On 18th June, at the very start of the UK’s mini heatwave, DeeperThanBlue hosted an in-person event at The Victoria in Sheffield. Our guest speakers – Carolyn Shepherd, Duncan Payne and Mark Ketteman – joined attendees to explore how AI is evolving, what it’s already being used for, and how businesses can adopt Agentic AI (as well as other flavours of AI) to streamline operations, integrate systems, and support growth. While the rest of Sheffield basked in the sunshine, we were busy soaking up ideas and inspiration.
It had been a while since we’d held an in-person event. In recent years, we’ve relied on webinars to bring people together to share knowledge, ideas, and experiences. From those virtual sessions, one common theme emerged: many businesses are turning to AI to reduce the time spent on repetitive, manual tasks.
This event was about going a step further. We wanted to bring together the AI-curious in one room to explore the exciting potential of Agentic AI in a business context, and to offer practical insights on how it can be used meaningfully and effectively. Real-world examples were shared, including Duncan Payne’s take on how AI is revolutionising movie production and dubbing, as well as thought leadership from IBM, a global pioneer in enterprise AI and more recently Agentic AI.
The afternoon featured four presentations from guest speakers and from DeeperThanBlue, as well as a panel discussion, all packed with insights and actionable takeaways. There was also a Q&A session, along with plenty of time for networking and enjoying the refreshments provided by The Victoria.
We’re trying to grow what we do rather than shrink what we’ve got. Duncan Payne, Zoo Digital
Keep reading for a summary of the afternoon, including highlights from the talks, the panel, and audience questions. You’ll also find reflections from attendees on where they are in their AI journey, plus some practical resources and quick tips to take away.
Let’s dive in.
AI: A Year in Review – Stephen Birch and Chris Booker, DeeperThanBlue
We kicked off with a warm welcome from Stephen Birch, Marketing Manager at DeeperThanBlue. Stephen shared some examples of how the team is using AI to save time, from helping with event planning and content ideation to image creation and generating the script for our website chatbot. And depending on the use-case, Stephen estimated that using AI to assist him in his role was saving him 10-20% of his time every week, simply by overcoming thought-blockers or reducing the number of repetitive tasks he was undertaking.
Then, Chris Booker, Sales and Marketing Director, shared his perspective on the current AI landscape and the powerful potential of using AI video to get internal experts in front of potential customers in video much faster and with much lower costs. Chris observed that the AI world has progressed ‘even further and faster’ than they had envisaged, and dubbed 2025 as ‘the year of agentic.’ To demonstrate the speed at which the use of AI is growing, he pointed out that the IDC – the International Data Corporation – expects 1 billion new AI and low code applications to be developed and running by 2028.
He explained that, unlike AI assistants or generative AI, AI agents can reason, plan and execute with minimal human input. They can operate autonomously on tasks, access information on their own, and interact with other software applications including other agents. ‘Agents act more like a digital employee than a passive assistant,’ Chris continued. ‘They operate based on intent, context and longer-term memory. They will collaborate with one another and form a business process.’
While chatbots can be rigid and fairly narrow in scope, AI agents are pre-trained, reactive, and proactive. They are goal-orientated, adaptive and self-improving. The future will see AI agents helping with finance, HR, calendaring, email, travel, logistics, procurement, lead generation, campaign execution, and customer support, with multi-agent systems coordinating across teams and departments to solve complex business challenges.
Chris also outlined some of the business impact of agents, such as efficiency, customer experience, and always-on business operations, with ‘9-5 businesses’ able to respond to people 24/7. AI agents also allow rapid scaling without additional headcount and can be considered a new ‘digital workforce’, giving businesses a competitive edge in agility and innovation.
AI that Lands: Where Strategy Meets Human Experience, Carolyn Shepherd, Emmeline.ai
Carolyn Shepherd is the founder of Emmeline.AI Consultancy, which helps R&D and HR leaders harness AI without the jargon or overwhelm. She is also the author of The ChatGPT Advantage, which she wrote in order to make AI less scary. Through stories, Carolyn says, people can connect and find their courage in a time of momentous technological change.
Carolyn asked the question, ‘Why is AI not delivering results?’ Her answer is that people need to catch up. ‘People are not moving as fast as we’d like them to. It’s quite scary to a lot of people – most people go to work, do their thing and don’t give a lot of thought to business processes.’ But once people have got the fear out of the way, they can open their minds to learn something new.
Carolyn has found that the people who didn’t start using AI simply lacked confidence, even to ask questions to a machine, because they would be afraid of asking the wrong thing. The solution, she argues, is to build people’s confidence through stories so they can see people like themselves using AI to automate the tasks that drain them.
‘Start by asking yourself, “What’s one task that drains you?” Then, ask ChatGPT, or whichever AI tool you and your employer favours, “Help me reduce or automate this task,” and see what becomes possible.’
