1. Introduction – Generative AI and its Relevance to Private Equity
Since OpenAI released ChatGPT in late 2022, Generative Artificial Intelligence (GenAI) has been the talk of the town and it is easy to see why: it brings capabilities which most humans had never seen from machines before. Unlike previous AI architectures, Generative AI can create new content – enabling greater creativity and broader applications. This content can be in multiple formats, such as text, images, audio files and videos. Tools such as ChatGPT (for text) and DALL-E (for images) have gained significant traction since their launch: in fact, ChatGPT became the fastest-growing consumer software in history, reaching 100 million users only 2 months after its release. GenAI offers immense potential and its advent seems like the start of a new era – similar to the creation of the Internet.
Few articles, however, explore how Private Equity firms can use Generative AI. This article aims to fill this gap.
Thinking about the PE lifecycle (see Figure 1 below), GenAI has the highest potential impact in 2 of the PE lifecycle’s 4 stages: investing in companies and creating value in portfolio companies. These 2 use case categories are the focus of this article.
Figure 1: Impact of Generative AI across the Private Equity Lifecycle
2. Investing in Companies
2.1. Deal Sourcing
In terms of use cases that apply to PE firms themselves, “deal origination and curation is where PE firms are making the biggest investments today, which makes sense given competition for deals,” according to Rich Klee, Director at Palladium Digital, a digital consultancy for PE firms. The traditional deal sourcing process involves manual research and networking, which are time-consuming. Generative AI can help augment and speed up this process in several ways:
Getting up-to-speed quickly on new markets – Thanks to its ability to efficiently summarize large amounts of data and answer questions on topics where information exists, GenAI acts as a skill-leveler. It helps investors get up-to-speed faster regarding new areas of interest.
Identifying potential investment opportunities, including add-on targets – “In our sectors and geographies, we generally know which target companies meet our investment criteria; where GenAI can be very helpful is in identifying add-on opportunities in the private markets”, says Lakshman Charanjiva, Partner and Co-Head of Portfolio Operations at BC Partners, a global investment firm with USD 40+ billion in assets under management. Charanjiva mentions that his firm is building APIs to let Large Language Models (LLMs) analyze unstructured data (e.g., news, industry conferences, media statements), which would otherwise be challenging for associates to sort through quickly.
Automating outreach to company executives – GenAI can also be used to personalize and automate outreach emails to companies of interest, by scouring data sets (including recent company/industry news) and crafting customized messages to maximize response rate. While this may require custom IT work, some deal sourcing tools are starting to integrate similar capabilities. For instance, SourceScrub has released a beta version of “Sourcing GPT”, a tool which combines ChatGPT’s generative capabilities with SourceScrub’s private market data to allow users to automatically draft personalized emails to companies that emerge as potential investment targets. The company claims that these tools enable users to complete the process in a few minutes compared to half an hour previously.
2.2. Due Diligence
Another area where Generative AI can be helpful to PE firms is due diligence. During the due diligence process, investors sift through extensive collections of corporate documents, financial data and market information to assess risks and opportunities.
Here, Generative AI can help PE firms in multiple types of due diligence:
During commercial due diligence, GenAI tools can help analyze large data sets and surface insights. As BC Partners’ Lakshman Charanjiva highlights, GenAI “can be very useful in rapidly searching VDRs [Virtual Data Rooms] and publicly available data via prompts, and in market scans and surfacing competitor information for our commercial diligence”. However, in the current state of technology, tools such as ChatGPT often yield unreliable results when it comes to, for instance, identifying relevant competitors: they should therefore be used discerningly.
During financial due diligence, GenAI tools can be used to analyze target companies’ financials. As Charanjiva mentions, “We can apply [ChatGPT’s] ‘code interpreter’ to use Python to rapidly create data cuts on large data sets we receive from the target (e.g., customer/spend cube) – without our associates needing to learn Python coding”.
In technology due diligence, tools leveraging GenAI can be used to automate code review, assess technical debt and highlight security risks. For instance, GitHub Copilot can help quickly understand and navigate a target’s codebase and documentation. Security tools such as Snyk can also assess a target’s cybersecurity posture, by detecting vulnerabilities across the target’s software supply chain.
In legal due diligence, GenAI tools can be used to automate contract review and highlight related risks. Tools such as Kira Systems and Luminance can help accelerate the due diligence process and uncover insights which might have been missed in a manual review.
3. Creating Value – Improving Portfolio Companies’ Performance and Valuation
In addition to leveraging AI to source and analyze potential investments, PE investors can use AI to create value in their existing portfolio. “At the moment, private equity firms are less focused on developing AI tools for their own operations and more on understanding how a burgeoning AI revolution could impact their portfolio,” said Richard Lichtenstein, a partner at Bain & Co, a consulting firm.
Use cases and their impact vary by industry and firm: they may pertain to horizontal functions (e.g., IT, Finance, Sales & Marketing) and/or to specific industry verticals (e.g., Technology, Manufacturing, Healthcare). We will now explore both types.
3.1. By Horizontal Function
Figure 2 below summarizes major GenAI use cases by horizontal function, as well as their potential impact and feasibility.
Figure 2: Impact and Feasibility of Key Generative AI Use Cases by Horizontal Function
When considering which initiatives to implement first, companies should prioritize high impact / feasibility use cases – such as in software engineering, sales & marketing, customer support and R&D.
Of course, impact and feasibility vary by industry and firm. Therefore, companies need to set up a task force to identify the highest-potential use cases, assess which technologies can best address them, and design pilot projects, testing and rollout plans.
3.2.1. Industry Vertical Example 1: Manufacturing
Use cases also have specific variations by industry. For example, in Manufacturing, Generative AI can be used to improve production operations, enhance product development, optimize supply chain management, and improve customer engagement (see Figure 3 below).
