AI-Native Investment Screen
Test necessity, user value, defensibility, durability, and adaptability
- Difficulty
- Advanced
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 97%
The screen begins with necessity: remove AI conceptually and ask whether the company can still deliver essentially the same product. If it can, the technology may be decorative rather than foundational. Next, evaluate the real value created for end users instead of rewarding an impressive demonstration or isolated feature. For infrastructure, examine whether the technology can support many future applications as platforms evolve. For vertical businesses, inspect access to customer and operational data, the company's ability to build unique technology on that data, and whether usage creates a compounding flywheel. Finally, assess durability under changing models, architectures, regulations, and geopolitics. Strong candidates combine indispensable technology, genuine customer outcomes, founder insight, accumulating defensibility, and enough organizational flexibility to survive a decade of change.
Origin
Extracted from Masters in Business through Songyee Yoon's explanations of how Principal Venture Partners evaluates AI-native infrastructure and application companies.
Core principles
- 01AI must be indispensable to the business rather than decorative
- 02End-user value matters more than impressive demonstrations
- 03Foundational infrastructure should serve multiple future applications
- 04Proprietary data access can compound into a defensible moat
- 05Long-duration investments must withstand technical and policy change
How to run it
- 1
Run the removal test
Imagine removing AI from the product and operating model. Determine whether the same value could be delivered with conventional software or analytics.
Pro tip Ask directly why AI is indispensable rather than merely useful.
Watch out Agent terminology does not prove that an application needs an agent.
- 2
Verify end-user value
Identify the real problem solved and the measurable improvement experienced by the customer. Separate a compelling demonstration from a sustainable business outcome.
Pro tip Ask what becomes more efficient, effective, accessible, or newly possible.
Watch out Do not equate technical novelty with customer demand.
- 3
Inspect native design
Assess whether the technology shapes the company's workflow, organizational design, staffing, and product architecture. Examine whether the founders understand both its capabilities and limitations.
Pro tip Compare the organization with a company built on the prior technology stack.
Watch out Adding an AI feature to an unchanged business does not make it AI-native.
- 4
Evaluate platform breadth
For infrastructure companies, determine whether the foundational technology can serve multiple applications as models and platforms evolve. Favor capabilities that remain useful across changing product categories.
Pro tip Map several credible downstream uses rather than relying on one current trend.
Watch out A single fashionable use case can disappear as architectures change.
- 5
Trace the data flywheel
For vertical applications, identify lawful and appropriate access to customer or operational data. Verify that continued use improves unique technology and strengthens the product's competitive position.
Pro tip Describe explicitly how each unit of use produces better data, decisions, or performance.
Watch out Data volume alone is not a moat if competitors can access equivalent information.
- 6
Pressure-test durability
Model significant changes in technical architecture, policy, litigation, and geopolitics. Determine whether the company can adapt while preserving customer value and defensibility.
Pro tip Evaluate investments against a decade-long holding period.
Watch out Do not treat today's dominant models or rules as permanent.
In the wild
Yoon highlighted Together and Cartesia as companies led by strong founders building foundational technology intended for use across many future AI platform companies. Their appeal was not limited to a single application available at the time of investment.
→ The investments aligned with a thesis favoring infrastructure that can remain useful as the platform evolves.
Yoon's gaming experience taught her the importance of individual features in shaping user experience. She believed Sesame's founders understood which aspects of voice technology mattered and had concentrated their capabilities on advancing those aspects.
→ Founder insight and focused technical execution supported the decision to invest.
A hypothetical AI-native logistics company begins with authorized operational data from customers. Each deployment improves its ability to predict delays and recommend interventions, while its workflow is designed around those predictions rather than adding them to a conventional dashboard.
→ Customer usage improves the product and compounds into a defensible operational data advantage.
Common mistakes
Backing an AI demonstration
An impressive technical feature may have no durable end-user value, repeatable demand, or viable business model.
Mistaking analytics for AI necessity
Some purported agent applications could solve the same problem with conventional data analytics, making their AI positioning superficial.
Assuming today's stack is permanent
Models, architectures, and regulations can change substantially during a venture investment's life, so rigidity creates long-term risk.
Is it for you?
Best for
It is best for evaluating early-stage AI infrastructure companies and vertical applications during rapid technological change.
Not ideal for
It is not ideal as a standalone valuation method or a substitute for financial, legal, and market due diligence.
From the transcript
“can you do the same thing without AI? Like why do you need it? Why is it dispensable?”
“I always try to ask what is the real value that is going to bring to the end users”
“the ones who have appropriate access to data, the data, understanding customers and consumers and the business and build a technology, unique technology on top…”
From the episode
Investing for the AI Shift: Masters in Business with Songyee Yoon
Songyee Yoon