4. AI Due Diligence - Supporting PE investors to assess what good AI looks like

 AI DD - for PE Investments:

PE investors are increasingly looking to understand how strong, reliable, and scalable the company’s AI ecosystem actually is. Hence, alongside a traditional IT & Tech/Software DD, PE investors are keen to add an AI DD as a separate scope to fully understand AI nuances, risks & associated one-off & yearly costs.

The question now is not just 'does the AI work?'

The real question is - 'does it work reliably, fairly, legally, and can the team prove it?

AI DD - Suggested Scope:

Best practices suggest an AI DD should have the following scope covering key AI elements & integrations woven into the overall IT & Tech landscape

• AI Strategy: assess how AI supports the business and creates value. Is AI actually helping generate revenue, keep customers, or reduce costs?
• Data Governance: assess if data is used safely, properly, and without bias, analysis of how data is implemented and where the training data comes from
• Model Development: assess how AI models are built and trained, is AI just a basic layer built on top of someone else’s API, or does the company understand every AI model being used with clear documentations, AI framework being used; LLM, RAG, Agentic AI, AI Agents
• Model Monitoring (Model Training, Evaluation & Testing): assess how the AI is trained and tested, and how its performance is measured (like accuracy, testing methods, latency, etc.)
• MLOps maturity assessment: asses if the models can be updated automatically, fix issues quickly, and monitor for unusual behaviour, assess how do they manage deployment, automation, and smooth running of AI systems
• Security: assess how the company protect AI systems and data from risks and attacks, do they follow rules, check risks in LLMs, and make sure data and models are properly owned and licensed
• Explainability & Compliance: Makes AI decisions understandable and ensures rules are followed
• Team Capability: checks if the right people and skills are in place to manage AI, assesses role coverage (data engineers, ML engineers, MLOps, annotation), documentation culture, and key-person dependencies
  

Typical AI DD Assessment:

An example of an AI assessment at a high level is shown below. This is typically more nuanced, with a lot more complexity encountered while completing the assessment.

AI Due Diligence Risk Heatmap & Scoring Model

Example of AI DD conducted by CIDEK for an AI driven SaaS Platform

We conducted AI assessment for Company X, an AI driven SaaS platform providing job market insights, aggregating job postings and LinkedIn profiles to match labour supply and job demand insights.       

In our AI assessment of the platform, it appeared that the usage of programmatic AI in the platform was minimal, needing significant work to make this a more fully fledged AI driven platform. There was little documentation to support the company's explanation of the end to end process of data sourcing, data storage, data processing, data analytics, and data retrieval. AI models were not trained, resulting in poor predictability. AI frameworks were very basic just to identify natural language processing. 

Overall investment readiness was scored 53/100, highlighting significant risks needing investment and remediations post PE investment. We provided an AI improvement roadmap and associated costs, which was successfully implemented to improve the overall AI maturity. 

Example of AI assessment for a PE backed PortCo

Several PE backed PortCos are in process of AI implementation, not fully knowing what to implement or not. Many a times, AI implementation is starting without any preparation or an AI strategy, mainly as the PortCos are in pressure to show some AI investments. A PortCo justified its AI investment in an ITSM platform's AI featured license, as they had communicated to employees AI driven chatbots to resolve IT issues faster. While the license cost increased multifold for the AI driven ITSM platform, users didn't receive the full benefit of a truly AI driven strategic ecosystem. 

CIDEK's AI DD & AI Implementation capability 

Alongside traditional IT DD and Tech/Software DD, CIDEK has know-how of what good looks like for an AI DD, both for AI driven SaaS platforms, and to assess AI maturity of a PE backed PortCo. Furthermore, CIDEK has supported AI implementation to improve the overall AI maturity of PE PortCos. 

CIDEK is available to support and keen to discuss any AI related programs for PE investors and PE backed PortCos.