top of page
  • Writer's pictureAdam Votava

AI Due Diligence: evaluating substance over hype in investment opportunities

Artificial intelligence (AI) is becoming a cornerstone of business innovation. It's no surprise that the majority of investment memorandums now mention AI in one form or another. As an investor, you need to ask yourself – is it just marketing, or is it something worth exploring?

 

The Importance of Data Quality

Having data is great, but having analytics-ready data is better. This means the data should be anonymised, standardised, centralised, well-described, and free from biases. For example, in a recent engagement with a healthcare company, we found millions of clinical data points scattered across various operational databases. While a valuable asset, transforming this data into a format suitable for analytics will require substantial effort.

 

Takeaway: Ensure your data is analytics-ready to maximise its potential.

 

Data Usage Terms & Conditions (T&Cs)

Understanding the terms and conditions of data usage is crucial. Can the data be used beyond its original purpose? Can it be leveraged for data monetisation or algorithm training? And under what conditions? An example from a recent project illustrates this perfectly: a company with unusually comprehensive data usage T&Cs has the rights to use any anonymised data and owns all anonymised data, derived insights, and analytics models. This legal coverage provides them with significant flexibility and potential for innovation.

 

Takeaway: Clear data usage T&Cs are essential for flexibility and innovation.

 

Robustness and Scalability of AI & ML Models

Evaluating the robustness, scalability, proprietary nature, explainability, and flexibility of AI and ML models is vital. The best companies integrate AI into their business and operational models, creating a competitive advantage. One notable example is a company that tied together 15 machine learning and analytics models to form an end-to-end robust process. This comprehensive approach enabled them to smartly collect data across the customer journey and apply analytics to achieve market-leading performance. This contrasts with many companies that claim to leverage AI but are merely experimenting with off-the-shelf solutions.

 

Takeaway: Robust, scalable AI models integrated into your business can create significant competitive advantages.

 

Machine Learning Operations (MLOps)

Effective MLOps practices are essential to ensure AI models are maintained, updated, and improved over time. Deploying and operating AI & ML solutions differs significantly from traditional software. For instance, in a due diligence engagement with a UK company, we found that the MLOps were completely lacking, and the company used an ad-hoc trained ML model in production without any data or model monitoring.

 

Takeaway: Implement effective MLOps to ensure continuous improvement and reliability of AI models.

 

Automation in AI Development and Operations

The extent to which AI development, deployment, and operation rely on human intervention should be minimal. Manual processes can become problematic when scaling rapidly, whereas automation drives efficiency and reduces operational costs. For instance, in a due diligence engagement with a fintech scale-up, we found that while their ML development and operations processes were robust, the lack of automation was identified as a potential bottleneck for rapid geographical expansion.

 

Takeaway: Prioritise automation in AI processes to ensure scalability and efficiency.

 

Conclusion

As AI continues to evolve, distinguishing between genuine innovation and marketing hype is critical. The importance of robust AI due diligence in private equity is growing. The landscape is changing, and so must investors’ strategies. Are you prepared to navigate this new terrain?

 

Five questions all investors should ask themselves during due diligence:

  1. How much does the target’s underlying data ecosystem play into value creation, today and in the future?

  2. Are we investing in data due diligence proportionately to the perceived value?

  3. Do we have the experience —or know where to get it— to truly understand the data nuances in this particular space and to develop unique insights?

  4. Is our data diligence integrated with the broader commercial and financial due diligence effort, so the insights and recommended actions are consistent with where the value lies?

  5. Are these insights flowing directly into the value-creation plan to jump-start delivery on the investment thesis post-acquisition?


Our clients appreciate the findings of our AI due diligence, which have helped them make better-informed investment decisions. At DataDiligence, we have delivered dozens of AI and data due diligence engagements. If you want to hear more about our approach to AI and data due diligence, reach out to us today. Let’s increase your confidence in truly understanding AI-related risks and opportunities. Contact us now to start the conversation.

Comments


bottom of page