By Dave Perske, RES. Business IT
AI is clearly going to be part of how work gets done. Teams are already experimenting, but many businesses are still trying to work out what’s safe, what’s useful, and what’s just noise.
That uncertainty often leads to two responses: some businesses freeze, while others look the other way and hope it doesn’t become their problem.
AI will get used whether a business “allows it” or not. What changes the outcome is whether it’s adopted deliberately, with a sensible approach the team can follow, or whether it spreads quietly through unsanctioned tools and habits as shadow IT.
The opportunity is there, and you don’t need a huge transformation program to get started.
What’s holding businesses back
In recent conversations, a few common themes have started to emerge.
Businesses want to move but don’t know what the first step should be, so it stays stuck in discussion. Leaders worry about risk, uncontrolled use is already a concern, and policy doesn’t change behaviour if people aren’t given a safe option. Some teams are moving early to get an edge, while others avoid it because they assume they can’t use AI at all, even though there’s usually still value in the right scenarios.
This matters because AI affects productivity, process, and quality. With more customers and partners asking direct questions about data, privacy, and what tools are in play, a clear starting point now makes those conversations easier later.
The misconception we still hear is “My people know they aren’t allowed to use AI, so they don’t.” In reality, people are often just trying to get work done, and leaders may not have visibility of which tools are being used or what information is being entered.
A simple way to start: Write, meet, process, decide, share and scale
1. Write better: Everyday writing help with Copilot
Early exploration often starts here because it’s immediate and low friction. Copilot can turn rough notes into a clearer email, sharpen a customer letter, summarise a long document, or improve tone.
Try prompts like:
- “Rewrite this email to be clearer and more confident, but still friendly.”
- “Summarise this document into 5 key points and 3 recommended actions.”
- “Tighten this proposal section and make the value clearer.”
Keep in mind: Treat outputs as drafts and avoid pasting sensitive client information into non-sanctioned tools.
2. Meet better: Easy recaps and action tracking in Teams
For organisations working in Teams, meeting recap and action tracking is often the quickest way to see real value. People spend less time taking notes, and follow-ups get sharper.
Try prompts like:
- “Summarise this meeting for someone who wasn’t there.”
- “List decisions made, risks raised, and actions with owners.”
- “Draft the follow-up email with next steps.”
Keep in mind: Be deliberate about which meetings are recorded or transcribed and set expectations with staff and customers.
3. Process better: Turning repetitive work into a streamlined workflow
This is where AI starts moving from personal productivity into process improvement. We often see jobs like converting supplier quotes into a client-facing format, where traditional automation can struggle because the inputs are not always consistent.
With curated prompts and clear business rules, teams can extract key fields, apply rules, and produce a consistent output that still gets reviewed before it’s sent.
Try prompts like:
- “Extract line items, quantities, pricing, and assumptions into a clean table.”
- “Map this supplier format to our standard quote template.”
- “Highlight anomalies: missing SKUs, unusual pricing, inconsistent units.”
Keep in mind: Maintain explicit business rules, don’t let AI guess, and validate outputs before sending.
4. Decide better: Research and analysis using business information
Once teams get comfortable, AI can help make sense of information spread across files, emails, meeting notes, and spreadsheets. Done well, it’s like giving your team a first-pass analyst that summarises themes and highlights where to look deeper.
Try prompts like:
- “Summarise themes and open issues from the last 30 days of client meeting notes.”
- “Review this Excel file and highlight clients with margin anomalies or unusual discounts.”
- “Based on these incident reports, list recurring root causes and suggested fixes.”
Keep in mind: This only works safely if SharePoint and Teams permissions are in good shape, because AI will amplify messy access.
Getting the most from AI in Microsoft 365 also depends on having a solid data foundation. Our Data & AI Consulting helps organisations structure their data environment so Copilot and other tools can surface reliable, governed insights.
5. Share and scale better: Turn individual wins into repeatable capability
This is the step most businesses miss. Adoption usually moves from individual tips to shared patterns and reusable assets, then into deeper integration.
If a prompt helps one person turn meeting notes into a useful customer summary, turn it into a shared template. If a process keeps repeating, document the rules and make it easier for the team to follow the same approach.
Keep in mind: Don’t roll this out wider until it’s well understood and you’re confident with the results. Standardise what repeats and only integrate deeper once data access and controls are in good shape.
