Scaling Release Communications with AI
How GPT Assistants and Notion templates transformed release communications, enabling clearer customer-facing messaging, improving feedback loops, and fostering cross-team knowledge sharing.
Context
The majority of this team’s portfolio was the backend engine that calculated and continuously optimized for the best possible route for pickups and dropoffs. We had an understaffed product team that caused bottlenecks, so we empowered engineers to take ownership alongside PMs from discovery through delivery, including presenting demos and being involved in the process of developing release communications. This was a brilliant technical team that faced challenges with communicating their under-the-hood optimizations and upgrades, especially when no front-end changes were made—resulting in customer confusion about how our product worked.
Discovery
I worked with the engineers to break down the challenges they faced in release messaging, creating a psychologically safe environment for them to admit what made them nervous and where they felt they needed more support. The common themes I noticed were that, despite having deep product knowledge, the gaps lay in storytelling, framing, and confidence in crafting messaging that conveyed value in customer-facing language.
Delivery and Process
At this particular company, the output for customer-facing release communications was called a Product Explainer - to share the value of new features with our customers. I created Product Explainers that met our criteria for tone with simple, clear and concise communications. These explainers became my dataset for a GPT Assistant, allowing engineers to easily input context about the newest features, with the output copied into a Notion template. This solution simplified their process for creating release messaging and sharing it in a templated manner. With each feature near release, I paired with the engineer to see how we could improve the tools with their real-time usage feedback.
Transformation
The Product Explainer GPT Assistant and Notion Template transformed this team’s ability and confidence to translate their technical updates into customer-facing language. This greatly improved our customers’ awareness of backend changes, allowing them to provide meaningful feedback and enabling greater iterative product development.
Retrospective
In hindsight, I wish I could have experimented with ChatGPT’s Memory using these explainers to support “just-in-time learning”. Once engineers or product managers refined the outputs, they could update the Memory with the latest product changes, feeding them back into the GPT Assistant. This could change the narrative that “only PMs are the subject matter experts of how a feature works”, and instead, any team in the organization can easily learn about a feature in their own time. For example, on a sales engineering call, team members would be empowered to share more details immediately instead of booking another call with a PM or sending a follow-up email. This would require review from Product Operations to ensure clean and accurate data in Memory, but I believe the cross-pollination of knowledge could have a significant impact and free up Product team bandwidth.