 The blog series "Crossing the AI Chasm" aims to identify in practical form, critical actions organizations must take in order to fully and successfully integrate GenAI into their day-to-day operations.
Episode three in the series focuses on finding "the specific" application to choose, and also who should be responsible to implement it in your organization.
Topics:
Selecting the GenAI application(s)
We've already covered in the series two fundamental pre-requisites of injecting GenAI into an organization. You must understand the different forms of AI, and how to link them to the strategic processes of the organization. The next step in the process is selecting the variant of GenAI (the specific application) that works best for your organization.
Staying away, and beyond recommending a "brand", here are elements to consider while selecting your GenAI application(s).
Consider these factors when making an App selection:
Desired Deliverables and Outcomes - If you are going to make the investment in GenAI applications, make sure that you will get deliverables in the form that you need them, and into the applications where you will use them. Otherwise you may have really cool single purpose technology that will create more work for your team to do.
Feasibility and Impact - Be aware of tradeoffs you'll have to make when deploying GenAI technology. Some will be positive, some less so. If the GenAI tool you plan on implementing forces your team to be trained beyond their job description, you may be creating tension and workload that is unintended. You don't want or need your business development manager to become an expert in prompts; you need them to be experts at selling, that happen to have tools that use GenAI.
Scalability and Integration - Closely related to #1, make sure that your tools will be able to support your planned growth and the volume of work they are expected to do over the next two to three years. It is one thing to prove that you can convert a video to text using GenAI. It is completely another having to process 1000 videos a week. As well, make sure that the integration workflows you establish are not linked 1:1, application to application. Avoid having your implementation be torn down because you integrated the applications in a way that can't scale.
Future-Proofing - Choose tools that can grow with your business and can be replaced. Yes - replaced. Don't integrate or select a tool that will force such extreme measures that it becomes the immediate critical path and single point of failure. Avoid unique applications that work with one vendor or format only.
Compatibility - Ensure the AI tools can integrate with your existing systems.
Vendor Support: - Consider the level of support and updates provided by the tool vendors. I'm looking at you "open source". Consider using applications that have documentation and customer service, such that they can help you when technology throws a wrench in your day... because it will.
Selecting the implementation/integration party
Even if you make a perfect match between your organizational needs and one or more GenAI applications to support your business mission and goals; all may be in vane, if the implementation is left to chance or to "the smart person in the team who knows how to use ChatGPT".
Many initiatives fail in the implementation phase because the internal or external, person or team in charge of implementing the initiative (let's call them SMEs for now) "know about" the technology, but fail to see how it should be integrated to the business goals and beyond it's technical stack.
Consider these factors when selecting who will integrate the GenAI applications to your organization. These apply to both internal and external implementing parties.
Expertise - Ensure your SMEs have more than working knowledge about the technology and how it supports your business needs. You can have a PhD in AI fail the implementation because they are not a SME in your business mission and goals. Look for SMEs that can adapt to the changes of your organizational plans as they may change even while you are implementing a small project.
Experience - A track record of successful implementations is indicative of a reliable partner. Do yourself a favor too, and understand that some experience must come from outside of the world of GenAI, because it is an emerging field. Similar experience in adjacent technology fields is OK. The key here is that you need not train anyone at the risk of the mission. Remember, experience is that thing you gain the hard way, and immediately after you needed it the most. Hire people that have it already.
Cultural/Â Mission Fit - Your internal or external stakeholder SMES should align with your company's values and work culture. Recall that beyond incorporating the technology to the business processes and technology stack, the SMEs will also have to get the workforce to drive how GenAI should be used, and will set the tone on how it will be adopted.
Clear Objectives - Define what success looks like for both parties; from implementation through deployment and workforce training. If you don't, you will never know if you accomplished the mission. These should also help to make sure you don't let the idea fairy give you "technology indigestion". Start small, hit milestone after milestone and then expand from there, always aligning your objective to your business goals and strategic plans.
Roles, Responsibilities and Authority - Establish clear expectations for each stakeholder and make sure that the SMEs have the authority go get things done.
Backbone and brain - Do yourself a favor and make sure that your SME has the backbone to say when things are not going well, and the brain to recover and get back on track.
Bonus - Resist the urge to do this "organically" from within the organization. You can't drive the car looking in the rearview mirror, or being distracted. No one says you can't have "a smart employee who knows ChatGPT" implement GenAI in house... but know you are asking for a heartbreak and flirting with a failed program. It's not about the cool features, but how the technology is immersed into your business ecosystem.
Select AI applications that will deliver value and can be integrated in a way where no one in your team has the slightest clue that they are using AI. If the addition of AI is imperceptible in its form, but obvious and measurable in its return, you will know you have chosen wisely.