Look beyond
the AI claim.
Buy-side AI and technology due diligence for private equity, venture capital, investment banks, and enterprise M&A teams.
Understand what you are investing in
AI investment due diligence tests the technology behind an investment claim. RPA2AI supports buy-side evaluation of technology platforms, software, and data assets for investors and enterprise deal teams.
The aim is to understand what the technology can deliver, what makes it distinctive, and what could put its value at risk.
What diligence should examine
Capability: what actually works?
Ask what the technology does, where it performs well, and which limitations remain. Examine whether the evaluation is meaningful for the intended use.
Differentiation: what is defensible?
Consider ownership and access to data and models, alongside the team’s ability to develop the product. Clarify what depends on external technology.
Architecture: what happens at scale?
Explore technical debt, scalability, and the effort required to support additional uses. A working demonstration leaves important operational questions unanswered.
Risk: what could change the thesis?
Identify material uncertainties for further investigation, including technology dependencies and risks associated with intended use. Align technical findings with the broader transaction review.
AI through an investor’s lens
Examine what an AI business actually owns, whether its team can improve the technology, and how model performance compares with alternatives. Kashyap’s session on AI investment due diligence connects these questions to readiness, scalability, and the risks an investor needs to price in.
Buy-side focus. RPA2AI’s published policy is not to accept software-vendor mandates for this due diligence service.
Start with the transaction question
Share the type of transaction, the technology area, the decision timetable, and the questions the deal team needs to resolve. Confidential material can follow after an appropriate process has been agreed.
For procurement rather than a transaction, see technology strategy and selection. For finance-team learning, see professional education.
Questions about AI investment due diligence.
What is AI due diligence?
AI due diligence examines the technology and data behind an investment claim. RPA2AI’s published offering covers buy-side assessment of technology, software and data assets.
The scope should reflect the investment decision: what is being acquired, what could undermine its usefulness and what evidence is available. Data, development practices and actual operation deserve examination together. Kashyap’s AI-audit article describes a structured assessment that can also support mergers and acquisitions. Read more about the scope of an AI audit.
How can investors identify AI washing?
Look beyond the AI label. Establish what the product actually does and what evidence supports its claims. Kashyap has highlighted exaggerated AI marketing as an investor and customer concern.
Require an explanation of the system that can be tested. Assess task performance, reliability and operational constraints rather than accepting a broad label as proof of capability. Kashyap’s model-evaluation article shows why a product’s enterprise usefulness requires evidence beyond impressive benchmark results. Read more about testing enterprise AI claims.
Does a strong model establish a strong AI business?
No. Technical quality needs to translate into a usable product and a viable business. Examine benchmarking, differentiation and the ability to deploy the technology.
The work needed around the model can determine whether it becomes a dependable product. Data access, application connections and deployment arrangements all need sustained support. Kashyap’s enterprise technology-stack guide makes these dependencies visible and helps distinguish a working component from a complete solution. Read more about the systems behind enterprise AI.
Why do data and model rights matter in an AI investment?
A business may depend on assets it does not fully control. Establish ownership and usage rights for data and models, and assess dependencies that could affect the investment.
Information dependencies deserve attention even where the model itself is supplied by a third party. Establish whether the product can access the data it needs without exposing information to unauthorized users. Kashyap’s co-authored privacy article explains why data access and governance must be examined together. Read more about privacy and AI.
What changes in the economics of an agentic AI business?
One request can trigger many paid actions. Assess the cost of completing a customer task, including retries and supervision, rather than relying on token prices alone.
Test how the cost changes when tasks become longer or less predictable. Repeated calls and failed steps can make an apparently affordable application expensive to deliver. Kashyap’s cost-optimization guidance helps investors ask how the product controls those behaviors rather than relying on a favorable demonstration. Read more about agentic AI cost controls.
What is your next AI decision?
Talk to us about your business priorities, an investment, or your team’s learning needs.