‘The pace is gathering and AI agents will soon be swarming,’ Carolyn added. So, it’s time to get involved. Carolyn compared thinking about AI to thinking about electricity, in the sense that it would be strange to ask people, ‘What’s your electricity strategy?’ Just like electricity, AI will be a fundamental and embedded thing in our lives and work – something we don’t need to think about. Therefore, we need to focus on the outcomes we want, not the strategy itself or the tools we use to achieve these outcomes.
Leveraging AI for Enhancing Media Localisation – Duncan Payne, ZOO Digital
Duncan Payne is Head of Software Engineering at ZOO Digital, a Sheffield-founded company that provides cloud-native media localisation services globally. ZOO Digital’s mission is to make life easier for the people who entertain the world, and the team works with clients such as Netflix, Facebook, and Disney.
ZOO specialises in dubbing and subtitling film dialogue from one language to another, and one significant challenge in this area is finding professional voice actors, particularly where child voice actors are involved, due to the strict legal limits on the amount of time that children can work. Duncan shared an example of one child actor who could do one hour per week, and explained that one hour of recording would produce just 10 minutes of usable audio.
So, ZOO worked with the Speech and Hearing Research Group at the University of Sheffield to take a recording of a grown-up’s voice and use AI to make it sound like a child’s. This means changing both the pitch and the style of delivery. It took lots of training data and examples of speech from both adults and children. The project was supported by Knowledge Transfer Partnership, a UK government-funded program that facilitates collaboration between businesses and universities (or other knowledge-based institutions) to drive innovation and growth.
Other areas where ZOO uses AI include talking through ideas and working out problems, test generation, code completion and developing management practice.
Duncan believes AI is going to revolutionise the media industry, but that there is no intention to replace real actors, and in fact, AI is going to create new job prospects for people. Skilled human talent will remain crucial, especially for premium entertainment content, and the roles of creative professionals are likely to transform,’ he said.
After Duncan’s presentation, we paused for a tea and coffee break, and spoke to attendees about where they are on their AI journey.
Gavin Kimpton, Pipe Ten, said that everyone on the team has a paid ChatGPT licence, which they mainly use internally for things like transcribing and summarising meetings, and identifying key action points. Pipe Ten also uses it to help with coding and creating scripts. Staff weren’t resistant to AI, but staff weren’t sure how to use it. When they first introduced ChatGPT, they sat down with everyone on the team to see how they would use it, to support adoption.
Bryce Elvin, Motorscrubber Ltd, has been diving into AI recently and came up with a time-saving solution that impressed the leadership team. Emails from customers are read automatically by large language models, and the key data points in the email are then converted into JSON files and automatically added to the CRM under the relevant client. Bryce built this using the OpenAI-hosted GPT API. It all started with a complaint from the warranty department, and Bryce had an idea of how it could be solved. Bryce’s solution means that the human agents have more time, and he tells us that, quite rightly, the CEO found it ‘very cool.’
AI presents a game-changing opportunity for businesses of all sizes and sectors, no matter where they’re based. By adopting intelligent automation, leveraging data-driven insights, and delivering more personalised customer experiences, organisations can significantly boost productivity, drive innovation, and remain competitive in a fast-moving landscape. Kir Nuthi, Head of AI and Data at techUK
Panel discussion
Q. What makes a use case ‘AI-worthy’?
‘The problem needs to have business value, which is up to the judgment of the business, and then how much it will cost to use AI to solve it,’ said Carolyn. ‘“What tasks are draining you, and draining your time?”’ It’s the moments when you feel unproductive, and when it feels like a robot should be doing the task you’re doing, that there might be an AI-worthy use case.
‘Customers sometimes bring a problem statement that identifies an issue,’ said Chris, ‘but there needs to be a strong ROI case.’ He recommends leaning on people who have done this before and remembering that not everything needs to be solved with AI, since there are other ways of automating things without it.
Stephen adds that AI can be a catalyst to getting started on creative projects, helping you to get off a blank piece of paper or empty screen. If you’re lacking inspiration, you can ask for an idea of where to start filling that page and develop the content from there. Jake Appleyard, Partnerships Manager at The Curve, echoed this when we spoke to him in the break. The Curve uses AI to come up with ideas for content, which they then run past the dev team for feedback and a sense check. Jake then gives AI the audience and the topic, and asks it to write the content, and the experts at The Curve ‘augment’ it. ‘AI is good for refining content to make it sharper and to the point,’ Jake said.
Q. What’s been your biggest lesson in developing or deploying an AI solution – and what would you do differently next time?
