Microsoft AI Foundry: a practical guide for UK SMBs
What Microsoft AI Foundry actually does, when an SMB needs it, and how to get started without overspending. Written for finance directors, managing directors and operations directors, not developers.
If you have sat through a Microsoft AI pitch in the last six months, you have probably heard the phrase Microsoft AI Foundry. Sometimes you will have heard the older name, Azure AI Foundry. The two refer to the same platform. Microsoft renamed it in 2025 and the new name has not fully propagated through every Microsoft slide deck or partner conversation yet. This guide is the plain-English read on what Microsoft AI Foundry is, when a UK SMB actually needs it, what it costs, and how to start without overspending.
Lanmark is a Microsoft Direct Cloud Solution Provider, which means we hold the licensing and support relationship directly with Microsoft rather than through an intermediary, and we build on Azure AI stacks and Microsoft AI Foundry on that basis. This article is written from that perspective, for the businesses we usually work with: UK SMBs in the 20 to 500 user range.
What Microsoft AI Foundry is, and what it is not
Microsoft AI Foundry (formerly Azure AI Foundry) is the unified platform Microsoft uses to build, deploy and govern AI workloads on Azure. In plain English, it is the place where the AI is actually running. Microsoft 365 Copilot is a user-facing product that sits in front of people; Microsoft AI Foundry is the engine room that lets an organisation build its own AI products on the same infrastructure.
It contains four things that matter to most SMBs. First, a catalogue of foundation models from several providers, each available behind a Microsoft commercial agreement so the data does not leave the Microsoft trust boundary. Second, a workspace for grounding those models on your own data, which is the technique behind any chatbot that appears to know your documents. The technical name is retrieval-augmented generation, or RAG; in plain English, the model looks up the answer in your files before it writes a reply. Third, a workspace for building AI agents that plan and carry out multi-step tasks. Fourth, a governance layer giving you content safety, prompt security, observability and audit logging across all of the above.
What Microsoft AI Foundry is not: it is not a productivity tool you hand to staff (that is Microsoft 365 Copilot), it is not a website chatbot you can buy off the shelf, and it is not an alternative to Microsoft 365. It is the infrastructure underneath the AI products you build for your own business.
Microsoft AI Foundry compared with Microsoft 365 Copilot
The single most useful distinction is this. Microsoft 365 Copilot is consumed by your users. Microsoft AI Foundry is consumed by your business systems.
Copilot helps your team draft, summarise and answer questions inside Word, Outlook, Excel and Teams. It is licensed per user at a published UK list price and is bought as a productivity tool. Microsoft AI Foundry is licensed by consumption, meaning a charge per million tokens of model usage plus storage and other Azure consumption. It is bought as a platform, by a project sponsor, to build something specific.
Most SMBs we meet need Copilot first and Foundry second. Some need only Copilot for a long time. A smaller group, where the value of AI sits in answering questions over their own data or in automating multi-step tasks, will need Foundry within twelve months. The rest of this guide explains how to tell which group you are in.
When a UK SMB actually needs Microsoft AI Foundry
A useful test: if the AI you want involves your own documents, your own records or a multi-step business process, you are likely on the Foundry side of the line. If the AI you want is simply to make your staff faster in the apps they already use, you are on the Copilot side.
The four most common SMB use cases that justify Microsoft AI Foundry today are a grounded internal chatbot that answers questions over your own policies, contracts and SharePoint files (a knowledge agent); a customer-facing assistant that reads product documentation, knowledge base articles and structured data and responds accurately (a customer agent); a back-office assistant that reads incoming emails or documents, classifies them, drafts a reply and queues an action for human approval (a triage agent); and a specialist assistant for a regulated process where a generic Copilot is not controlled enough, for example a compliance check or a financial review (a process agent).
If none of those resembles what you are trying to do, you probably do not need Foundry yet. Copilot, possibly Copilot Studio, and good Microsoft 365 hygiene will take you a long way before the platform underneath becomes the bottleneck.
What you can build with Microsoft AI Foundry
Those four use cases translate into four buildable assets. Each starts with a chosen foundation model from the catalogue, with a current general-purpose model used where reasoning quality matters and a smaller, cheaper model used for high-volume, cost-sensitive tasks. The model is then grounded against your own data using a vector index, prompt-engineered for the task, wrapped in a content-safety layer that filters harmful inputs and outputs, and exposed to your own application or to a Copilot Studio agent.
Foundry also supports light fine-tuning, where the model is taught the tone or the domain language of your business, but most SMBs do not need this in year one. Grounding the model on good data, plus careful prompting, outperforms naive fine-tuning on most operational tasks.
The output of a Foundry build is usually invisible to users. They see a chatbot, a recommendation in their CRM, an enriched record in their ERP, or a draft reply in Outlook. Foundry is the platform underneath.
Microsoft AI Foundry pricing, in plain English
Foundry is not a per-user licence. It is consumption-based, and the consumption sits inside your Azure subscription. There are three meaningful cost lines.
The first is model usage, charged per million input and output tokens and varying by model. The second is vector index and storage cost, in Azure AI Search or Azure Cosmos DB, which for an SMB-scale knowledge base is a modest monthly line rather than a major one. The third is content safety, monitoring and orchestration, which is largely included but can add a small premium at higher tiers.
As an indicative planning range rather than a quotation, an SMB project of the scale described above would typically sit in the hundreds to low thousands of pounds per month in Azure consumption once live, plus the partner build cost, which we usually structure as a fixed-scope four to eight week pilot. The figure moves with data volume, query rate and the model you choose, and Microsoft revises its own pricing regularly. We will give you a costed estimate once we know the use case, which is a more useful number than any range published on a website.
