What a day of disruption tells us about resilience, ownership and building AI that supports people.
On 12–13 November 2026, a Microsoft 365 outage disrupted Exchange Online and affected features across Teams, SharePoint and Copilot. Lasting almost 24 hours, the incident raised an important question for businesses increasingly reliant on AI: how much control do you have over the systems your operations depend on?
No cloud provider is immune to outages. But as AI moves from everyday productivity into business-critical workflows, understanding the technology and dependencies behind it becomes essential.
When convenience becomes dependency
Off-the-shelf AI assistants are useful for everyday tasks such as drafting documents, summarising meetings and retrieving information. But relying on one platform for essential business workflows introduces a different level of risk.
Consider an organisation using AI to prepare technical reports, process quotations or review compliance documentation. If its AI interface, document connections, workflow logic and model access all depend on one platform, disruption can affect several parts of the process simultaneously.
Even when alternative AI models are available, switching may be difficult if the surrounding workflows are tied to the original provider. This creates vendor lock-in and limits an organisation's ability to respond.
The risk isn't simply downtime. It's having limited control over the workflows your business depends on.
The AI model is only one part of the system
An AI model is only one component of a working business solution. Much of the operational value comes from the systems around it, often referred to as an AI harness.
An AI harness connects a model to a business's data, tools and workflows, providing the controls and oversight needed for reliable operation.
Key components can include:
- Knowledge and retrieval: Connecting AI to relevant internal documents, records and business information.
- Workflow orchestration: Coordinating tasks and integrating with existing systems and software.
- Guardrails and human oversight: Establishing validation, controls and specialist approval for critical decisions.
- Evaluation and auditability: Testing outputs and maintaining visibility over how workflows operate.
Together, these components turn a general-purpose model into a solution designed for operational use. A flexible harness can also make it easier to adapt workflows as technology and business requirements change, without rebuilding everything around a single model provider.
AI should enhance expertise, not replace it
At Brainpool, AI is designed to enhance human expertise, not replace the people responsible for important decisions.
In specialist sectors such as testing, inspection and certification (TIC), AI can reduce repetitive work and improve access to information while qualified professionals retain oversight and approval.
That's why validation, traceability and human review belong within the AI workflow. They help specialists understand, challenge and approve outputs, keeping accountability where it belongs.
Human oversight is most effective when supported by appropriate technical controls, clear responsibilities and access to the evidence behind AI-generated results.

Your AI. Your control.
Business-critical AI needs to do more than work today. It needs to remain adaptable as technology, requirements and operations change.
Can your organisation choose where its AI is deployed? Can workflows be adapted without starting again? And who owns the solution developed around your business's processes?
At Brainpool, these principles underpin Brainpool Cortex, our AI agent platform. Through our Build-to-Own approach, organisations can benefit from:
- Deployment choice and control: The option to run Cortex in their own cloud environment or use Brainpool-hosted deployment, providing greater control over infrastructure and data.
- Flexibility: The ability to work with different AI model providers and integrate agents into existing systems and workflows. Switching providers still requires appropriate planning and testing, but a flexible architecture can reduce the need to rebuild the surrounding solution.
- Ownership: Retaining ownership of bespoke agents, workflows and configurations developed for the business, alongside a licence to the delivered Cortex platform.
The aim is to give businesses greater control over how their AI operates, evolves and supports their people, rather than tying essential workflows unnecessarily to a single provider.
Resilience still needs planning
Greater control doesn't mean immunity from outages. Cloud infrastructure, model providers, identity systems and integrations can still fail.
A flexible architecture gives organisations more options to plan for disruption, from alternative providers to documented manual fallback processes.
The advantage is greater choice and control, not guaranteed uptime.
Building AI for the long term
The Microsoft 365 outage is a reminder that as AI becomes more embedded in business operations, organisations need to look beyond what individual tools can do.
The questions that matter are who controls the technology, how easily it can adapt, and whether it supports the people responsible for the work.
Because the real value of AI isn't simply what it can do today. It's the freedom to adapt it, the confidence to rely on it and the control to make it your own.
Discover how an AI harness can help your business build AI around its people, processes and priorities.

