If you’re using ChatGPT within your organisation, are you ready to go to the next level?
Enter RAG AI – described by Kieron White, founder and CEO of Engine, as “the most important innovation that businesses should be looking at right now”.
So what is RAG AI – and how is it transforming business operations?
“It brings together the benefits of Large Language Models (LLMs) like ChatGPT but enables that kind of AI to work inside your organisation, trained on your documents and data,” White tells BusinessCloud.
“RAG AI therefore allows teams to access the 80% of company data that isn’t in structured databases: it exists in documents like emails, notes, marketing collateral or social media posts.”
RAG, which stands for retrieval augmented generation, is a relatively recent development but it’s driving a lot of the hype around AI as a tool for business. It uses a method called Generative Search to look through large sets of data that can be explored while remaining private.
“This kind of search is also much better than simple keyword searches because the AI understands the meaning behind questions, providing more accurate and relevant answers,” says White.
“RAG AI hugely reduces the risk of hallucination, which you get when AI is looking more widely across the internet to form a response from a mix of reliable and unreliable information.
“Businesses can obviously benefit from RAG AI with better-informed decision-making, but it also means they can offer more consistent customer service by generating good quality information without investing in a lot more training or hiring more staff.”
London-based Engine helps its clients to create new opportunities using this fledgling technology, which White says is not just for big businesses.
“Some of the most well-known enterprise solutions are out of reach for a lot of SMEs: an enterprise version of ChatGPT costs upwards of £100k per year, and Microsoft’s Copilot starts at around £20 per person, per month,” he admits.
“But the underpinning technology has become increasingly accessible and affordable, making it viable for start-ups and SMEs as well. Companies like ours offer more affordable, scalable solutions that can be tailored to the needs of smaller businesses from as little as £10k per year to build and run.
“This democratisation of AI technology is what we find exciting: it means that companies of all sizes can take advantage of the power of RAG AI to improve and compete.”
Finance, healthcare, retail, and legal services are leading the adoption of RAG AI in the UK, he says.
“In finance and elsewhere RAG AI is now widely used to improve customer service – giving staff immediate access to the latest regulatory information and market data. Healthcare providers use it to improve patient care by accessing medical research in a similar way.
“Retail businesses increasingly use RAG AI for better inventory management and to offer more personalised online shopping experiences. Legal firms have also been at the forefront of using RAG AI for document analysis and legal research.
“These uses have all helped manage data overload and tailor information retrieval to each person’s needs. But we’re starting to see that RAG AI is good for SMEs too, helping businesses grow and improving outcomes in areas like education and social care by providing quick access to relevant information and reducing admin burdens.
What are the main challenges companies face when integrating RAG AI and how can they be overcome?
“Money is the first challenge. For most businesses, AI is a new budget line for 2024/25 that won’t be instantly offset, although in time there will be savings and growth from AI solutions,” answers White.
“Integrating RAG AI into business operations can also present challenges around data privacy and the skills you would need to manage and maintain AI and your cloud infrastructure in-house. We think those can be managed by choosing the right expert partners.
“Resistance to change is a real challenge too, which feels very familiar after more than 20 years as a digital transformation consultant.
“The other challenge is good data. AI can only work with the data it’s given, so collecting and storing the right data is essential, even if AI can handle it being poorly structured. For example, customer feedback can only be used to train AI if the data can be accessed.
“This means bringing together email feedback, social media, call centre and ticket-desk data to create a full picture that an AI tool can use to answer future customer questions or provide insights into customer satisfaction.”
White likens the arrival of RAG AI to the industrial revolution: “Over the next decade, RAG AI will change the landscape completely. It will drive innovation, increase productivity, and enable more personalised and efficient services across every sector.
“The risk is that gaps will emerge between organisations that can and do invest in AI and those that don’t. We should expect to see a more data-driven approach to decision-making, with businesses using AI to gain deeper insights and stay ahead of the competition.
“The widespread adoption of RAG AI will likely lead to the creation of new industries and job roles, including those who will be needed to manage data. Data – in any format – is the rocket fuel for an AI solution. Better data means better AI output.
“As AI capability increases and technologies shift, their foundation will always be the data that they are trained on. Businesses that understand this and have the best data sets will emerge as the leaders in a future AI-first world.”
Tech unicorn to ‘invest heavily’ in UK after $800m investment