How you should think about AI

Hammers in search of nails

Amongst the seemingly never-ending onslaught of artificial intelligence hyperbole and catastrophisation, a lot of companies, perhaps yours, are beginning to drift away from the first product development principle of finding customer problems to solve and then developing solutions to them. Many are now entirely transfixed on developing solutions and then finding customer problems to solve with them.

The challenge is that AI in and of itself possesses no inherent value. Its value can only be derived through its application to specific problems and opportunities.

I thought this LinkedIn post from Allie K Miller articulated this brilliantly:

Oil is a raw commodity; the actual use case is the application of that commodity, i.e. the transportation provided by cars powered by oil. Electricity is a utility; its value is unlocked when used to enable an activity or solve a problem, i.e. improving productivity by allowing people to read after dark.  Cotton is a raw material; its value chain extends to the production of goods that directly impact the market and play a role in the life of consumers, i.e., school uniforms.

Artificial intelligence is a technology; it's not the technology that's transformative — it's the problems it solves for your customers and the benefits it creates for them.

You may very well develop new ventures, products and services powered by AI in the future, but the technology, as we've seen, can't be the starting point.

How we should be thinking about AI

I'd posit that as innovators, our time, money and headspace would be better spent figuring out how we can use AI to understand customer problems and de-risk the development of solutions to them more effectively and efficiently.

Corporations have for years bemoaned that they struggle to look outside the organisation for new growth opportunities, that talking to customers is too time-consuming, that they don't have the skills to prototype effectively and that running in-market experimentation to validate ideas before investing in them is too hard. More recently, the teams I've spoken to have lamented that their budgets don't enable them to place a large enough number of bets on potential new propositions and that by being unable to hedge against volatility with volume - they're not producing results.

But now, an emerging wave of AI tools allows you to smash through these blockers.

What you could use AI for

At Future Foundry, we've spent the last six months testing and assessing hundreds of tools to help power the Velocity™ method we've built that helps clients accelerate, de-risk and systematise the development of breakthrough propositions. There are four areas in which we think these tools, when used as a co-pilot to our teams, are incredibly powerful:

‍Horizon Scanning‍

Historically, getting a comprehensive understanding of all the opportunities a client could explore involved manually synthesising thousands of customer, competitor, technology and employee data points into digestible 'How might we...' statements. Now, we're using Scholarcy to read and summarise the research, saving countless hours of distillation and freeing up more time to look at other valuable inputs.

Customer Recruitment

We onboard around 100 customers to interview, test and iterate with throughout the Proposition Development phase of Velocity™. Profiling those customers, screening them and creating context-specific interview scripts involves a lot of people and a lot of hours. Now, we use QoQo to help us build better personas, flag the right customers to work with and create the right interview flows for them, giving us more time to speak to more people.

Prototype Experiments

We've used no-code and low-code tools for a while to help build, test and iterate on digital prototypes quickly, and we've used product boxes, 3D printing and explainer films to demonstrate physical products to customers. Tools like Biteable mean we don't need to spend days creating explainer films; Midjourney helps us to create physical product mock-ups in minutes; Uizard helps us to develop interfaces rapidly; and Galileo helps us to create digital mockups using text-based prompts, which leaves more budget to take those prototypes to a higher level of fidelity.

In-Market Experimentation

Creating experiments designed to test our assumptions about a new venture, product or service proposition and then running those tests in a live environment involves thousands of permutations. Tools like Kraftful, Ask Viable, Adwords AI and Mixpanel mean we can run millions of multi-variate customer acquisition, product and business model experiments simultaneously and double down on the winners even faster.

What you (probably) shouldn't use AI for

We've learned over the last six months of these tests that these tools are not a panacea. Building something new requires non-linear thinking grounded in breakthrough insights, which, by definition, can't be inferred from publicly available knowledge.

‍Opportunity Assessment

‍Assessing the potential of new growth opportunities is profoundly nuanced. It calls for a deeper understanding of cultural subtleties, market sentiments and sometimes unspoken behaviours that only humans can instinctively grasp. The grey areas of market dynamics and human intuition are not a place for AI to play in... yet.

‍Roadmap Prioritisation‍

With its reliance on existing data, there is an inherent unpredictability in innovation that AI is not equipped to handle. Deciding which opportunity areas to pursue next is as much an art as a science, involving stakeholder perspectives, long-term vision alignment, and, sometimes, a gamble on future trends that the tools can't simulate.

‍Customer Development‍

Building a fundamental understanding of customer problems will always be about connection, empathy, and trust. The qualitative aspects of customer development, especially understanding personal stories, individual contexts and their idiosyncracies, require emotional rather than artificial intelligence, which the tools we've trialled have failed to emulate.

‍MVP Development‍

None of the tools we've tested match the human capabilities required for designing business models, discerning assumptions from facts, creating experiments to test assumptions, and building pilots to validate or invalidate those hypotheses. There will be no autonomous development of MVPs, at least not in the near future.

What do you need to do next?

1. Understand the inefficiencies and ineffectiveness in your current working methods and test some tools with dummy data to see which could fix the most painful part of the process before building out from there.
2. Spend time with information and data security teams to ensure the tools you want to use don't compromise the safety of company or customer data and explore setting up a safe sandbox environment for testing on real-world projects.
3. Educate your teams to use these tools effectively and ask the right questions of them. AI won't replace them; critical thinking will always be required to turn an idea into a viable business model. But their jobs will fundamentally change over the coming months and years as AI becomes a powerful prosthetic that extends their existing capabilities in previously unimaginable ways.

‍For more insights like these, join The Fold, Future Foundry's free but highly vetted membership community for corporate innovation and strategy professionals.