How innovation workshops help to get smart about gen AI use cases

When it comes to identifying and prioritizing opportunities for business value across the enterprise, gen AI use cases, specifically, are arguably no different than any other technology-related ones. It all comes down to cost benefit, and having a set of priorities.

Advantages of gen AI workshops

In conducting hundreds of innovation workshops around the world for the public and private sectors, many of which have been focused on AI/ML, data, and analytics, I’ve found these design-based brainstorming sessions work well for gen AI as well as any other strategic technology trends.

The workshop methodology typically takes 10 to 25 individuals from across the organization and guides them through a two- to four-hour ideation session to identify and prioritize their ideas. There’s a “Goldilocks” principle here in terms of the right number of participants, the right duration, and the right number of voting criteria, so this has been learned over the course of many experiments and refinements. Of course, there are innumerable ways to run an innovation workshop or brainstorming session, so here’s one in particular that worked well. Whatever approach or format you ultimately decide on, consistency is key.

We start with a kickoff to share the overall goals and objectives of the workshop, then a pep talk from the project sponsor, followed by 30 minutes to capture ideas electronically. During this time, participants are able to enter their own ideas into group decision support software, either in-person, virtually, or in a hybrid setting, and also comment on and build upon each other’s ideas.

We follow this with one to two hours of hearing elevator pitches from each participant,  typically just a minute or two per idea. This is where we’re able to socialize the ideas with the broader group for general understanding, quick discussion for any clarifications, and, later, individual voting.

Finally, we spend 30 minutes letting each participant vote on all ideas, based on what they’ve heard. Then we wrap up by taking a preview of the voting results and outcomes. Then an executive summary and set of formal deliverables are shared a few days after the session.

To make the process consistent and repeatable, I like to use a well-defined and balanced set of voting criteria, which helps place each idea into a cost-benefit matrix, often called a project prioritization matrix. We use two voting criteria for the business benefit side of the equation, which are strategic fit and business value. Similarly, we use two voting criteria for the ease of implementation, or cost, side of the question, which are time and cost to implement and maintain, and project risk and complexity.

Each of these are scored on a 1-3-5 scale and added together to give an overall score in the subsequent cost-benefit chart; business benefit is plotted on the y-axis, and ease of implementation is plotted on the x-axis.

The result of this approach is that after voting, participants can clearly see how their ideas stack up. If they’re tactical opportunities, they’ll show up as quick wins in the top-right quadrant (i.e. high business benefit and easy to implement), and if they’re more strategic opportunities, they’ll show up as must haves in the top-left quadrant (i.e. high business benefit, but more time-consuming, costly, and risky to implement). Both of these types of opportunities are good, however. The quick wins naturally support the near-term implementation roadmap whereas the must haves support the longer-term implementation roadmap.

The benefits for gen AI

One clear benefit of this approach is it’s consensus-based, so with group participation, you’re able to have your own people come up with and share ideas, as well as vote on each other’s ideas. This approach can support both top-down and bottom-up philosophies in terms of capturing ideas from leadership and from end-users by carefully selecting the participants you’d like to attend the session. One of the side-benefits of these types of workshops is you can bring together team members who may not regularly see each other or be in everyday communication to help cross-pollenate ideas and bring in diverse perspectives.

Another is you quickly get a large number of ideas on the table for further investigation and analysis. The prioritization approach in the workshop takes a coarse-grained filter to help identify quick wins and must haves, and you can then conduct further analysis on each idea as needed to come up with the in-depth business case and ROI for each opportunity. Some ideas you can also group together into a broader project, simply park for future investigation, or even discard as needed.

Of course, many gen AI use cases happen organically and outside the scope of a workshop. For these, you can simply have someone note them down as pre-existing ideas and then have them prioritized by the group in the same way so they show up in the cost-benefit matrix. This helps to show the big picture and where these ideas fit in. You may also identify adjacent and similar projects, which can be rolled into the existing project, and vice versa.   

If you find there’s too many gen AI use cases floating around for a single session, you can also break the workshops into discrete sessions by department or division. The same can be done for functional areas such as IT, operations, sales, marketing, legal, finance, HR, and so on. Your chief AI officer can then take the results from each session and see how they support the broader strategy and implementation plan for AI across the enterprise.

While much is different with gen AI, there’s also a lot that’s the same. Using proven and reliable techniques such as innovation workshops and cost-benefit analysis — starting with a coarse-grained filter and then refining into specific business cases — is a great way to get organized and start to prioritize both your tactical and strategic opportunities.