TE
Travis Eric
HomeNow
Work with meBook a 30-minute call
Travis Eric LogoTravis Eric

AI consulting and custom software, based in Fort Collins. Explore the work. Bring your next idea.

Travis Eric AI Consulting(720) 770-2979

Navigate

  • Home
  • Now
  • Projects
  • How I work

Deeper

  • Work with me
  • About
  • Share your experience
  • Case Studies
  • Resources
  • Impossible Vision
  • Technical Stack
  • Manifesto
  • Writings

Follow the work

New systems and field notes, sent when they ship.

Communities

AI Builder's LabView membershipTeneo Publishing CollectiveView details
@TravisEric_Travis Eric@TravisericLive Builds@travis_eric
PrivacyTerms© 2026 Travis Eric
/
/
Back to writings
Skills and playbooks

Conversation bench

Travis EricNovember 4, 20262 min read
Share this post:

The conversation-bench feature is an intriguing part of our autonomous business system, designed to facilitate meaningful dialogues based on voice recordings or memos. This tool focuses on extracting core insights from the founder's ideas, as outlined in the C:\code\.claude\skills\conversation-bench\SKILL.md.

In our overnight build system, the conversation-bench is triggered with phrases like "ask me questions about this recording" or "talk to me about this memo." It aims to delve deeper into the founder's thoughts by asking questions that only he can answer, such as big picture concepts, values, and personal stories. The system avoids burdening him with micro-decisions, which are instead handled by the agent and listed under "My calls" for potential override.

The core mechanism relies on what the document refers to as the "gold path." This algorithm ranks questions to prioritize significant decisions and meaningful inquiries. If a thread becomes too deflective or gets a "You decide" response, it is deprioritized. Conversely, ideas that receive high ratings, like a 9 or 10, are temporarily set aside to allow them to mature.

The user interface integrates design elements from FortListOS/src/lib/design-tokens/fortlist-brand.v1.json. This includes brass actions, ink text, and a combination of Archivo Black and Inter fonts. The system is visually designed to maintain focus and clarity, with features like a brass glow on the active card and a subtle nudge to encourage engagement.

For practical deployment, the server script is run through PowerShell using the command:

node $env:USERPROFILE\.claude\skills\conversation-bench\scripts\server.mjs "<recording folder>" --port 4520

This setup ensures that the founder can access the tool via http://localhost:4520/, with the option to open a browser session upon request. This approach allows for a seamless, focused interaction without unnecessary distractions or interruptions.

Join Pattern Insights

Get weekly insights on AI architecture, pattern recognition, and building platforms without permission.

Tags:
build-in-public

Read next

November 3, 2026

target-scout — aim premium models at the right work

In the overnight build system, I stumbled upon a useful tool called target-scout. Its purpose is clear: to direct premium models like Fable to the most...

Read it

November 2, 2026

UI Automation Harness (Track-A watchdog scaffold)

The UI Automation Harness, specifically the Track-A watchdog scaffold, offers a practical approach to maintaining site integrity through an overnight build...

Read it

November 1, 2026

UE5 GamesOS MCP

Last night, while running the UE5 GamesOS build with MCP, I noticed a few things that could be crucial for anyone in the thick of Unreal Engine 5 production....

Read it

Join the Discussion

Have thoughts on this post? I'd love to hear them! Join the conversation on X where we can discuss AI architecture, pattern recognition, and building platforms.

Discuss on X

Or reach out directly at @TravisEric_