Tinderbox 15AUG26: AI Intern to Chief of Staff — Thinking, Building & Learning with AI & Tinderbox
| Level | Intermediate |
| Published Date | 8/17/26 |
| Revision | 1 |
| Revision Note | |
| Acknowledgments | |
| Type | |
| Tags | AI, Chief of Staff, Claude, Intern, 5CKM, 5Cs of Knowledge Management, Eastgate, Identity Praxis, Inc., Mark Bernstein, Michael Becker, Tinderbox |
| Video Length | 01:56:35 |
| Video URL | https://youtu.be/loO0EtSkQYw |
| Example File | TBX Meetup 15AUG26.tbx (303.9 KB) |
| Chat File | TBX Meetup 15AUG26_Chat.txt (12.3 KB) |
| TBX Version | 11.5 |
| Host | Michael Becker |
In this Tinderbox Meetup, we began with a lively discussion prompted by Andreas Grimm’s previous demonstration on using AI to build a new web-based interface to work with Tinderbox, think, and work with your notes. The conversation ranged from AI hype and skepticism to the practical value of AI as a tool for research, knowledge management, problem solving, and learning. A central insight emerged: rather than asking AI simply to solve problems, use it as a partner to help understand the problem, explore possibilities, and incrementally develop solutions. This reflects the principle of incremental formalization—allowing structure to emerge through interaction rather than prematurely defining the final system.
The second half put those ideas into practice through a demonstration of creating an AI “chief of staff” and connecting that relationship to Tinderbox. We explored the distinctions among conversational AI, collaborative workspaces, and agentic/code environments; how context, memory, instructions, terminology, privacy, and boundaries shape an AI relationship; and how an initially inexperienced “AI intern” can become increasingly useful as it learns the user’s work. The demonstration culminated in having AI construct and populate a Tinderbox knowledge-management document, including structured notes, prototypes, attributes, maps, and timeline views. Importantly, the discussion also cautioned against the collection fallacy: having AI generate mountains of information without personally engaging with, understanding, and metabolizing it into knowledge.
Learning Outcomes
By the end of this session, participants should be able to:
- Distinguish among AI interaction modes — conversational chat, collaborative workspaces, agents, and code/execution environments — and understand where each fits within a knowledge-management workflow.
- Develop an AI “intern” toward a chief-of-staff role by supplying context, responsibilities, terminology, standards, boundaries, and ongoing feedback rather than relying on isolated prompts.
- Apply incremental formalization to AI-assisted knowledge work, beginning with goals and dialogue and allowing useful structures, workflows, and representations to emerge over time.
- Use AI alongside Tinderbox to create and manipulate knowledge structures, including notes, prototypes, attributes, maps, timelines, and other views, while retaining human oversight of the resulting knowledge.
- Recognize the importance of human agency and governance — deciding what AI may access, what it may change, how it should behave, and when the human needs to step back in to interpret and learn from the material rather than merely collecting AI-generated output.