Making the case for building an AI roadmap

By Iris de Jong
Blog header image for making the case for building an AI roadmap Inviqa blog

Today – as with any digital transformation and technological advancement – there’s a lot of hype and fear around artificial intelligence (AI). Hype about what it is capable of and fear it might take your job. To dispel the myths, highlight the opportunities, and explain how organisations can get started with building an AI roadmap to leverage this fast-evolving technology, Kaustav Bhattacharya, Chief Technology Officer at Inviqa, and Ankit Bharadwaj, AI Lead at Gate One, hosted a morning event at the Havas Village London.

The session kicked off with Kaustav asking the room how attendees have been using AI to date. While everyone in the room had used AI tools to some extent, the main use case had been for content creation, with some using chatbots with varying degrees of success. 

With an understanding of where everyone is at with their AI adoption, Kaustav and Ankit delved into the history of AI, shared examples of how AI is being used, and explained how attendees could get started with the meaningful adoption of AI tools in their organisation.

The evolution of AI

In the early days, Kaustav explains, AI was all about probability and prediction – it was very functional. Today, AI technology is more human – catapulted to the forefront of everybody’s mind (and at the tips of everyone’s fingers) with the launch of ChatGPT in 2022. 

In the short time since then, many technology vendors have started offering generative AI as part of their solution. Smaller AI disruptors – like Dall-E, ChatGPT etc - are partnering with the big vendors – like Microsoft, Adobe, and AWS - to support AI solutions as demand grows and the technology matures.

But why is this important? Investment in AI is only going to increase, so organisations need to understand its potential for them and plan adoption of this technology accordingly. While AI is evolving at the rate that it is, now is the opportunity to start thinking about what this means for your business, for your team, and how you should start building your strategy and roadmap and determine what use cases you can implement these solutions for. 

How can organisations begin to incorporate AI in their day-to-day?

According to both Kaustav and Ankit, the key is to start experimenting. Play with the technology to see how it works and understand how it could help your organisation improve current workflows.

Havas, for example, Ankit explained, has a ‘Play don’t publish’ policy. This allows people to experiment with AI technology, see what it’s capable of, and become proficient in the tools while not allowing these experiments to go ‘live’ in client work.

This kind of experimentation shows what the opportunities of AI are and where its limitations lie, so more informed decisions can be made on how this technology could then be incorporated into day-to-day operations.

Examples of what can be done with AI

Kaustav then shared a few examples of these types of experiments, using AI tools to create something tangible – even if only for fictitious clients. 

In the first example, a video ad was created using AI for the fictitious beverage brand Double L lemonade.

How was it done?

He started by researching the brand using Copilot. The output from this was then fed into Adobe Firely to create the imagery which was then passed into RunwayML to bring motion effects into the advert. Music (because how could you have an ad without music) was generated through Mubert, while ElevenLabs was used to clone a voice to generate the voice-over.

The thing to understand about this example, though, is not the impressive video ad created using AI tools, but rather the creative human input that was still required at each step of the process: making decisions, finetuning outputs, driving the ‘creative process’, and bringing human connection to the finished product.

Without this human element – also known as keeping the human in the loop – it’s unlikely the ad would have landed as positively as it did.

In the second example, Kaustav showed how the AI wireframing tool, Relume, can be used to quickly stand up a new website and speed up the timeframe between an idea and workable digital prototype. All that was needed was a prompt and Relume took care of the rest, even inserting placeholder copy into the wireframe. These wireframes were then imported into Figma where designers and copywriters were able to refine the design and content to create the final site.

The final example Kaustav shared was from Contentstack. Like many CMSs, it offers a generative AI tool for creating content for a website page. But the platform has taken it one step further to support Retrieval-Augmented Generation (RAG). 

How is this different? Rather than generating content based on worldwide data, RAG allows you to upload your brand documents into the platform. Now when you put in a content prompt, it generates copy based on the information from these documents, so the output is going to be more relevant and accurate for your business, and in your brand’s tone of voice.

What are the wider benefits of AI?

What these examples show is the variety of use cases in which AI technology can be utilised in organisations to help people improve processes or speed up their work. 

