The Living Gem Philosophy – Iteration & Second Memory
Nothing is Ever Finished
In traditional software or documentation, you build it once, and you are done. That mindset fails in AI.
The landscape of AI (Gemini, Anthropic, OpenAI) moves at breakneck speed. New features—like Google's agentic "Anti-Gravity" solution—are released constantly. Simultaneously, the brands we manage (like Lambent or Lumikha) are evolving. We learn new things about our audience every day.
Our Goal:
To treat Gems not as static tools, but as living documents and a Second Memory that evolves alongside the technology and the brand.
Part 1: The Beta Mindset
Because the platforms change so fast, anything we build today might be obsolete in three months.
- Expect change: We will continually revisit and rework Gems to reflect new AI capabilities.
- Don't aim for permanent perfection: Build for now, knowing you will improve it later.
- Flexibility is key: Our approach to prompting must be flexible enough to pivot when the underlying technology shifts.
Part 2: Gems as a Second Memory
Developing a brand is a journey of discovery. Sometimes we learn from audience feedback, and sometimes we learn from silence (lack of feedback).
If we just keep that knowledge in our heads, it is lost. We use Gems to store this accumulated wisdom.
- The Feedback Loop: When we discover a new keyword that works, or a tone of voice that resonates, we must feed that back into the Gem.
- The Result: The Gem becomes a "Second Memory." It remembers the brand's history and evolution so you don't have to start from scratch every time.
Part 3: The Capture & Refine Workflow
How do we practically manage this evolution? We don't update the Gem every single minute, but we do capture insights constantly.
Step 1: Capture the Insight
When you have an idea or learn something about the brand, capture it immediately. Don't let the thought vanish.
- Tools: Use whatever is fastest for you:
- Capacities
- Google Docs
- Google Keep.
Step 2: The Retrofit
At your leisure, take those captured notes and feed them back into the system.
- Update the First Principles logic.
- Update the Brand-Specific Gem knowledge files.
- Create a New Prompt if the insight reveals a gap in our toolkit.
Summary
- Embrace Impermanence: The video you watch today might be re-recorded in three months. That is normal.
- AI Moves Fast: Platforms like Gemini update constantly (e.g., adding agentic capabilities). Our Gems must update to match them.
- Brands Evolve: As we learn what works for a brand, we must update the Gem.
- Capture Everything: Use notes (Google Keep/Docs) to capture insights, then retro-fit them into the Gems later. This turns the Gem into a powerful "Second Memory."