A simple starting point we recommend
Before you invest in more tools or write a long policy, these three moves tend to do the most work when you’re first starting off:
- Understand what’s already happening: Where is AI being used today, and what problems are people trying to solve? A curious approach works best.
- Choose a sanctioned starting point: Give staff a safe option to start working with AI (eg, Copilot Chat), with clear guidance on what should never go into prompts.
- Run a short pilot: Pick one or two scenarios above, like meeting recap and follow-ups in Teams or drafting and summarising documents. Then observe the impact and feedback. Time saved and consistency are usually the easiest to track.
Where RES can help you get started
Most organisations don’t need an “AI program” to begin. What they need is a sensible starting point, a small number of use cases that map to real work, and the basics in place so the rollout doesn’t create new risk.
This is where RES can help. We support organisations to make practical use of Microsoft Copilot, focus on repeatable tasks that genuinely save time, and strengthen the foundations around SharePoint, Teams, permissions and data handling.
These foundations sit within our broader modern workplace solutions capability, helping businesses configure Microsoft 365, Intune, Teams and SharePoint so that AI tools operate securely and consistently from day one.
The aim is to get value early, learn quickly, and build confidence without turning AI into a compliance headache.
If you want help choosing a sensible first step, or you’d like someone to sense-check what’s safe in your environment, we’re happy to talk it through.
Frequently Asked Questions
How should a business start using AI at work?
A business should start with one or two practical tasks that employees already complete regularly. Suitable examples may include drafting documents, summarising information, preparing meeting follow-ups or organising repetitive inputs. A small starting point makes it easier to observe results, identify risks and improve guidance before expanding AI use across more teams or processes.
What business tasks are suitable for an initial AI pilot?
An initial AI pilot should focus on repeatable, low-complexity tasks with outputs that employees can review. Examples include improving written communication, summarising documents, extracting information into a consistent format or identifying themes across business records. Tasks involving sensitive decisions or unreviewed customer-facing outputs generally require stronger controls and more careful assessment.
What is shadow AI and why is it a business risk?
Shadow AI is the use of artificial intelligence tools without the organisation’s approval, visibility or oversight. Employees may adopt these tools to complete work faster, but the business may not know what information is being entered or how outputs are being used. Providing an approved option and practical guidance can reduce uncontrolled use, although it cannot remove every risk.
What information should employees avoid entering into AI tools?
Employees should avoid entering sensitive, confidential or restricted information into tools that have not been approved for that purpose. This may include customer data, internal financial information, credentials, commercially sensitive documents or personal information. The organisation should define clear data-handling rules so employees understand which tools are approved and what information can be used safely.
Why should AI-generated content always be reviewed?
AI-generated content should always be reviewed because it may be incomplete, inaccurate or inconsistent with the organisation’s requirements. Artificial intelligence can produce a useful first draft, but it does not understand business context in the same way as an accountable employee. Reviewers should check facts, calculations, tone, assumptions and any proposed actions before using or sharing the output.
How can AI improve meetings and follow-up work?
AI can support meetings by preparing summaries, identifying decisions and organising actions for review. This may reduce manual note-taking and help participants follow up more consistently. Organisations should still decide when recording or transcription is appropriate, explain the process to participants and confirm that generated summaries accurately reflect the discussion before relying on them.
Why do data access and permissions matter when using workplace AI?
Data access and permissions matter because an AI tool may surface information that a user is already authorised to access across connected business systems. Poorly managed permissions can therefore make existing access problems more visible or more influential. Businesses should review document ownership, shared locations and user permissions before applying AI broadly across internal information.
How should a business measure the value of an AI pilot?
A business should measure an AI pilot against a small number of practical outcomes linked to the selected task. Useful measures may include time required, consistency of output, rework, employee feedback and whether the process became easier to follow. The purpose is to understand where AI provides useful support, not to assume that every task should be automated.
When should a business expand AI use beyond a pilot?
A business should expand AI use when the initial task is understood, the outputs are reliable enough for their intended purpose and appropriate controls are in place. Repeating successful prompts, documenting business rules and creating shared templates can help turn individual results into a consistent team capability. Deeper integration should follow only after data access, ownership and review processes have been addressed.
How can RES. Business IT help a business adopt AI responsibly?
RES. Business IT can help organisations identify practical AI use cases, assess their technology and data foundations, and establish a controlled starting point. Support can include reviewing permissions, data handling and repeatable workplace processes before wider adoption. The focus is on helping teams gain useful experience while maintaining appropriate oversight and avoiding unnecessary complexity.
