What the AI?!... Triton Partners
The Drawdown
Connecting an exclusive community of operational leaders in the private capital fund industry.
Ubiquitous product launches have AI enthusiasts claiming it will revolutionise work by increasing productivity, removing repetitive processes, reducing human errors and providing detailed data analysis.
Meanwhile, concerns over the implications of AI on ethics, data quality, transparency and accountability are being voiced just as loudly, with regulatory restrictions on the horizon.
But what does all this mean for PE in practice and how can the market adapt to these developments?
In a series of interviews,?The Drawdown?has sought to speak to various industry stakeholders about how they are using AI and how they see the technology shaping the industry in the future.
The following interview is with Triton Partners' head of technology Lyndon Arnold.
The Drawdown?(TDD): What does AI mean to you?
Lyndon Arnold (LA):?I think of AI as any technology that can significantly beat the human or be augmented with the human to optimise results. AI can be used to improve efficiency, productivity and gain deeper insights into data, faster. AI can help automate tasks, optimise processes and enhance the digital experience for investors. In the case of generative AI, this can be used to provide investment professionals with an alternative view.
Triton has been running for more than 25 years and has accumulated a wealth of private data. We are therefore constantly thinking about how this can be used internally. We treat AI as a technology that can accelerate data analysis, deal sourcing and information dissemination.
TDD: How do you want to use AI internally at Triton?
LA:?We have split the use cases into three pillars.
From an investment perspective, we want to improve how we source companies and find add-on opportunities for our portfolio companies. We can also use AI to strengthen due diligence processes and in the creation of deal proposal documentation.
Then, there is portfolio company value creation. We help portfolio companies apply generative AI use cases to improve their business processes or find new business models and alternative revenue streams.
Lastly, we look at how we can use AI within Triton to administer our funds. In our view, document processing, market mapping, managing fundraising and onboarding processes, cybersecurity and personal workflow efficiency are all strategic areas for us to deploy AI in.
TDD: What is your technology strategy – do you prefer to develop it internally or outsource it?
LA:?Our strategy is, again, threefold. Either we use the technology that we've got and make best use of that, or we look to the market to buy existing tools. And when we can't do that, we look to develop that technology internally.
In terms of making use of our current technology, we are developing alongside Microsoft ways to use its version of generative AI and Copilot, which is going to help optimise individual efficiency levels at the firm.
But when it comes to tools that are built for private equity, we look to buy. There are more than a dozen vendors on the market building specific AI large language models designed for our particular use case. We want a tool to help us centralise knowledge from our internal data sources.
Aside from that, we are developing our own in-house administration tool built on the Neptune platform, which has generative AI capabilities. Some possible use cases include the digital onboarding of investors, improving the digitalisation of side letters processes and their operational management.
领英推荐
TDD: How have you found the implementation of Microsoft Copilot so far?
LA:?I think we're at the early stages with this technology. Parts of the Microsoft AI stack have been rolled out to the end users, allowing them to find their own use cases in a private and protected way. I think Microsoft’s Bing Chat Enterprise is a safe and secure way for our end users to test the product. We are still understanding the best way this tool can be used, taking into account any risks.
TDD: You mentioned your in-house tool Neptune. Can you tell me more about it?
LA:?Neptune is a ‘low code no code’ solution. It helps you rapidly deploy applications and digitalise workflows. Neptune happens to be owned by Triton and we like to use what we buy, where possible.
Anything that is currently sitting on spreadsheets and requires a manual email-based response is destined to be systemised with Neptune, enabling us to leverage that data across the organisation. We can ensure that there's a single source of truth because it sits on top of the Triton data platform that we've built.
We’re currently using it for our quarterly investor reporting, side letters and AML process. We're in production on the first half a dozen use cases and we've got a backlog of many dozens more that are going to be deployed onto the platform during the next year or so.
TDD: What do you consider to be the most important risk factor to take into consideration when implementing generative AI into workflows?
LA:?For me, it’s an overreliance on the technology that poses the biggest threat, especially where decision making is concerned. There are many examples of people or institutions excessively relying on AI to the detriment of the operation they're trying to carry out. There always needs to be that extra level of human supervision. Hallucinations, for example, are where it’s particularly dangerous. When AI has got it slightly wrong, it is very difficult to spot.
TDD: How else can AI optimise operational processes in the future?
LA:?If I look to the future, I think there's a place for AI and automation across private capital markets. If you think about how we operate as an industry, there's lots of duplication. We need to solve the problem of data silos, for one.
Once we have consolidated that data, it will give the industry more options to democratise data assets and make use of digital tokens. But we should be cautious of inflated expectations of this technology. I always remind myself that it was probably close to 25 years ago that I first used voice recognition software. I remember thinking that we won't be typing out emails for very much longer. Of course, 25 years on, expectations never quite match reality.
This article is part of an AI series and was originally published in?The Drawdown, a publication?which?provides?vital insight and analysis for operational professionals?in private equity and venture capital.
For more articles from this series, please visit?The Drawdown ?website.
Articles from this series include: