From 'AI' to 'assistant': the reframing that unlocked adoption at bet365
How we built our AI assistant to support decision making and improve efficiency securely, responsibly and at scale
When generative AI broke into the mainstream, the conversation across most organisations followed a familiar pattern: excitement on one side, caution on the other. At bet365, we wanted to move fast, but thoughtfully.
We could see the potential immediately: the ability to summarise complex information, accelerate the software delivery lifecycle and support decision-making across the business was compelling. But so too were the risks. Uncontrolled use of public AI tools raised real concerns around data exposure, confidentiality, compliance and cost.
The goal was clear: embrace the potential of AI while ensuring data security, compliance and trust across every team. The result was assist365, bet365’s in-house AI platform that puts the power of generative AI safely into the hands of thousands of employees.
From controlled testing to company-wide platform
assist365 began as a simple idea: create a secure environment where people could explore generative AI safely.
Our first iteration was effectively a secure wrapper around ChatGPT. It allowed employees to test the benefits of AI in a way that kept all prompts and outputs contained within bet365’s infrastructure.
This gave us two things. First, it removed the immediate risk of data leakage. Second, it allowed us to observe how people wanted to use the technology in their day-to-day work. What we learned quickly was that the opportunity was far broader than we initially imagined.
As usage grew, so did the platform. Today, assist365 is a multi-model AI assistant, integrating both OpenAI and Google Gemini models, and supporting a wide range of use cases across the organisation. But the real shift wasn’t just technical - it was cultural.
Designing for everyone, not just specialists
From the outset, we were clear that this couldn’t be a tool for a small group of experts. If AI was going to deliver meaningful value, it had to be accessible to everyone. That meant treating usability as a first-order concern.
The aim was, and continues to be, to make people’s lives easier through a personal assistant that increases efficiency by supporting the execution of mundane administrative tasks; and in doing so, allow everyone to focus on higher value initiatives and interactions.
To ensure everyone could use it, we deliberately designed assist365 to feel familiar. The interface mirrors patterns people already recognise from consumer tools, removing the need for training and lowering the barrier to entry.
At the same time, we ensured consistency with our existing internal systems. If someone had used one of our Business Systems Solutions tools before, assist365 would feel instantly recognisable.
The goal was simple: make it obvious, make it intuitive, and make it safe, because adoption doesn’t come from capability alone. It comes from confidence.
Balancing innovation with control
One of the biggest challenges in enterprise AI adoption is balancing access with oversight. We approached this by making security foundational. The system is hosted on bet365 Google Cloud infrastructure which connects to LLMs in our private tenants. This ensures that all prompts and outputs are securely contained and usage is fully auditable.
Costs are tracked centrally, allowing the business to monitor usage. This enables us to keep monthly spend tightly controlled while ensuring every employee has more than enough access to do meaningful work. All interactions with assist365 remain within bet365’s environment, with full auditability built in. That gives teams the confidence to work with sensitive financial, regulatory, and operational data without fear of exposure.
At the same time, we introduced clear cost controls through a token-based allowance system. Every interaction has a transparent cost, and usage can be monitored and adjusted centrally.
This creates an important balance. People are encouraged to explore and experiment, but within a framework that keeps spend visible and controlled. The result is enterprise-scale capability delivered at minimal cost.
The power of reframing: From ai365 to assist365
Interestingly, one of the most important decisions we made wasn’t technical at all. Originally, the platform was called ai365. It was accurate, but it had an unintended consequence. People approached it cautiously. AI still felt experimental, even risky. So, we reframed it.
assist365 better reflected what the platform is: a digital assistant designed to support everyday work. That shift in language, combined with clear assurances around governance and safety, had a dramatic impact. Adoption accelerated, and the platform became part of people’s daily routines.
It was a reminder that successful AI adoption is as much about perception and culture as it is about technology.
From simple tasks to sophisticated workflows
Today, assist365 is used to support Finance, Legal, Tax, Technology and beyond. Some use cases are straightforward: drafting emails, summarising documents, generating job descriptions and managing staff schedules. Others are more advanced.
We’re seeing non-technical users transform Excel reports into interactive dashboards for senior management. Software Engineers are using AI to understand legacy code and support and document greenfield development. Finance teams are analysing legacy spreadsheets and understanding complex macros. Regulatory teams are interpreting new compliance requirements and assessing their impact.
In each case, the pattern is the same. assist365 doesn’t replace human judgement, it accelerates it. It removes friction from the process, allowing people to focus on higher-value work.
Scaling adoption without losing control
In 2025 alone, we onboarded more than 5,000 employees onto assist365, with the vast majority becoming active users. Over that period, the platform has supported hundreds of thousands of sessions, all within a controlled cost framework.
That scale matters. It demonstrates that enterprise AI adoption doesn’t have to be fragmented or experimental. With the right foundations, it can be consistent, governed, and widely accessible. Just as importantly, it shows that innovation doesn’t need to be centralised. When you give people the tools and the confidence to use them, innovation happens everywhere.
A cultural shift, not just a technical one
What began as a secure wrapper around ChatGPT has evolved into a multi-model assistant connecting to OpenAI and Gemini, supporting text, image and soon video generation. But its real success isn’t just technical, it’s cultural. By re-framing the project from ai365 to assist365, we’ve helped shift perceptions from risk to reassurance, driving confident adoption across departments.
AI is no longer something reserved for technical teams or isolated projects. It has become part of how the business operates day to day. People are more willing to experiment, more confident in exploring new ways of working, and more open to rethinking how tasks are done. That shift is difficult to measure, but its impact is undeniable.
At scale, small improvements in how individuals work quickly compound into meaningful organisational change.
What comes next
Today, assist365 is a trusted everyday tool used across much of the business. It enables employees to summarise complex documents, generate code documentation and translate regulatory material, all within a compliant, centrally managed environment.
We’re continuing to expand assist365’s capabilities, including image and video generation, richer document handling, and deeper integration into existing tools. Future enhancements include richer multimodal capabilities, department-specific assistants and deeper integration into workflows — all while maintaining the same balance of innovation, control and responsibility that has defined the project so far.
The longer-term ambition is to embed assist365 directly into the flow of work. Not as a separate destination, but as a contextual assistant that surfaces insights, recommendations and automation exactly where they are needed.
At the same time, our focus remains unchanged. Innovation must be balanced with control, capability must be matched with trust and technology must always support people, not replace them.
assist365 is our blueprint for doing that. A way to bring safe and smart AI to the enterprise, at scale, and with people firmly at the centre.