Figure 3: Manufacturing Example – Key Generative AI Use Cases across the Value Chain
3.2.2. Industry Vertical Example 2: Technology
Similarly, in Technology, major areas where Generative AI can be used by tech firms are Development & Engineering, Security, as well as Sales & Customer Support (see Figure 4 below).
Figure 4: Technology Example – Key Generative AI Use Cases across the Value Chain
As we’ve seen, from product development to production operations and customer engagement, Generative AI has the potential to make a significant impact across multiple industries’ value chains.
4. Implementing Generative AI
Having established key use cases and potential impact of Generative AI for PE firms and their portfolio companies, it’s important to understand how this technology can be implemented.
PE firms have two major ways – buy and build – of implementing Generative AI depending on their needs, priorities, capabilities and budget.
Buy: PE firms looking to leverage AI capabilities, while staying focused on their core competencies, can subscribe to AI-powered platforms, such as SourceScrub (for deal sourcing), AlphaSense (for market intelligence) and Kira Systems (for contract analysis). With the technology landscape rapidly evolving and new startups emerging, investors would be wise to continuously monitor which available tools best address high-impact use cases, and trial the most promising ones. Also, tools which PE investors already use today are expected to gradually integrate AI capabilities over time.
Build : PE firms looking for additional customization and/or differentiation can explore developing proprietary GenAI-powered tools and workflows – either in-house or with outside support. Deal sourcing – identifying and assessing investment opportunities – is a prime use case. Many large investors have been building AI and data science capabilities for years; for instance, Blackstone has built a 50+ members data science team since 2016, while European PE firm EQT developed Motherbrain, a data-driven investment platform it uses to identify/source investment targets by sifting through vast amounts of publicly-available data. At the same time, some smaller investment firms have also built tools: for instance, Pilot Growth, a growth equity investor based in San Francisco, developed NavPod, a deal-sourcing and workflow tool that uses data from CB Insights and processes it with IBM’s Watson AI service. Jolt Capital, a France-based investor in deeptech, has built Jolt.Ninja, a proprietary tool used to collect and analyze company and market data. For investors who may not have the in-house team to build such tools, two types of solutions exist: for simple applications like chatbots, companies can leverage “drag and drop” tools (such as Microsoft Copilot Studio, “Use Your Own Data” feature of Azure OpenAI service and Dante AI); or they can work with AI consultancies, such as Tribe AI or AI4SP, which have emerged to help.
Buy and build decisions have their respective benefits and drawbacks, as shown in Figure 5 (below):
Figure 5: GenAI – Benefits and Drawbacks of Building In-House Solutions vs Using Third-Party Tools
5. Risks and Limitations associated with Generative AI
While GenAI presents numerous opportunities, PE firms should also be aware of its current limitations:
Results Inaccuracy and Hallucinations – This refers to GenAI’s ability to produce false information. OpenAI itself recognizes the challenge, as it explains on its blog: “ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers”. Thus, users must always critically assess output from AI models, and check it against reliable sources. This also means that GenAI tools are best suited for “error-tolerant” use cases where creativity is an advantage – for instance, generating draft marketing copy or advertising text/images.
Data Security – This is a major obstacle for adoption in an enterprise context. To address this, ChatGPT, for instance, rolled out an Enterprise solution in August 2023 (in which customer prompts and company data are not used to train OpenAI models); however, its consumer-facing versions (including paid ones) currently do use such data (unless chat history is disabled), which raises concerns about data privacy and security. Moreover, like any information system, GenAI tools have vulnerabilities and can be hacked and/or hijacked – such as in the case of prompt injections (which aim to elicit unintended responses from AI tools). In Private Equity, where confidentiality is key, this could have very significant consequences.
Transparency and Auditability – Due to their probabilistic nature and the complexity of their calculations, AI models can appear like “black boxes”: understanding and auditing how models derived specific outcomes can be very challenging – making it difficult to trust them for high-impact decisions and in regulated contexts.
Bias and Ethical Considerations – Models (like humans) have biases based on their structure and training data, which can lead to sub-optimal results. PE firms need to understand these potential biases and find ways to mitigate their impact (e.g., by prompting tools differently, comparing results between tools, fine-tuning models with curated datasets). Down the line, regulation on AI may also require companies using AI models to demonstrate proof of non-bias: in that process, they will need to show transparency, while not revealing proprietary algorithms.
Integration with Existing Systems and Workflows – For productivity gains to be significant, AI tools need to be integrated into firms’ existing systems and workflows. This can be complex, and require significant time, resources and effort, including workforce training.
6. Conclusion – Looking Ahead
Due to the aforementioned challenges and the technology’s novelty, adoption of Generative AI remains relatively limited in Private Equity and Enterprise contexts, but it is growing fast. GenAI, with its unprecedented capabilities, offers significant potential for productivity gains. As PE firms navigate complex markets, GenAI can help automate tasks and deliver insights – with applications ranging from deal sourcing to due diligence and portfolio value creation.
As such, it is important for companies to stay abreast of developments in this field, while identifying and experimenting with high-potential use cases. Technology will continue to evolve. New models – including powerful open-source models like the ones by Mistral – are being built, and new companies are emerging – for instance focused on applying GenAI to specific use cases. Generative AI itself is only one iteration – albeit a breakthrough one – in the history of Artificial Intelligence, and future architectures may have even more powerful capabilities, such as objective-driven AI, as Yann LeCun, Chief AI Scientist at Meta, has pointed out.
As the landscape and technology continue to evolve, Private Equity investors and their portfolio companies need to continue monitoring the space – building capabilities in-house when it makes sense, and working with trusted partners when needed.
Comments