From Hype to Reality: 5 Critical AI Implementation Lessons (Learned from RPA)

Navigating the landscape of selling AI involves a blend of excitement and complexity. Given that generative AI may undergo similar phases as the RPA hype cycle, here are some reflections distilled into 5 strategies for early AI providers. These insights may prove valuable for Tech Enthusiasts and Product Managers navigating the AI terrain. 🧠

1. Prioritize Skill Development Early: At the peak of the hype cycle and slightly beyond, the target persona you’re selling to requires efficient upskilling to manage their expectations effectively. Whether through unpaid presales efforts or paid workshops, it’s crucial to acknowledge that technological understanding will not be at the same level as with early adopters. Prepare them gently 🎓 for potential challenges where their cooperation is indispensable.

2. Demystify the Jargon: Ensure clarity around AI buzzwords within among stakeholders and within your team. Understanding terms like ‘hallucination,’ ‘models,’ and ‘temperature’ is essential for alignment. Just as we’ve had to clarify misconceptions around the term ‘robot’ 🤖 in RPA, it’s important to avoid cementing misunderstandings that could persist for years to come.

3. Showcase Transparent Demonstrations: Demonstrate the true capabilities and limitations of AI through live demos. Witnessing these capabilities firsthand sets realistic expectations and lays the groundwork for success. Implementing quick prototypes, such as a demo featuring a visual effect e.g. using a HTML slider for the ‘temperature’, can be particularly effective, as seen in early RPA demos with their “attended” clickpaths. This will lay the groundwork for the charging model of the run costs (topic of a future post).

4. Implement Rigorous Monitoring from Day One: Continuous oversight is imperative. Unlike in the early days of RPA, where reporting requests often arose during UAT, transparency from the outset is key. It is imperative that requestors have a sufficient visibility into the decisions taken by e.g. an LLM. Failure to deliver on this (even if unasked for) will stall the adoption of AI and will stall your project right before RtP!

5. Bridge the Expertise Gap: Early stages of technology adoption often reveal a significant disparity between the provider’s expertise and the buyer’s understanding. Misaligned expectations regarding the provider’s expertise can lead to unrealistic demands. Do not let your expert status trap you into potentially (legal?) demands that you (or the tech) cannot fulfill. 🤝

AI adoption entails more than just integrating new technology — it’s about aligning requirements and capabilities. What strategies have you found indispensable in your AI adoption journey? Perhaps you have experiences to share from other technologies and their early hype cycle phases? Feel free to share your thoughts and experiences below. hashtag#AIAdoption hashtag#Strategy hashtag#RPA hashtag#TechTrends hashtag#LessonsLearned