Governance, security and the UK regulatory context
For most UK SMBs the governance position is the reason Microsoft AI Foundry gets approved over a generic build on a consumer AI service. The platform sits inside your Microsoft tenant, runs on Azure, respects Entra ID identity, integrates with Microsoft Purview classification and data loss prevention, and offers region-locked data residency. UK South and UK West regions are available for both compute and storage, which matters for FCA-regulated firms, NHS suppliers and any organisation whose clients ask due diligence questions.
The platform also provides content safety filters, prompt injection protections and the audit trail you need to demonstrate that AI is being used responsibly. For background, see Microsoft’s own Microsoft AI Foundry product page and the UK government’s approach to AI regulation.
That said, the platform is only as good as the way it is set up. The single most common AI governance failure we see is not technical: it is allowing a Foundry build to read SharePoint content that the user was never supposed to see. An AI agent inherits whatever permissions you give it, and most SharePoint estates have accumulated sharing decisions nobody has reviewed in years. We look at the SharePoint sharing model, Purview classification and Entra ID Conditional Access policies before any AI workload goes live.
How Microsoft AI Foundry fits the AI Capability Ladder
Lanmark uses the AI Capability Ladder framework with clients to keep the AI conversation in business language rather than technical language. Microsoft AI Foundry sits at rungs two to four: knowledge AI (Foundry plus a grounded knowledge base), agentic AI (Foundry plus orchestration, with people approving the steps that matter), and action-taking AI (Foundry integrated with your line-of-business systems, with audit and rollback designed in from the start).
Most SMBs are working at rung one, which is productivity AI and mostly means Copilot, and looking up at rung two. The right first Foundry project for a business at rung one is almost always a single grounded knowledge agent on a single high-value data set, kept deliberately narrow so that the build, the cost and the governance all stay controllable.
You can read more about how we sequence the move up the ladder in our wider AI consultancy London work.
How Lanmark approaches Microsoft AI Foundry for SMBs
Our engagement model is deliberately small at the start. The first step is an AI Readiness Review, a focused conversation that places your business on the AI Capability Ladder and establishes whether Foundry is the right next step or whether Copilot adoption alone is. There is no obligation and no sales pitch. If Foundry is right, the second step is a four to eight week fixed-scope pilot on a single use case, with a written success measure agreed before we start. If the pilot lands, the third step is a rolling adoption programme, adding use cases at the rate the business can absorb them.
Lanmark is a Microsoft Direct Cloud Solution Provider and holds Microsoft’s Support Service Designation, an accreditation held by only a handful of companies worldwide. Our technical team is completing a structured programme of training delivered directly by senior Microsoft AI architects, covering Microsoft AI Foundry, Azure OpenAI on enterprise data, and the design, deployment and governance of AI agents.
Our work is run from our offices in Holborn and Dartford, serving SMBs across central London and the South East.
Frequently asked questions
What is Microsoft AI Foundry in plain English? It is the place where Microsoft’s enterprise-grade AI runs on Azure. Foundation models, your own data, AI agents, and the security and governance wrapped around them all live in one platform. It is the engine room behind any AI product a business builds for itself.
How is Microsoft AI Foundry different from Microsoft 365 Copilot? Copilot is bought per user as a productivity tool inside Word, Outlook, Excel and Teams. Microsoft AI Foundry is bought by consumption as a platform, by a project sponsor, to build a specific AI product. Most SMBs need Copilot first and Foundry second.
Why is it sometimes still called Azure AI Foundry? Microsoft rebranded the platform in 2025. The older name Azure AI Foundry has not yet been replaced everywhere, particularly in older documentation, partner slide decks and search results. The product is the same.
When does an SMB actually need Microsoft AI Foundry? When the AI you want involves your own documents, your own records, or a multi-step business process. If you only want to make staff faster inside the apps they already use, Copilot is enough.
What does Microsoft AI Foundry cost? It is consumption-based rather than per user, so the bill sits inside your Azure subscription and moves with usage. As an indicative planning range rather than a quotation, an SMB project of the scale described on this page would typically sit in the hundreds to low thousands of pounds per month in Azure consumption once live, plus the partner build cost. The figure depends on data volume, query rate and the model you choose, and we will give you a costed estimate once we know the use case.
Is my data safe in Microsoft AI Foundry? Foundry runs inside your Microsoft tenant, on Azure, with UK South and UK West data residency available. It respects Entra ID identity, integrates with Microsoft Purview, and gives you content safety and audit logging. The most common governance failure is letting an AI agent read content the user should not see, which is a setup discipline rather than a platform limitation.
Can we build with Microsoft AI Foundry without an in-house development team? Yes. Most of the SMB Foundry work we do is delivered as a fixed-scope, partner-led pilot. Microsoft Copilot Studio also lets you build lighter-weight agents on top of Foundry without writing code, which is enough for many first projects.
How quickly can a UK SMB get a first Microsoft AI Foundry project live? Four to eight weeks is realistic for a single grounded knowledge agent, assuming the data is reasonably clean and the use case is narrow. We start with an AI Readiness Review, which tells you whether that timeline is realistic for your business.
Talk to Lanmark about Microsoft AI Foundry
If you are weighing up whether Microsoft AI Foundry is the right next step for your business, the cheapest first move is an AI Readiness Review. There is no obligation and no sales pitch, and the only follow-up is the one you ask for. Get in touch using the form, or call 020 7123 4910 and ask for the AI consultancy team.