The role of AI in transforming software investing

The latest wave of AI innovation has brought a familiar combination of technological progress and market recalibration.

Frontier model announcements, including Anthropic’s latest product release, continue to reinforce the pace at which AI capabilities are advancing. At the same time, public market valuations for listed software companies have come under pressure, reflecting investor concern about how these advances may reshape competitive dynamics and long-term value capture in the sector.

From a software-focused private equity perspective, these developments are not contradictory. Periods of rapid technological progress have historically been accompanied by uncertainty around where, and how quickly, economic value ultimately accrues.

Across multiple software cycles, breakthroughs in capability have rarely translated directly into durable advantage without disciplined execution, organisational readiness and a clear understanding of where value is created and protected. Valuation pressure has also been unevenly distributed: more pronounced in horizontal application software than in systems software, and more acute in North America than in Europe. Markets are beginning to differentiate between software categories based on perceived AI exposure, even if the frameworks for making those distinctions remain early-stage.

The question, therefore, is not whether AI will matter. That has been decisively answered. The more relevant question for long-term investors is how to act in this environment. In our experience, lasting value is created where new technology is absorbed into systems, workflows and decision-making processes in a way that compounds over time.

What follows reflects how we think about AI and its implications for our investment activity, grounded in our experience of building, scaling and owning software businesses.

As autonomous coding agents compress the marginal cost of bespoke software development, the binding constraints shift. Domain expertise to define what should be built becomes more valuable.The ability to verify the correctness, reliability and compliance of outputs becomes more important. The scarcity moves from code production to judgment and accountability.

In practice, this produces a lumpy and uneven adoption curve. Early deployments may generate compelling demonstrations yet limited economic impact. More substantial returns often emerge only once AI is embedded within core systems and operating routines. The speed at which AI capabilities improve should not be confused with the speed at which value is realised.

Attempting to deploy AI broadly and simultaneously across an organisation can dilute focus and overwhelm teams, particularly in mid-market software businesses with finite management attention.

In our experience, selective sequencing – concentrating effort on a limited number of high-impact areas – is more likely to produce durable outcomes than blanket adoption strategies.

For investors, this distinction between capability and absorption matters in two ways: in selecting which companies to back, and in determining how to create value alongside management teams during the ownership period. Companies that can absorb AI in a disciplined, system-level way are better positioned to convert technological change into compounding advantage.

Where AI creates value – and where it can destroy it

If organisational absorption is the binding constraint, AI’s impact on software markets will necessarily be uneven.

The same technologies that dramatically increase productivity or enhance the value proposition of some software companies can undermine the economic logic of others. Understanding this divergence is central to assessing both opportunity and risk.

The impact of AI spans a wide spectrum. At one end are businesses where AI can be absorbed into existing systems and workflows, acting as an accelerator of established value creation mechanisms. At the other end are products whose functionality may be replicated or subsumed by more general-purpose AI systems, leaving limited scope for durable differentiation. Between these poles lies one of the most fertile environments for software innovation and investment in recent years.

New applications are becoming economically viable. Existing platforms can extend their relevance and scope. The boundaries of what software can automate and orchestrate are shifting rapidly. Distinguishing between these cases requires attention to several dimensions.

Software products that primarily organise, search or summarise information are more exposed to displacement as AI models improve in reasoning, synthesis and domain adaptation. By contrast, products that combine data capture, workflow orchestration and action tend to be more resilient, because value is created not only through insight but through execution within a defined system.

AI-assisted development reduces the cost and time required to build and iterate on software. This accelerates innovation but can compress differentiation where products compete mainly on feature breadth.

As the ability to generate bespoke code for user-defined workflows becomes abundant, the constraint shifts from writing software to shipping trusted systems, and then to changing how work is performed within organisations.

Faster analysis to accountable action

AI reduces the latency and cost of analysis, increasing managerial leverage and enabling earlier intervention. However, this benefit is realised only where processes, incentives and accountability structures are sufficiently clear to translate information into action.

AI can therefore act as both a disruptive force and a powerful amplifier of existing strengths. 

Where it is absorbed into broader systems of record, workflows and outcome-oriented use cases, it tends to reinforce and compound established advantages. Where it cannot be embedded at the system level, it risks bypassing incumbent software altogether.

Durability and value protection in an AI-rich environment

Market reactions to advances in AI reflect concern that new capabilities may erode the position of incumbent software providers or compress long-term returns. In our experience, both risks and opportunities are real but unevenly distributed. 

Applying the absorption framework above, below are four structural characteristics that can help to protect and enhance the value of software businesses as AI becomes more powerful and more widely available.

Enduring principles in a shifting cycle

The evaluation of these characteristics has gained urgency in the past 24 months, but they are not new to long-term software investors. They are enduring attributes of healthy, capital-efficient mid-market B2B software companies. Such businesses can thrive amid an accelerating pace of innovation, and in some cases, despite the presence of larger competitors with substantial R&D budgets.

The core principles remain consistent: clear economic relevance, strong execution layers and the ability to absorb new technology in a way that compounds over time.

Viewed through this lens, the role of the long-term software investor is to remain deliberate and selective, attentive to technological and strategic developments, while grounded in durable convictions about how value in software is created, protected and compounded.

Disclaimer: This presentation is provided for information purposes only and is not intended to be relied upon as the basis for an investment decision. The contents herein are not to be construed as legal, business, or tax advice, and each individual should consult their own attorney, business advisor, and tax advisor as to legal, business, and tax advice. No representation or warranty, express or implied, is given as to the accuracy or completeness of material contained herein on the part of Bregal Milestone. Any investment is subject to various risks, all of which should be carefully considered by prospective investors before they make any investment decision.