Building AI Features with AI Squad

Lightful, a tech-for-good company that helps nonprofits develop digital skills, recognized the need to systematically explore generative AI integration rather than approach it ad-hoc. The key challenge was understanding how to build useful AI features when working with unpredictable API responses, unlike traditional integrations where all possible outcomes are known. Additionally, they needed to navigate rapidly evolving AI technologies, establish best practices in uncharted territory, and overcome steep learning curves.

The company assembled a cross-functional "AI Squad" with dedicated designers, engineers, digital coaches, and product experts, securing senior buy-in through a company OKR focused on completing an AI pilot. They took a problem-first approach, mapping charity challenges using opportunity solution trees before exploring AI solutions. The team added "prompt design" as a new step in their development cycle and used collaborative techniques including user story mapping, continuous prompt testing by all team members, daily cross-functional feedback loops, and human-centered design principles.

The squad successfully launched an AI-powered social media feedback feature integrated with their existing platform, using GPT-4 to provide three specific improvement suggestions with educational insights. This feature maintains human-in-the-loop collaboration, helping nonprofits stay authentic while learning best practices.

Key takeaways for product managers include: start with user problems rather than AI capabilities, establish dedicated cross-functional teams for AI exploration, build collective mental models through continuous testing, focus on AI as collaborative enhancement rather than replacement, and adapt traditional agile practices to accommodate the unpredictable nature of generative AI responses.