Common AI Questions
Governance and risk
Do we need an AI policy before staff start using AI tools?
Yes. Most businesses discover staff are already using AI informally, often pasting company or customer information into free tools. A short, practical AI policy sets out which tools are approved, what data can and cannot be used, and who is accountable. It is the fastest way to reduce risk without blocking progress.
What are the biggest risks of unmanaged AI use?
Data leaking into consumer tools, inaccurate outputs being used unchecked, breaches of client confidentiality or UK GDPR, and inconsistent practice across teams. None of these require banning AI. They require clear rules, approved tools and basic training.
What is ISO 42001 and does it apply to us?
ISO 42001 is the international standard for AI management systems. Most businesses do not need certification, but matching your AI policy and risk assessment with it gives you a recognised framework and assures clients, boards and regulators that AI is being managed properly. Our governance work is delivered by an ISO 42001 Lead Implementer.
Who in the business should own AI?
Senior ownership matters more than job title. In most businesses it sits with the managing director, operations director or a named board sponsor, supported by an internal advocate who drives day-to-day adoption. Treating AI as an IT-only issue is the most common reason rollouts stall.
Choosing the right tools
Copilot, Claude or ChatGPT: which is right for our business?
It depends on your systems, your data sensitivity and what your teams actually do. Copilot suits businesses embedded in Microsoft 365. Claude is strong for writing, analysis and working with long documents. ChatGPT offers breadth and familiarity. We score the options against your specific requirements rather than recommending one by default.
We already pay for Microsoft 365. Should we just roll out Copilot?
Not automatically. Copilot is often the right answer for Microsoft 365 businesses, but licence costs add up quickly and results disappoint if governance, data hygiene and training are not in place first. Sequencing matters: readiness first, licences second.
Is our data safe in these tools?
Business-grade versions of Copilot, Claude and ChatGPT offer contractual protections that free consumer versions do not, including commitments not to train on your data. The bigger risk is usually staff using personal accounts. An approved-tool list and clear data rules close that gap.
Training and rollout
Why in-person training rather than online courses?
Generic online courses teach the tool, not your business. In-person training uses your real workflows, gives people room to ask the questions they would never type into a webinar, and builds momentum as a team. Adoption is faster because every example is theirs.
How do we get staff to actually use AI after training?
Adoption fails when training is a one-off event. It succeeds when each team leaves with agreed use cases, there is an internal advocate, and support continues in the weeks after the session. That is why our engagements include adoption structure, not just a training day.
How do we measure return on AI?
Start with time. Identify specific tasks, baseline how long they take today, and measure again after rollout. Time saved converts into capacity, faster turnaround or reduced cost. We define these measures with you during discovery so the board sees evidence, not anecdotes.
Working with fortysix ai
What does working with fortysix ai look like?
A discovery call first. Then, typically, a governance session to establish your policy and select the right tools, followed by bespoke in-person training and adoption support. Every engagement is built around your business; nothing is off the shelf.
How much does it cost?
Governance sessions and training days are quoted after discovery, so you know the full cost before committing. Programme pricing varies with team size and scope.