A conversation from the shop floor between Commercial Director Dave and MD Justin about some of the challenges our clients have and how AI can be used to help solve them.
With most businesses having websites, customer platforms, portals and eCommerce capabilities, what are the common challenges they face with their current systems?
At a higher level, a longstanding challenge for businesses has been managing data and establishing a single source of truth. For B2C companies with eCommerce sites or customer platforms, consistently integrating data from various acquisition channels, onsite and in-application usage, CRM systems, and third-party sources into a centralised system remains a persistent challenge. Different teams within the organisation need access to this unified data, but achieving this isn’t easy. Solutions often address parts of the problem, but then fail to scale. Or new technologies emerge that are incompatible with existing systems. This challenge is equally significant for SaaS businesses, which need to track sales activities, such as licence purchases, and platform usage. Merging this data with marketing information adds another layer of complexity.
Over the past year we've seen a reduction in marketing spend (Gartner), particularly as a percentage of clients' revenue. Clients are faced with tough choices about how to allocate their budgets to maximise impact. In a typical marketing team there’s competition between maintaining and adding new features to the website, producing content and funding ad campaigns. To build efficiencies and cut costs, clients are increasingly turning to automation and generative AI in these areas.
A third challenge is staying up-to-date with the rapid pace of innovation and maintaining a competitive edge with customer platforms and SaaS offerings. New competitors emerge every few months, forcing businesses to ensure their products and customer experiences are at least on par by regularly releasing new features and updates. Without investment in technology updates, companies risk inefficiencies in their back-office operations due to outdated systems. Many businesses are still performing manual data extracts, cleaning data manually, and uploading it into other systems just to conduct basic analytics or gain insights. Ideally, they need to be more responsive, accessing real-time or near real-time data, rather than relying on a process that takes a team member several days.
What philosophy guides our approach in helping businesses keep their platforms at the forefront of technological capabilities?
Our commitment to continuous improvement is fundamental to this, shaping not only our internal operations, but also our collaboration with clients. We have a proactive, incremental approach to refining websites, platforms and eCommerce stores. By taking small, regular steps and embracing an experimental mindset, we can develop proof-of-concept models and conduct simple user tests to validate the value of new features. If a feature proves to be beneficial, we incorporate it into our agile, sprint-based workflow, enabling us to release new enhancements quickly each month, thereby continuously enhancing the customer experience.
This investigative, small-steps approach is presumably not only efficient, but also significantly more cost-effective?
Yes, definitely. It's often more effective to identify exactly what needs testing, and just implement that - for example, by using no-code platforms and simple tools to create landing pages. We use a component-based system that allows us to quickly and cost-effectively build UI elements, enabling us to test ideas before developing them into a full product.
When would you say we first identified the potential to integrate AI, such as smart features, into one of our existing client’s platform?
In the early days, going back about 5 years, it was about using machine learning algorithms to help with classification tasks. Where there were large volumes of content and data being manually classified and given taxonomies by editors, one of our first implementations would have been automating that process, and training an algorithm around the specifics of a client’s particular knowledge-set/data domain.
How did that set the foundation for the work we’re doing now?
Augmenting clients’ content and data using AI technologies, with generative AI having come so far, large language models enable us to do that both from a back-end perspective of cleaning, tagging and preparing data - and content - for different use cases. It also now lends itself through the natural language processing part of what’s baked into the modern AI stack to enable those interactions with end users and enable them to query data and content, and interact with it - using natural language.
There are common areas where AI has improved our clients’ customer platforms, regardless of sector. Our areas of expertise include research and insight, and learning (particularly eLearning), education and upskilling.
Large language models now enable us to augment clients’ content and data using AI technologies to clean, tag and prepare data and content for different use cases.
Through the natural language processing capabilities inherent in modern AI technology, we can facilitate interactions with end users, allowing them to query and engage with data and content using natural language.
