# Technical DD Framework A technical due diligence for deep tech companies answers one question: is the claimed technology real, and can it scale? **Evidence hierarchy** (strongest to weakest): 1. Production system running at a paying customer with measurable outcomes 2. Pilot system with before/after data and controlled baselines 3. Internal demo with synthetic data 4. Published academic paper by the team 5. Slide deck claims with no supporting evidence **The narrative-evidence gap.** Every startup has a vision architecture. The DD maps what's actually running vs. what's planned. The widest gaps between narrative and evidence are usually the most important capabilities. If the claimed key differentiator is the least evidenced component, that's the reddest flag. **Baseline attribution.** When a company claims "X% improvement," always ask: compared to what? Manual spreadsheet planning? Previous vendor? Doing nothing? A 6% improvement over spreadsheets might be genuine algorithmic value, or it might just be the value of automation. Distinguish AI value from automation value. **Patent vs. product.** A filed patent means the team believes something is novel. A conceptualized patent means they've thought about it. A "proprietary" anything with no patent filed means it's unprotectable know-how. Unfiled patents on core claimed capabilities are a major gap at any stage past seed. **What must be true at scale.** List the 3-5 conditions that must all hold for the business to work at venture scale. If any single condition fails, the company hits a structural ceiling. This is the investment framework: each condition is a de-risking milestone, and funding should be staged against them. Related: [[Technical Moat Assessment Framework]], [[IP Strategy for Deep Tech Startups]], [[Key-Person Risk in Deep Tech]], [[Industrial AI MOC]], [[Investing/Investing Principles]] --- Tags: #deeptech #investing #systems #firstprinciple