Members-Only
Recent Talks & Demos are for members only
You must be an AI Tinkerers active member to view these talks and demos.
Claude Agent: Positioning Clarity Score
This talk details an AI agent built in Claude Projects that quantifies positioning clarity across consistency, focus, and alignment using engineered prompts and psychology research.
I built an AI agent (Claude Project) that solves a problem every technical
company faces: positioning clarity degrades over time without anyone noticing.
THE PROBLEM:
Year 1: Your pitch deck, website, and sales materials are aligned
Year 2: Different team members write different content
Year 3: Your materials contradict each other—investors can’t explain what you do
This isn’t a strategy problem. It’s a communications architecture problem.
THE SOLUTION:
A Claude Project configured to quantify clarity across three dimensions:
- Consistency: Do materials emphasize the same core concepts?
- Focus: Is messaging concentrated or fragmented?
- Alignment: Do materials support the same strategic positioning?
Result: Concept Clarity Score (0-100) + specific divergence points
THE DEMO:
I’ll analyze volunteer company materials live using the Claude Project:
- Upload their website/deck content to the project
- Show the agent generate Concept Clarity Score in real-time
- Reveal top 3-5 divergence points with specific examples from their content
- Walk through the system prompt engineering that makes this work
- Explain the research foundation (100+ psychology studies on memory,
comprehension, consistency)
THE TECHNICAL DEPTH:
- System prompt design for multi-dimensional analysis
- Knowledge base integration (research papers, scoring frameworks)
- Prompt engineering for quantitative scoring vs. qualitative feedback
- How to get consistent, measurable outputs from LLMs
- Building reliable agentic workflows in Claude Projects
WHAT YOU’LL LEARN:
- Advanced system prompt engineering for complex analysis tasks
- Using LLMs for quantitative measurement (not just text generation)
- Applying psychology research through AI agent configuration
- Creating reproducible frameworks from subjective qualities
- Real consulting use case (I’m using this with actual clients)
This is a working tool solving real problems, not a prototype. I’ll show
you the actual system instructions, knowledge base structure, and live
workflow.
Adding a chat interface (https://github.com/SEMalytics/claude_project_chat) to help simplify the demo process.