AI-assisted approval of incoming invoices
Invoices from e-mail are extracted, checked against rules and reach accounting after a person approves.
Read the scenario
Resources · Reference scenarios
Reference scenarios
How we would approach typical problems in AI, data, commerce, cloud, delivery, operations and security — with the architecture, limits and trade-offs, not promises.
Every scenario is fictional. None is a client project, a certified architecture or a measured result. Companies, systems and numbers in them are examples.
AI workflows and agents
AI automation, agents, RAG and support with a human handoff.
Fictional scenario Fictional scenario Multi-agent service desk
A coordinator routes requests between classification, knowledge and action agents, with approval before any effect.
Read the scenario
Fictional scenario Internal policy assistant (RAG)
Employees ask about policies and procedures and get an answer with its source, respecting access rights.
Read the scenario
Fictional scenario AI support with a human handoff
An assistant answers frequent questions and hands complex or sensitive ones to an agent with full context.
Read the scenario
Documents and data
Document processing, data pipelines and hybrid integration.
Fictional scenario Contract document intelligence
Contracts and supplier documents are classified, extracted and checked before they enter the register.
Read the scenario
Fictional scenario Hybrid integration of an on-premises ERP
The on-premises ERP exchanges data with cloud services through a secure channel and a queue, without opening it to the internet.
Read the scenario
Fictional scenario Data pipeline to an analytics warehouse
Sales, marketing and operations data is collected, checked and modelled in a warehouse with clear owners.
Read the scenario
E-commerce
ERP synchronisation and product catalog enrichment.
Fictional scenario Shop and ERP synchronisation
Orders, stock and prices are synchronised between the online shop and the ERP with events and checks.
Read the scenario
Fictional scenario Product catalogue enrichment
Descriptions and attributes are completed from supplier data and approved before they are published.
Read the scenario
Cloud and platform
Azure and AWS landing zones, Kubernetes and cloud migration.
Fictional scenario Azure landing zone for regulated workloads
A management group hierarchy, hub-spoke network, policies and central logging as the foundation for new applications.
Read the scenario
Fictional scenario Multi-account AWS landing zone
An organisation with separate accounts, SCPs, central logging and a Transit Gateway as the foundation for teams.
Read the scenario
Fictional scenario Internal Kubernetes platform
A managed cluster with team namespaces, GitOps, policies and observability as a paved road for services.
Read the scenario
Fictional scenario Migrating a monolith to the cloud
A phased move of a monolithic application with assessment, pilot, parallel running and a rollback plan.
Read the scenario
Software delivery
CI/CD and GitOps with change control.
Fictional scenario CI/CD and GitOps with change control
Every change goes through a pull request, tests, a signed artefact and GitOps to the environments.
Read the scenario
Operations and reliability
Observability, reliability, FinOps and AI operations.
Fictional scenario Observability with OpenTelemetry
Metrics, logs and traces from services are collected through a collector, enriched and feed SLOs.
Read the scenario
Fictional scenario Platform reliability
SLOs, on-call, runbooks and post-incident reviews turn reliability into a process rather than heroics.
Read the scenario
Fictional scenario FinOps and AI costs
Allocation of cloud and AI spend, budgets, alerts and reviews of models and tokens.
Read the scenario
Fictional scenario Operating AI systems
A version registry, evaluations before changes, controlled releases and incidents for AI assistants in production.
Read the scenario
Identity and security
Single sign-on, least privilege, secrets and a secure AI gateway.
Fictional scenario Identity, least privilege and secrets
Single sign-on, task-based roles, temporary privileges and a central secrets store for people and services.
Read the scenario
Fictional scenario Secure gateway for AI models
Every AI request goes through a shared gateway with masking, policies, quotas and cost metadata.
Read the scenario
Facing something similar?
We start with an assessment of where you are and propose a realistic first phase.