AI Academy 2.0 manifesto

Originally published on the AI Academy Blog on May 03, 2019.

Three years ago, two good friends and I decided to start a project called AI Academy. All of us were just back to Italy from Silicon Valley, where we witnessed the rise of AI companies first-hand. It was clear that AI was about to play a key role in the technological development of every organization, yet we found that many companies outside of the Bay Area were still sleeping on it.

We embraced the mission of speeding up the acquisition of Artificial Intelligence in non-tech organizations. At first, we decided to do that by bringing awareness and running educational workshops for professionals and businesses. This was our manifesto at the time. People that followed our workshops started getting excited and soon asked us to build products for them. We thought that there was a business there and AI Academy became a real company, adding technical consulting on top of educational workshops.

In the past three years, we’ve been blessed with the opportunity of working with companies ranging from pre-prototype startups to multinational corporates with more than $20B in revenues.

We never lost track of our mission. Yet, what we learned made it clear that we needed to change the way we help people and organizations.

Here’s what we learned, and how our learnings shaped our values:

1. Having a lean AI strategy is the key to success.

From our experience, this is the single most important success factor when building innovative AI solutions. What does “lean AI strategy” mean? It means designing an AI roadmap that breaks down projects into small milestones that are carefully crafted based on key risk areas (business, data, technology). This allows organizations to address and solve the most pressing challenges early on and maximizes the chances of success.

Developing innovative products or services isn’t easy, no matter how powerful your technology is. Actually, bringing a technology like AI into the mix adds another layer of complexity (to put that straight: AI isn’t a silver bullet and doesn’t always work). We’ve seen companies kick off the development of a new product worried about the technology feasibility, and forget to do enough research on how it would fit in their offering. We also saw the opposite: startups excited about a market opportunity but risked wasting tons of money on immature technology. With a lean strategy, you can identify (and solve) these problems early on, and have a more informed strategy to reach your goals.

This is how we recently helped a Swedish company in the energy business: they wanted to add a new AI-based functionality to their platform, had proof of its business value, but were unsure of the technology feasibility. We identified a simple proof of concept that was fast and simple to build: it was carried out by a single intern in a month and proved that they could reach the performance they needed.

How this shapes our values: we promote experimentation, we challenge every assumption (ours and yours), we approach AI innovation thinking and acting with a startup mindset, whether you’re a 3-person team or a 30,000-person multinational.

2. Building AI is neither a marathon nor a sprint. It’s a relay.

Some companies see AI as a sprint: they strive for quick wins that look good on press releases but hold little value. Others see it as a marathon, with a 3-year plan and a beautiful long term vision, yet they loose momentum halfway and fail to bring serious results to the table.

We argue that the best AI strategy looks more like a relay. AI isn’t a technology gimmick, it has some serious potential and needs to be well rooted into your organization’s long-term strategy. Yet, you’ll need to see short-term results that speak for themselves. If your AI journey has more than one intermediate step that delivers tangible results, you’ll be able to:

• Keep momentum
• Gain experience and create organizational learning
• Refresh (or increase) executive’s buy-in
• Draw interest in different organizational departments and develop an AI culture
• Simply, improve your organization sooner rather than later

When we started helping a customer working in the customer operations industry, we embraced this approach from the beginning, breaking down their ambitious vision into 4 smaller steps, so that they could get market validation (and cash) in weeks instead of months or years.

How this shapes our values: we are not afraid of taking big bets and chase moonshot projects. We like keeping our feet solidly anchored to the ground though, and design strategies that aim at the moon one significant milestone at a time.

3. Education is still key, but better if tailored and case-based

When we started training non-technical people three years ago there was a lot of confusion about AI. Today, it can get even worse. As the buzz grew, so did the confusion and mismatched expectations around what AI is and what it can/can’t do, and therefore we are still strongly convinced that education is paramount to help organizations reach their goals. Yet, the way we train people today is fundamentally different from three years ago.

We embraced Harvard’s case methodology, and invested time to create comprehensive case studies on companies that achieved outstanding results implementing AI in their business. We found that a case-based method allows us to put our clients in the driving seat, making them think deeply about the opportunities and consequences of what they want to achieve with AI.

How this shapes our values: we believe everyone can (and should) understand AI, from the IT to the marketing department. We value practical teachings, real-life examples and hands-on workshops to get life-long learnings.

4. Technology is not the bottleneck anymore (with a little help from good mentors).

Three years ago, if you wanted to build any sort of Machine Learning project you’d have to face the nightmare of finding skilled people to build it. Today it’s different, for two main reasons:

1. There’s a large offering of MLAAS (Machine Learning As A Service) solutions. This means that many products (or parts of products) can be built fast and cheap using APIs. This may not be a viable long-term strategic decision (the build vs buy strategic decision is not an easy one), but it definitely helps kickstarting AI initiatives with low effort and risk.
2. Opensource resources and online courses like Andrew Ng ‘s dramatically increased the availability of (junior) ML developers. These people may be too unexperienced to lead an AI project, but with the right help of good mentors they can do wonders.

This is what we’ve done with a company that needed to kickstart a simple AI proof of concept. Their most skilled engineers were overwhelmed by the development of their platform, so we assigned the project to an intern that was self-taught in Machine Learning. We reviewed his progress every week and removed roadblocks. It proved to be a winning strategy as the POC was finished in a month and the intern gained valuable experience that will stay with him (and the company) forever.

How this shapes our values: we want young developers or IT teams on their first AI projects to grow and learn. We value working together instead of completely taking over projects, to promote learning and cooperation.

These four pillars can be summed up in a single sentence:

Technology isn’t the key anymore. Knowledge, Strategy and Vision are.

We believe in this so much that we decided to stop building algorithms, and focus on assisting our clients with strategic consulting instead. It’s hard to turn down tech work, especially when you still like technology, but it was necessary to fulfill our mission. As a boutique firm, we can deliver higher value to our clients by focusing on what matters the most: developing their knowledge and strategy.

We still want to make it easy for organizations to build their vision, so we’re building a network of partners and AI experts that can support us and our clients in transforming the vision and strategy we build together into real software. As technical people, we supervise the development of these solutions. Actually, we found that getting rid of the heavy-lifting job of writing code allows us to spend more time staying up to date on the latest AI developments, and we can be even more useful in our role as advisors.

We are excited about this new path, not just because it’s what companies really need, but also because we really enjoy our new role. We look forward to sharing our enthusiasm and moving even faster towards our ultimate goal: making organizations ready for the AI era.

AI Academy is a boutique consulting firm. If you want to exchange some ideas on your AI projects, shoot us an email at info@ai-academy.com.