Top 5 Tips for a Successful AI Engagement

3. Be Able to Hear “Bad News”

Sometimes during the engagement, the project will hit a snag with data that is unexpected – perhaps the amount and quality of client data is not sufficient to build the solution we want. It’s never fun delivering this type of news. And, sometimes we don’t “see” the problem until we are deep into the data of the organization. But as your trusted partner, we will move mountains to solve the root problem, get you back on track, and get you to where you need to be. And what we need from you, our client, is hearing us, being ok with bad news, and working with us to put our heads together to find solutions quickly.

4. Assemble a Great Internal Team

Who you assign to this project – and especially who is managing your service provider relationship – is one of the most important decisions you will make. From an organizational perspective, you need someone to serve as Project Manager, but this does not necessarily need to be a full-time role. It should be someone who is strong at communicating internally and externally, coordinating schedules, and is collaborative about addressing any issues that may arise. 

Another key person for this team is someone who really knows your technology architecture, where data lives in it, and how best to access it. This technical person may be a liaison to a larger technical team that is at the appropriate level to be able to make decisions, but also able to bring decisions back to the larger team, to make sure we are in line with their expectations. The third “must have” player on this team is a subject matter expert; they may not be very technical, but they are deeply embedded in the business of the organization: they understand the business model, business processes, and all the domain terminology and underpinnings to work closely with us and guide us as we dig deep into the data.

Finally, this team should include the sponsor of the project. This is someone who is on (or just below) the executive team. This person is an evangelist for the project, communicating regularly on the trials and tribulations of the effort. Part of this role is to over-communicate and educate the executive team on what projects like this really take to succeed. There should be real-time communication with the service provider (via Slack or Microsoft Teams for example), but also plugging in updates on this project to regular internal standing meetings is recommended. Of course, much of this depends on the internal culture of the company, but ideally, this individual should want to bring the rest of the company along in this journey, in a balanced way.

5. Understand AI is a Science Experiment

At the end of the day, it is important to remind yourself and each other that these types of engagements are science experiments. So much is “unknown” at the outset – but if goals are achieved, the results can have huge potential impact. This is where the “art and science” of where business strategy and artificial intelligence come into play – and much of it involves ensuring that the science “experiment” itself is set up correctly in the first place: the more the client can share examples of their data earlier, the more we can stress-test to ensure the viability, and the higher the chance that we can prove their hypothesis is correct. 

The Scientific method is defined as “an empirical method of acquiring knowledge that has characterized the development of science for centuries.” It involves observation, applying rigorous skepticism about what is observed. It also involves formulating hypotheses based on such observations, experimental and measurement-based testing of deductions drawn from the hypotheses, and refinement of the hypotheses based on the experimental findings. 

Internally, the AI project should be positioned as something that you will learn from now, and for the future. Sometimes “success” is knowing that this type of approach will not work with your organization’s data, and you need to find another approach. The learning process itself – specifically, whether we are able to prove the hypothesis or not – is enormously valuable, as your team will gain the experience in AI to take advantage of it in the future.

Conclusion

If all of these “tips for success” could be condensed into just one word it would be communication. The relationship between a service provider and client is built on trust, and that, as they say, is a two-way street. Frequently we try to look at as much data as possible during the sales process to mitigate the stress of any potential surprises down the road but this is not always possible. The “Proof of Concept” engagements that begin with data strategy frequently help set us on the right course so expectations are clear on both sides. It’s not always a “sure thing” but we walk into the engagement feeling strong about our understanding of the state of the organization and its data. 

From a technical perspective, before we begin any project, we share a “delivery checklist” with our clients, which includes specific questions about the following: business problem to solve, POC vs production model and audience, training data, live data, computing environment, acceptance criteria, and timelines. One critical issue that frequently surfaces in this questionnaire is that the training data available is not representative of the data the client is actually going to use in a live system. Addressing this – and fixing this early on – will save a lot of time and pain. So, as you are building your all-star internal team, make sure you have the right people thinking about your company data – even as early as during the vendor selection process!

We wish you the best of luck in your engagement, and hopefully, you have found these tips for success helpful as you begin your journey.

Photo by Fabian Gieske on Unsplash