Indeed, one of the biggest impacts that the use of AI has had is speeding up processes.

Early adopters of the Microsoft Copilot tool, for example, reported they were more productive (70%), felt the quality of their work improved (68%) and were faster at tasks like searching, writing and summarising (29%).

AI also has the potential to remove or speed up the ‘drudgery’ work. Those who work in web development probably aren’t most excited about creating sitemaps and wireframes, preferring to get stuck into the actual designing of a beautiful, functional website. Using a tool like Relume can speed up this process, allowing them to spend more time and effort on the actual design and functionality of the site.

Another example of where an AI tool can help speed up drudge work is in the product information management (PIM) space. If you’re a retailer with a large catalogue of products, the opportunity with generative AI is that it can crawl through your entire catalogue and automatically write new product descriptions based on the information you already have in your system. It could also then translate these descriptions into multiple different languages.

While AI isn’t going to replace humans, those who have AI skills may have a competitive advantage when it comes to landing their next job, with 66% of leaders saying they wouldn’t hire someone without AI skills.

How businesses can embrace the opportunities of AI

Even with the potential that AI has to offer, much of what we’re still seeing right now is businesses focusing on everyday AI, like basic Q&A with generative AI tools. People are dabbling in the technology to see what its capabilities are, but not fully committing to better understanding how the technology can benefit the wider business, or how it can be embedded organisation-wide.

To take that next step, organisations need to understand how to move AI adoption forward: what should AI investment look like, and how does AI relate to the corporate strategy, vision, and objectives? What are the key measures and outcomes that organisations should be looking at when considering the adoption of AI? 

This is where building a roadmap and strategy becomes crucial – and it should be the next step for any organisation if they’re considering utilising AI.

What should be reassuring is that building a roadmap and strategy for AI is no different from any other tech transformation: it’s about people, processes, technology, and data.

Getting started with creating an AI roadmap for your organisation

The first step of creating an AI roadmap for your organisation is aligning on business objectives and landing on a consensus of why you want to use AI and how you think it can boost workflows and productivity.

Then it’s about evaluation – getting your hands dirty, registering for tools, using them, testing them and establishing the use cases for each of these tools.

When you’re doing this testing, you do want to be careful about which tools you’re using and how you’re using them. You don’t want your data or your customer’s data to go public, so read the Ts & Cs or speak to experts about which vendors provide the right level of security you need.

Then, when it comes to mass adoption across your organisation, how do you train everybody so they are empowered to use these tools? How do you make sure everyone’s aware of the technology available to them to use, and how best to use these and incorporate it into their everyday lives?

The graphic below shows the four steps for creating an AI roadmap:

A graph showing the four steps to creating your AI roadmap: Alignment, Evaluation, High-level Plan, Roadmap

What's worth keeping in mind is that you don’t need to wait until you’re at step four to start investigating and building your AI proof of concept or sorting the data which will drive the AI. Both can be done during the earlier stages of the process. Then when it's time to adopt the chosen AI solutions, you know how these can be utilised within the organisation and the data is ready to go.

And what about the ethics of AI?

Of course, there can’t be a discussion about AI without a discussion about ethics.

There have been worrying headlines over the years about how AI has gotten it wrong: gender stereotyping, ethnicity biases etc. A lot of this comes down to the data on which the AI has been trained. But this is improving as more people from around the world (and not just the tech bros from Silicon Valley) are creating generative AI tools.

But when implementing AI in your organisation it’s important to understand that bias is often baked into the technology and how you will mitigate these risks.

You must consider data integrity when exploring AI use cases and determine what your data protection policy is, understand the output your AI is creating, and who’s reviewing that output.

AI adoption in your organisation isn’t going to happen by happy accident. As with the adoption of any technology it requires understanding what the objectives are and how the technology will help achieve these. It’s also important that people are brought along on the journey – because AI isn’t there to replace people, rather it’s there to augment and complement their skills so they can do their job better.

If you’re interested in kick-starting AI adoption in your organisation, we’d love to chat. We’re offering a one-day AI accelerator which includes a half-day Gate One workshop to help you kick-off your roadmap creation and a half-day proof-of-concept supported by Inviqa.

Get in touch for more details.