On-premise generative AI in the enterprise
What sovereign really means, the five layers of an internal platform, what it costs line by line, the four most common pitfalls, and a decision grid for on-premise, SaaS and hybrid.
Read the guide
Resources
Reference points to decide with, not a sales pitch.
Deploying generative AI in an enterprise raises questions of architecture, law and cost long before it raises questions of tooling. These guides treat those questions on their own terms. They stay useful even if you choose a solution other than ours.
Choosing your architecture
On-premise, SaaS or hybrid, open or proprietary models, AI or rules: the structural choices, and what they really cost.
On-premise, SaaS or hybrid AI: choosing use case by use case
Classify your use cases by the data they handle, recognise when SaaS is enough and when on-premise is a must, and build a hybrid setup that does not rely on users' goodwill.
Read the guideThe real cost of generative AI in the enterprise: building your TCO
The lines of a full cost, the mechanisms that push the inference bill up, how to read a pricing model, and a method to build a defensible TCO before you buy.
Read the guideGenerative AI or deterministic rules: when to use an LLM
The criteria for choosing between an LLM and business rules, the cases where deterministic logic wins, and why the right architecture often combines both.
Read the guideOpen source or proprietary LLMs: choosing models for the enterprise
Open source, open weights or proprietary: what the licences really say, the criteria that separate Mistral, GPT and the others, and a method to evaluate models on your own cases.
Read the guideRAG in the enterprise: what makes a good document assistant
How RAG works, the real reasons document assistants disappoint, the question of access rights, the criteria of a good tool and a method to measure its quality.
Read the guideSovereignty and compliance
Cloud Act, GDPR, AI Act, anonymisation: what the law really requires, and how to check it before you deploy.
Data sovereignty and generative AI: what sovereign AI really guarantees
What sovereign AI guarantees and what it does not, the laws that apply, how to map a project's data flows, and the questions to ask a vendor.
Read the guideThe Cloud Act and generative AI: what is really exposed
The Cloud Act targets the entity operating a service, not the address of the data centre. What is really exposed in a generative AI project, what is not, and the checks to make.
Read the guideAI Act obligations for businesses that deploy AI
Deployer or provider, the timetable after the 2026 omnibus, what already applies to any business using AI, and what changes when a use is high-risk.
Read the guideGDPR and generative AI: the questions to settle before deploying
Roles, lawful basis, impact assessment, processor contract, transfers outside the EU, minimisation: the decisions to make and document before putting generative AI in your teams' hands.
Read the guideAnonymise or pseudonymise data before sending it to an LLM
What the law distinguishes between anonymisation and pseudonymisation, why free text resists anonymisation, what 2025 case law changed, and how to protect a prompt before it is sent.
Read the guideAgentic AI
Agents that act inside your tools: what sets them apart from a chatbot, what needs approval, and how to connect them to your systems.
Enterprise agentic AI platform: what IT should demand
The requirements that separate an enterprise agentic platform from a demo: permissions, human approval, connectors, audit trail, models and costs, with a buying checklist.
Read the guideAI agent, assistant, chatbot: what are the differences
What sets a chatbot, an AI assistant and an AI agent apart: who decides the next step, who acts, and what that changes for risk and for choosing.
Read the guideHuman approval of an AI agent: which actions to approve
Which AI agent actions a person should approve, on which criteria, and how to design human approval that does not turn into a reflex click.
Read the guideConnecting an AI agent to your systems: ERP, CRM, email
Connecting an AI agent to ERP, CRM and email: with which permissions, through which connectors, which operations to expose, and the risks specific to each choice.
Read the guideGovernance and rollout
Framing usage, controlling consumption, moving from proof of concept to production.
Generative AI governance in the enterprise: roles, rules, logs and costs
The six workstreams of a generative AI governance that holds up in operation: roles, usage policy, model catalogue, access rights, traceability and cost control.
Read the guideShadow AI: why banning fails, and what replaces it
Why banning public AI tools moves usage instead of stopping it, what to keep from the ban, and how a governed internal alternative, measured before and after, actually reduces shadow AI.
Read the guideAI quotas and spending caps: controlling consumption day to day
Why generative AI consumption drifts, the difference between a quota, a cap and an alert, at which level to set limits, and how to size them from real usage.
Read the guideGenerative AI from POC to production: deploying what has proved its worth
Why so many generative AI proofs of concept remain demos, how to scope a POC that can reach production, which exit criteria to set, and what changes at rollout.
Read the guide
SOVEREIGN BY ARCHITECTURE
A question these guides