In what ways does AI enhance efficiencies and customer experiences across our familiar client sectors, such as research & insights, and eLearning, training and upskilling?
In the research and insights sector, machine learning and AI excel at organising and classifying data. Many companies have a mix of internal content and data, as well as data from external sources. Traditionally, aligning these disparate sources and formatting the data uniformly for widespread use was a time-consuming task typically handled manually by a data scientist role. Whereas now, much of it can be automated quite quickly. There is also the potential for generating synthetic data and audiences to speed up gaining insight into new product ideas and ways to market, as well as validating blue-sky thinking before conducting more traditional user research.
Using AI to help remove friction from key user needs, such as finding information and understanding insights, is a key use case. This can be achieved through advanced search capabilities using RAG frameworks and generative AI to provide more accurate results, or by implementing genuine Virtual Assistants that users can interact with to accomplish tasks and reach their goals.
There are, of course, concerns regarding privacy and accuracy that must be considered when implementing these solutions. Spicerack’s latest approach and frameworks address these issues.
Additionally, we’re introducing smaller AI-augmented interactions, particularly in form-filling tasks, using NLP generative AI to improve the speed of data entry and the quality of the data captured.
Over the last two years, we've collaborated on AI projects with numerous clients in both the UK and US. What commonalities do these clients share in their perspectives on AI for their current and future business requirements?
Businesses are enthusiastic about the potential benefits AI can bring, and many have a good grasp of its possible applications. However, across both the US and the UK, businesses are still not sufficiently informed to build out a strategy and roadmap for testing and validating use cases across their business and product/services portfolios. Moreover, there's some apprehension regarding privacy and transparency issues, and uncertainty about integrating generative AI into products, with concern about maintaining robust guardrails to protect brands and reputation. And of course, worries persist about potential data breaches when data is shared with external providers like OpenAI.
How can businesses be reassured they’re protected from such a scenario?
The most effective way currently is to self-host large language models and the necessary infrastructure, rather than use public APIs. Major providers like Amazon have woken up to this common requirement. For instance, OpenAI’s enterprise plan now allows businesses to use their own trained models, and they’ve strengthened their privacy policy to mitigate risks. Amazon Bedrock offers comparable solutions. In the coming months, these cloud services will enhance their security measures and improve SLAs and privacy policies for paying customers.
For businesses overwhelmed by the AI hype and looking to integrate it, how can specialists like Spicerack assist them in navigating the available options successfully and cost-effectively?
Some businesses have already started recruiting AI leads to explore how AI can enhance internal workflows and improve efficiency. They’re also examining how to develop their products and services to leverage the benefits of AI integration.
As an agency, we’re well positioned to assist businesses not yet at that stage, but who need help with developing AI strategies tailored to their data and content. We offer guidance on assessing its value and exploring how AI can enhance operations and efficiency. Additionally, we specialise in improving customer experiences and creating new features and products that leverage AI capabilities, enabling businesses to monetize their content and data in innovative ways.
To wrap this up with some blatant self-promotion, what aspects of Spicerack’s digital media and tech heritage make us particularly well-suited for integrating machine learning and AI into our clients’ products?
We’ve been using AI for some years now, so we’re experienced in the pros and cons of different models, and ways of mitigating some of the risks and concerns around privacy and guardrails. We come from a very customer and user experience heritage and mindset. Innovation and emerging technologies are about improving people’s interactions with technology, for example in their working day, or just making it easier to find information and interact online.
Making them want to come back for more - and ultimately stick around?
Yes, and ultimately making the whole experience frictionless and more enjoyable. Because we come from that background, and have a deep technical understanding of how to integrate AI, we can make the connection between good user experience and the latest emerging technologies which support it. Currently, AI is that new technology. But that will change, and before long we'll be focusing on something new. However, the core principle of enhancing user experience by integrating cutting-edge technologies into workflows will remain constant, and that’s ingrained in our experience as an agency.