Unlocking the Power of AI for Small Businesses: Biztech Roundtable

On 4th February 2026, Biztech hosted its inaugural Biztech Innovation Roundtable webinar. Exploring how small and medium-sized enterprises can harness artificial intelligence. Chaired by Professor Ben Allen, a visiting professor at the University of Surrey’s 5G/6G Innovation Centre. The session featured a keynote presentation from Dr Ed Braund, Head of School of Computing, Engineering and Creative Industries at the University of Bedfordshire. Followed by a panel discussion with contributions from Henry Kafeman (engineering consultant at HDK Solutions Ltd. and Biztech director). And Axel Segebrecht (technology consultant at Be Braver Online). With questions from attending business owners and technologists.

Dr Braund opened by introducing the Luton AI Initiative, a civic organisation he directs that has engaged over 500 organisations. Luton AI has delivered more than 60 AI projects since its launch. Its mission is to help businesses navigate AI adoption at no cost, offering support ranging from skills training and security development to incubating start-ups.

Presentation

The presentation provided a grounding in what AI actually is and where it stands today. Dr Braund explained that artificial intelligence rests on three pillars: data, models, and computing power. While the underlying mathematics has existed for decades, recent advances in data availability, processing power, and a model architecture called Transformers have enabled the current wave of large language models such as ChatGPT, Copilot, and Gemini. He stressed that today’s AI remains narrow, excelling at specific tasks. But it is proving brittle when taken outside its trained domain. The notion of artificial general intelligence, he said, is still firmly in the realm of science fiction.

AI Avator Demo

To illustrate AI’s practical capabilities, Dr Braund used ChatGPT to generate survey questions on the spot. He then fed them into a presentation tool called Gamma to produce a polished slide deck in moments. And created an AI avatar to present the questions to the audience. The demonstration was deliberately playful but made a serious point: with the right training, readily available AI tools can dramatically accelerate everyday tasks.

Dr Braund outlined six core opportunities for SMEs: automating repetitive tasks, accelerating creative workflows such as marketing copy and image generation, reducing costs through predictive maintenance, optimising workflows, personalising customer experiences, and improving data-driven decision-making. He also highlighted the importance of staff empowerment Noting that some of the most impactful work Luton AI has done involves equipping employees with approved AI tools and clear usage guidance.

Important Questions

However, the presentation also raised important cautions. AI models are trained on vast historical datasets that can embed biases. Dr Braund demonstrated this by showing AI-generated marketing images for university courses that reproduced gender and ethnic stereotypes. He urged businesses to scrutinise AI outputs to ensure they reflect their values. He recommended a measured adoption roadmap: choose the right problem, define what success looks like before starting, build a small pilot, evaluate it against clear criteria, and only then decide whether to scale.

Panel Discussion

The panel discussion that followed brought additional perspectives. Henry raised concerns about the pace of change, warning that SMEs risk investing in workflows or tools that become obsolete within months. Dr Braund agreed, reinforcing the message to avoid adopting AI for its own sake and to focus on clear return on investment.

Axel Segebrecht shifted the conversation towards risk and security. Drawing on over 25 years of consultancy experience, he highlighted the dangers of data leakage, reputational damage, and the sharp rise in AI-powered phishing scams. Which he said have become far more effective because even non-technical bad actors can now craft convincing, targeted emails using chatbots. And prompts sourced from dark web forums. He emphasised the importance of sandboxing AI tools. Particularly agentic systems that run locally on a user’s machine.

Participant Contribution

Peter, a participant with a banking background running development teams, expanded on the data security theme. Describing the challenge of running AI pilots in controlled environments when staff may download tools with unknown security properties. Tom, who heads a digital agency, echoed these concerns and asked about trustworthy vendors and local model deployment as a way to air-gap sensitive data.

Dr Braund responded with practical suggestions. Recommending edge devices such as the NVIDIA DGX Spark for running models locally in a stateless configuration. And ensuring no training occurs on company data. For cloud-based solutions, he pointed to established vendors like Microsoft and AWS. Who offer robust security documentation and are willing to walk organisations through their requirements. He also shared a cautionary tale about a legal firm that banned AI tools outright, only to discover staff were using personal devices and accounts instead. Stressing that empowering people to use tools securely is more effective than prohibition.

Professor Allen also recommended the Royal Society’s published reports on AI as a well-founded resource for further reading.

Closing remarks

The inaugural Biztech Innovation Roundtable session closed by Biztech Chair and Founder of Briteyellow Fredi Nonyelu. Confirming plans to make the roundtable a recurring series, with the next event dates to be announced by the end of February.