‘First of all, approaching AI in the mindset that it can do anything is probably a mistake because you’re not going to meet your business’s needs,’ said Duncan. ‘Starting smaller is key. We’ve often talked about artificial assistance instead of artificial intelligence.’ He said that AI is for supporting skilled workers and despite its many benefits, one can’t assume that everyone’s on board. ‘You need to involve people early in the process, be task-specific in how you approach it, and lean into partners, people with the experience and knowledge, because that will give you a good starting point.’
‘Prepare ahead of time and give people as much of a runway as you can,’ said Carolyn. ‘People are using it, but not as you’d like them to be using it. Get people on the AI literacy journey as soon as you can.’
Q. How important is modern infrastructure in the success of AI use cases – and what are the core requirements for getting started?
Infrastructure is the unsung hero of AI. ‘Leveraging your data is what’s going to help you create secret sauce,’ said Chris. ‘Some applications make it easier than others.’
Carolyn often hears clients say that their data won’t be usable in an AI context, because that’s what their IT team has told them. So, she recommends speaking to a third party like DeeperThanBlue, and asking them to have a good look at your data, because you might be surprised. It could be that your data can be leveraged for the purpose you have in mind.
Similarly, Mark Ketteman from IBM emphasised that the way your IT and cloud systems are structured (the ‘underlying architecture’) has a huge impact on how well AI works and how efficiently it runs. He said that the architecture behind your AI systems really matters, especially when using cloud platforms.
Q. How do you align IT and business teams around an AI project, especially when expectations are high and understanding varies?
‘Variance in understanding is a huge piece of the puzzle. People don’t know what to expect,’ Duncan commented. Some people want AI to solve every problem, while others want to avoid anything that involves AI. ‘Transparency is key – liaising between teams and making sure communication is effective.’ Duncan recommends a ‘human in the loop’ approach and being clear that using AI is a strategic decision and that it’s not about trying to replace jobs, but about driving efficiencies.
‘We’re trying to grow what we do rather than shrink what we’ve got. Managing expectations is key – not overpromising. AI is not a magic bullet, it’s not going to solve all of your problems, and human expertise remains crucial.’
Q. With all the opportunities that exist with AI, what do you think are the main stumbling blocks to scaling them?
‘This is a journey of discovery, and what you’re aiming for initially is progression,’ said Chris. ‘Start small, find a pilot you can carry out for 6-8 weeks, and expand it from there. Keep a tight rein on how you iterate because the path might change, and there may be different use cases that emerge over time that you want to prioritise on your roadmap.’
‘Change is always difficult. Culture is a challenge, and the higher you go, the more ingrained it can be,’ Carolyn added. Mark Ketteman shared the same sentiment during his presentation, insisting that ‘you need to bring everyone along’ and that ‘the culture shift is the thing you need to pay most attention to.’
AI Agents: Unlocking new levels of productivity across the enterprise – Mark Ketteman and Dan Laycock, IBM
‘Right now, we’re at an IT inflection point, and a social inflection point, that reminds me of the early to mid-nineties with the invention of the World Wide Web,’ began Mark, Solutions Engineer for Digital Business Automation at IBM. ‘The big inflection point was the “dot-com boom.”’ Mark told the story of when he met with the marketing director of a major retailer in 1996 to present the idea of e-commerce to them. Were they interested? Absolutely not. They said their company ‘would never sell anything online’. Mark’s point being: we need to embrace new technologies like AI.
Always ahead of the curve, IBM began working with Agentic AI in 2017 and created AskHR to rationalise its HR system. This is IBM’s digital assistant for employees, designed to empower HR professionals to focus on higher-value tasks. 1.1 million automated tasks mean that 94% of employee enquiries are handled by AskHR, which is out now and available for organisations to adopt. It means HR staff can ‘do what they came into HR to do’: more human stuff rather than answering basic questions about annual leave and payslips. In this instance, it’s not a case of taking the human out of Human Resources; it’s more a case of maintaining the human connection where it really matters.
Mark also gave us a demo of IBM watsonx Orchestrate, a digital labour automation tool that enables workers to delegate repetitive, time-consuming tasks to a personal digital worker, powered by AI. Mark used it to scrape the public-facing content of the University of Sheffield website, which took approximately five minutes, before asking questions about booking accommodation (in both English and Chinese) via an internal equivalent of ChatGPT. The agent then responded in both languages using embedded translation functionality, and guided the user through a booking process which could be completed through the AI agent.
IBM watsonx Orchestrate also has agents that can help with a vast range of things, such as CRM, ERP, Warehouse and Order Management and Unstructured repositories such as MS SharePoint and Google Workspace. With more than 150 agents available off-the-shelf, including AskIT, AskProcurement, AskSales and AskHR, businesses already have tools they can implement straightaway, but Orchestrate makes it easy for people to build bespoke agents in a matter of minutes without writing a single line of code.
