AI adoption is the future, which is why enterprises need to find a way to make it work in their organizations where existing structures and established ways of working have not been designed to act and react with the agility required.
John Kennedy
|RSS FeedHead of Databases, Product at Aiven
Subscribe for the latest news and insights on open source, Aiven offerings, and more.
Artificial Intelligence (AI) is advancing rapidly. Much quicker than businesses can accommodate, which is leaving enterprises reeling. They’re struggling to keep up and take advantage because they’re an oil tanker in a speed boat race. But neglecting AI isn’t an option either. AI adoption is the future, which is why enterprises need to find a way to make it work in their organizations where existing structures and established ways of working have not been designed to act and react with the agility required.
The figures relating to the speed of AI development tell a staggering story when you dig into them.
According to GrandViewResearch, the global AI market is worth almost $197 billion. That's an increase of around $60 billion since 2022. Indeed, the global AI market size is expected to grow 37% every year until 2030 with the potential to contribute up to $15.7 trillion to the global economy by the end of the decade - more than the current output of China and India combined. Of this, $6.6 trillion is likely to come from increased productivity and just over $9 trillion from consumption-side effects.
Driven by fear of losing out on this opportunity, businesses are tempted to ‘dive in’, with many speeding up time to market on AI initiatives that tend to focus on two distinct areas. One is to improve internal efficiency and the other to ascertain whether and where in their own products they can build or integrate AI capabilities. This to add value to their customers and differentiate from the competition. But hitting the ‘AI button’ and hoping for results is not a strategy. Businesses can’t successfully adopt AI without a clear vision for value and if decades of technology adoption has taught us anything, it’s that you need a plan.
It’s why we believe that creating an AI Tactical Discovery team is the answer within an enterprise. A group of curated and trusted, domain technical experts for your organization. This team has the objective of using AI to solve an internally-facing process or challenges (tactical), and/or contribute to the value that an enterprise delivers to its customers (strategic). In doing so, the company will learn about AI, in a space that is close to its current value proposition and utilizes its unique datasets and domain expertise. This increases the likelihood of the learning being useful to the longer-term product strategy.
The first, and perhaps biggest, challenge for this team will be identifying the use-case and deciding whether this is something that should be built in-house or be bought. The former represents exactly the opportunity this team should be seeking in order to gain organizational knowledge while the latter involves leveraging capabilities in partner products. In reality, the likely outcome is that this team will chain together services already developed to solve a problem rather than build anything from scratch but, even so, this still delivers on acquisition of knowledge of the AI ecosystem.
Emerging knowledge and expertise within the partner ecosystem around GenAI means there are now increasing options where a company scales if building an internal AI Tactical Discovery team is not feasible. These partners can help get a proof of concept (POC) up and running that can be used by your teams to expose them to knowledge of the technology.
Of course, the prerequisite of creating the AI Tactical Discovery team is picking the right people. It may surprise many to hear that some of your best engineers may be skeptical about AI and have little or low interest in joining an effort around the simulation of human intelligence processes by machines. In this scenario, it is important to lean on your team’s interest and pick the engineers who are actually interested in this space. If you are a SaaS company, or have a SaaS arm, ensure you have an SRE, Cloud Architect, Data Scientist and someone from SecOps involved. These are the folks likely to keep the enterprise risk front and center as they work.
You’ll also need a mix of senior and junior folks in the team. What you want to avoid is a senior long-term architect, slowing or blocking the innovation, because “that’s not how we do it here”. This is an opportunity for the organization to gain new insights and skills and the hypothesis you are trying to prove through the effort of this AI Tactical Discovery team, is what is possible, not starting out with ideas about why things aren’t possible.
Once your team is compiled, the next stage is how best it can go about its business, which can be broken down into nine steps:
1). Identify a use-case candidate - Once identified, the Tactical Discovery team should develop a LLM or AI powered solution to that problem. The goal is to learn about AI and how models can be used in innovative ways with the assumption that usage and training will become increasingly cost effective over the next few years.
2). Stick to the task at hand - While the company should be spending time in marketing to tune and alter the company vision to highlight any/all existing AI supporting capabilities, the Tactical Discovery team should not include a marketing resource. The goal is organizational learning not specifically product discovery.
3). Resist the product manager (PM) temptation - At this stage, it’s wise not to involve a PM. This team is fuelled by the start-up, innovation culture where small failures are celebrated as opportunities to learn something new, iterate, and improve. They do not need to be burdened by the processes that come with a PM.
**
4). Keep ‘em separated - The AI Tactical Discovery team should be explicitly separate from any work in your SecOps team, who should be evaluating any and all AI capabilities arising in the tools the company is already using. However, the project targeting should be aware of the pipeline of AI capabilities being looked at so they don’t overlap. They should however ensure they are operating within the guidance and guardrails of SecOps to mitigate any potential data security issues.**
5). Embrace agility & expect change - Change and innovation happen through execution. This is not to say that strategies and roadmaps aren't important. Every organization needs a North Star for AI adoption. However, expect to pivot and change as the market evolves. Aim for multiple, connected wins as you make the journey towards the North Star, rather than one grand reveal.
6). But put some parameters around - Set a time limit of no more than three months to completion to encourage outcomes and set a benchmark for measurement against objectives.
7). Cut any admin burden - The Tactical Discovery team needs to be empowered to spend what they need and get access to new tech not currently at the company. Reducing the administrative burden as they progress so the rest of the business can stay informed and aligned is a good idea, depending on your scale, resources and culture.
8). Get clarity to avoid the anti-patterns of AI value - Your goal is to understand how this tech can work so that when/if you go to market with an AI powered capability, it is fit-for-purpose. You want to avoid the anti-pattern of just sticking a ChatGPT powered chatbot on or around your product and claiming to be AI powered. This false value will wear thin as the hype fades. Find real value for AI in your company and gain organizational knowledge, skills and muscle memory with AI close to where you currently add value.
9). Apply caution - GenAI is early in its adoption cycle but already the path is littered with the scare stories of how it can go wrong. Keep your company and user privacy, regulatory and compliance needs front and center as you learn.
The impact of taking such a radical path may be the only way for enterprises to keep pace with AI. This effort won’t be the silver bullet for your organization to “get AI”, but it’s a shot you have to take. The AI tools and capabilities available now are likely not to be the ones that you’ll use 18 months from now.
Outcomes will come to the fore, and you need your organization to have some knowledge about what outcomes can and can’t be achieved with current AI, and continue to flex that learning muscle at Tactical Discovery scale, as AI tech advances. And it will.
