Choose your own AI adventure, Brian Piper | EvolveDigital Toronto 2026
Summary
Brian Piper presented a comprehensive live demonstration of multi-tool AI workflows for higher education marketing, using Evolve University as a fictional case study. The session focused on practical implementation rather than theoretical concepts, showing how professionals can chain together different AI tools to solve real-world challenges. Piper began by establishing the premise that most AI demos show single tools in isolation, but effective work requires collaboration between platforms.
The audience chose to work on pitching an AI committee to leadership rather than launching a new program. The demonstration started with ChatGPT's custom Problems GPT for initial brainstorming about organizational barriers and strategic considerations. Piper emphasized the importance of custom GPTs, stating his rule: 'If I use a prompt more than three times, I'm creating a custom GPT out of it.
' The workflow then moved to Google Gemini for deep research, leveraging its ability to search hundreds of websites and provide cited sources. This research was fed into NotebookLM, which Piper described as 'specifically designed to hallucinate less' because it only returns information based on provided documents. NotebookLM's standout feature was its ability to generate interactive podcasts about the research, where users can call in and ask questions in real-time.
The session demonstrated persona creation using Christopher Penn's RACE model (Role, Action, Context, Execute) to build a Chief Information Officer persona. This persona was converted into a custom GPT and used to anticipate objections and challenges to the AI committee proposal. Piper stressed the 'human in the loop sandwich' concept, explaining that effective AI work requires human expertise at the beginning to provide context, and human validation at the end to check for accuracy and bias.
The final stage involved using Claude to create PowerPoint presentations, with Piper noting that while ChatGPT excels at organization and structure, Claude produces better long-form writing and integrates well with desktop applications. He demonstrated how the entire research and persona work could be synthesized into a presentation outline, then converted to slides that align with brand guidelines. Throughout the presentation, Piper emphasized practical tips like using WhisperFlow for voice-to-text input, keeping custom GPT instructions outside the AI platforms for portability, and the importance of collaborative validation when working with AI outputs.
Key Takeaways
- •Create custom GPTs for any prompt used more than three times to improve efficiency and consistency
- •Use the 'human in the loop sandwich' approach - provide human expertise at the beginning and validation at the end of AI workflows
- •Combine multiple AI tools strategically: ChatGPT for structure and brainstorming, Gemini for research, Claude for long-form writing
- •Store custom GPT instructions outside AI platforms in spreadsheets or documents to enable easy migration between tools
- •NotebookLM can generate interactive podcasts from research documents where users can call in and ask questions
- •Use the RACE model (Role, Action, Context, Execute) to create more effective AI prompts and reduce generic outputs
- •Test AI-generated personas against real audience data through surveys and feedback to validate accuracy
- •Deep research in AI tools should be validated by checking sources and cross-referencing multiple platforms
- •Voice-to-text tools like WhisperFlow improve AI interaction by condensing natural speech into clear prompts
- •AI workflows should incorporate institutional data like CRM, admissions data, and user surveys for more relevant outputs
- •Collaborative AI work with multiple team members provides better validation than individual use
- •Canvas view in ChatGPT enables easier editing of long-form content without regenerating entire outputs
Transcript
Hi everyone. Um, so I have the pleasure of introducing our next speaker. Uh, a few people have been at the intersection of content strategy and technology as long as our next speaker. Uh, Brian Kyper has been optimizing digital content for 30 years. He also co-authored the second edition of Epic Content Marketing. He hosts a AI for you podcast and now helps organizations build AI workflows but actually work for real people not just in theory and please give a warm welcome to Brian. >> Thank you....