‘We give agents something to do, the agent plans what to do with it, executes, then reflects on it, maybe more than once, then there’s the output of the execution. Meanwhile, it uses knowledge (search, glossary), tasks (tools, collaborators), and behaviour (guidelines, triggers),’ said Mark. It is the definition of the knowledge, tasks and behaviour that gives humans control over the processes; while the human is in the loop, the human is also at the helm.
Reflecting on the learnings from the day, Alekhya Ankarapu, Senior Product Manager from BigChange commented, “Coming from a world of data, I can already see how AI is making a huge difference. Customers are given reports and visualisations, but AI can help users understand the data.
“Often executives don’t have time to go deep into the data and reports, so we should be able to present data insights into the UI so that they can see and understand the data in a meaningful and accessible way to help them make informed decisions on their business processes. The future, with the help of Agentic AI, will be the ability to engage conversationally with the UI to answer the questions you really want to ask, rather than rely on other people’s interpretation of your data.”
Q&A
We asked attendees to save their questions for the end, so the Q&A session was a great chance to further the conversation. People shared concerns and solutions surrounding the implementation of AI in a business context.
For example, one person asked how to monitor the accuracy of answers from AI chatbots, and Mark Ketteman explained that there are a few ways to do this when using IBM’s agentic products. You can give a thumbs-up or thumbs-down to indicate the accuracy and relevance of the response, and there’s also the option to provide written feedback. Your private instance can collect user feedback that remains private to you so you can go into the back end and see the confidence and scoring. This then helps to refine the application and understand its accuracy as part of a feedback loop.
He explained that you can adjust your settings so that the AI chatbot only answers a question when it knows exactly what it’s talking about. ‘You can give it knowledge and then tell it to only use that knowledge,’ Mark added. ‘You can tell it, “please only answer with what you know,” to reduce the chance of hallucination.’
Another pertinent question was about how ethics fit into the AI for Business conversation. Mark explained that you need to trust the models you’re using – so, do you go with OpenAI or ChatGPT5 when it comes out next month (knowing that new releases require a period of road-testing and hardening to build stability and reliability), or do you go with a self-hosted model? He points out that there are limitations with self-hosted models and suggests looking at the guardrails the model gives you. ‘The companies doing it well have powerful AI ethics teams, led by board-level executives or CEOs,’ said Mark.
Wrap up
We certainly covered a lot of ground during our afternoon together and it is clear that there are still many more questions that need answering to give people the confidence to run full tilt towards AI. But the fact that the AI-curious gathered on a hot day in Sheffield shows that interest exists.
AI harnessed correctly has the potential to help and support SMEs to help grow, scale up and accelerate their productivity. We are particularly interested to hear further details of the new AI Adoption Fund announced in the recent Spending Review, and how this may benefit organisations across Sheffield and South Yorkshire. Louisa Harrison-Walker OBE, Chief Executive, Sheffield Chamber of Commerce
We heard Carolyn talk about the cultural side of AI and the importance of bringing colleagues on board with the right training and education. Duncan shared challenges and how Zoo have improved the time it takes to create dubbing of a particular piece of media, and how much they’re improving software development.
Mark talked about ‘humans at the helm’ but we need the right humans to help drive AI forward. We also need the infrastructure and architecture to be right to introduce AI solutions that will last.
An honest appraisal is that the proper implementation of AI solutions shouldn’t be rushed. It takes planning and preparation, but that doesn’t mean that progress will be glacial.
Sheffield has become the digital forge, the crucible in which elements are combined to create something that is greater than the sum of its parts. While the tech and the experts are ready and waiting, the catalyst for change will be the people. Business leaders, technical experts, the people whose jobs will be improved with appropriate solutions where AI agents can be brought together to optimise and automate business processes.
Getting these people on side through a process of education and AI literacy training, getting them to contribute to the strategy by defining the problems they need solving, is key to acceptance and establishing AI as another tool in the toolbox. You can work with DeeperThanBlue, IBM, or other vendors to help you learn and accelerate your journey. Bringing external capability into your business helps you to accelerate so your team can go faster and further on their own. Many use cases are starting to emerge, but the key is for organisations to take their first steps on the AI journey in a small yet impactful way, measure its value and be ready to scale up.
At Business Sheffield, we’re aware that AI is a subject generating a lot of interest, but also uncertainty, among local businesses. Some are already using AI tools to improve productivity and streamline operations, while others are still unsure about where to begin.
We see the opportunities, and risks, that AI presents, and we’re actively reviewing and updating our support to help Sheffield businesses make best use of it. Our workshop content is regularly updated to ensure it stays relevant and practical, including recent sessions that explore how AI is being used in different industries, real-world case studies, and tips for using AI effectively and responsibly. Robert Lawley, Service Manager at Business Sheffield
 |
 |
 |
 |
 |
|
 |
 |