§ 04 · AI-SDLC 04 / 10
AI-SDLC · PRD-043
Q2-S14 · pricing-rules-v2
REQ-PRC-03-014
The pricing engine SHALL apply competitor-aware adjustments to assortment SKUs
within CHF 50,000 daily impact without human approval. Above threshold:
hand off to the merchandising lead with a one-page rationale.
Requirements Engineer 14:32
The threshold is daily impact in CHF, but the rationale page is per-SKU. If a single batch exceeds CHF 50k across 200 SKUs, do you want one rationale for the batch, or 200 individual rationales for the merchant to triage?
You 14:33
One rationale per batch, grouped by category. Individual SKU detail behind a disclosure.
Requirements Engineer 14:33
Understood. I’ll lock that into the TDD and flag the disclosure pattern as a UI requirement. Proceeding.

Stylised view. Your real pipeline runs on your repo, your reviewers, your gates.

AI-SDLC.
Agentic software development life cycle.

Business intent. Specifications. Working code.

The pipeline begins with a short prompt from the business and ends with code in production. Between those, an agentic loop reads, clarifies, designs, writes, reviews, tests, documents, and maintains — drawing on a comprehensive understanding of the company. The agents do the lifting; the humans set direction and weigh trade-offs.

See the AI-SDLC flagship

§ 05 · Sovereign LLM 05 / 10

Your model. Your data. Your premises.

A complete vertical, end to end. The hardware on which the model runs. The open-weight model selected to fit the work. The agentic orchestration layer above it. The tools and integrations the agents touch. The AI-SDLC that builds on top. The Digital Employee that uses it. The company chatbot that runs on it.

No vendor lock-in. No token bill that grows with usage. No data that leaves the jurisdiction. Independent of OpenAI, Anthropic, anyone else.

For organisations whose regulators, customers, or boards have already decided that this is the only acceptable shape — and for those whose constraints will arrive at the same conclusion eventually.

See Sovereign LLM

Your premises
Hover or focus a layer to read what it means.
§ 06 · Integrations 06 / 10

Wherever the work happens.

Agents that meet your team in Slack and Teams. Workflows that run inside your AWS, GCP, Azure, or sovereign environment. Connectors to the systems of record you already operate — and to the ones you would rather replace. The agent fleet lives where the work lives.

Slack
Teams
AWS
GCP
Azure
Notion
GitHub
Linear
Jira
Salesforce
Snowflake
Custom REST
§ 07 · Capabilities 07 / 10
§ 08 · The Team 08 / 10

The Team.

A senior practice across systems engineering, product management, architecture, development, and AI specialism. The work has included state-of-the-art trading platforms with ultra-low-latency performance, institutional-scale programmes and roadmap design and execution, strong product implementations, complex automation, and full operating-model and AI-native transformations inside heavily regulated environments — Switzerland, the European Union, and the Middle East. Long-horizon engagements measured against precision and durable business growth, not slideware.

Technical excellence is the starting point. We have always built — the AI-native moment changed the tools and accelerated the work. The same people who designed and shipped those systems now design and ship the agent fleets that run alongside them. Builders, not users.

We make AI legible to founders, owners, boards, and executive teams. We make it real for the engineering and operations teams who will use it every day.

Headquartered in Switzerland, with a global delivery network. The core team is small by design; named senior partners across geographies are activated when an engagement calls for additional capacity — sub-contracted under one brand, one accountability, one point of contact.

You can outsource the thinking to AI, but not the understanding.

§ 09 · How the work actually runs 09 / 10

How the work actually runs.

An engagement isn’t a single delivery — it’s a layered sequence of parallel and dependent tracks. Audits run in parallel; design depends on what they surface; rollout sequences against the operating-model decisions made during design; sovereign infrastructure is delivered alongside the transformation, not after it. The diagram below maps a typical engagement shape. Every engagement is its own.

Phase 01Diagnosis — parallel W1 — W5
Technical Due Diligence
W1–W3
Codebase Analysis
W1–W3
Stakeholder Interviews
W1–W2
Landscape Assessment
W2–W4
AI-Readiness Audit
W3–W5
Phase 02Strategy & Operating Model Design W4 — W11
Vision & Strategy
W4–W6
Target Operating Model
W6–W9
Roadmap
W8–W10
Target Platform Architecture
W9–W11
Phase 03Sovereign Infrastructure — alongside W6 — W22
Hardware procurement
W6–W12
On-premise model deployment
W12–W14
Agentic orchestration layer
W14–W17
Tools & integrations
W16–W19
Company chatbot
W18–W20
AI-SDLC on sovereign stack
W19–W22
Phase 04Build & Phased Rollout W10 — W24+
Agentic Redesign — Function 1
W10–W14
Agentic Redesign — Function 2
W12–W16
Agentic Redesign — Function 3
W14–W18
AI-SDLC enablement
W14–W20
Digital Employee deployments
W18–W24+
Phase 01 · Diagnosis W1–W5
  • Technical Due DiligenceW1–W3
  • Codebase AnalysisW1–W3
  • Stakeholder InterviewsW1–W2
  • Landscape AssessmentW2–W4
  • AI-Readiness AuditW3–W5
Phase 02 · Strategy & Operating Model W4–W11
  • Vision & StrategyW4–W6
  • Target Operating ModelW6–W9
  • RoadmapW8–W10
  • Target Platform ArchitectureW9–W11
Phase 03 · Sovereign Infrastructure W6–W22
  • Hardware procurementW6–W12
  • On-premise model deploymentW12–W14
  • Agentic orchestration layerW14–W17
  • Tools & integrationsW16–W19
  • Company chatbotW18–W20
  • AI-SDLC on sovereign stackW19–W22
Phase 04 · Build & Phased Rollout W10–W24+
  • Agentic Redesign — Function 1W10–W14
  • Agentic Redesign — Function 2W12–W16
  • Agentic Redesign — Function 3W14–W18
  • AI-SDLC enablementW14–W20
  • Digital Employee deploymentsW18–W24+
§ 10 · Get in touch 10 / 10
Contact

Tell us what you’re working on.

A short note about what you are trying to do, what is stuck, and what success would look like.