Our position
The capability of foundation models continues to rise while the cost of using them falls. Strong models are available from several providers and in open-weight form. For most companies, access to a capable model is no longer a source of advantage.
We believe the durable AI companies of this cycle will look less like model companies and more like the best vertical software businesses: deeply embedded in one customer's workflow, holding proprietary data, and paid for the work they complete. These are the companies we want to back.
Three shifts we are underwriting
From assisting to completing work
The first wave of AI products helped people do their jobs faster. The next wave completes defined units of work end to end, such as processing a claim, reconciling an account or preparing a filing, with a person reviewing exceptions. That changes who the buyer is, how the product is priced, and how much value it captures.
From seats to outcomes
When software completes the work, charging per user makes less sense. We expect more AI companies to price per task, per outcome or as a share of value created. Companies that can measure the outcome they deliver will be able to price for it.
From demos to reliability
Many AI products impress in a demo and struggle in production. Customers in regulated industries need predictable behaviour, audit trails and clear accountability. Evaluation, monitoring and controls are becoming a core part of the product, and a category of infrastructure in their own right.
Where we see opportunity
- Vertical AI for regulated work. Financial services, insurance, legal, healthcare administration and government, where work is document-heavy and accuracy matters.
- Agents with real permissions. Systems that take actions inside customers' software, with the identity, access control and audit layers that make that safe.
- Evaluation and reliability. Tools that measure how AI systems behave, catch failures before customers do, and prove compliance.
- Data infrastructure for AI. Pipelines and platforms that turn a company's scattered data into something AI systems can use safely.
How we evaluate defensibility
We ask four questions of every applied AI company:
- Does the product improve with use? Each customer interaction should make the system better in ways a competitor cannot easily copy.
- How deep is the integration? Products that sit inside the system of record and change how work is done are harder to replace than a separate interface.
- Can the company survive a better model? If the next model release makes the product unnecessary, the business is a feature.
- Is accuracy measured in production? We want to see how the system performs on real customer work, not only on benchmarks.
What we are cautious about
- Thin interfaces on top of a model with no proprietary data or workflow ownership.
- Products whose main advantage is a prompt or a fine-tune that others can reproduce.
- Gross margins that depend on model prices staying where they are today.
Important information. This publication is provided for general information only and reflects the views of Cash Capital as of its date. It is not investment, legal, tax or accounting advice, and it is not an offer to sell, or a solicitation of an offer to buy, any security or any interest in any fund.
Figures, statistics and company information are drawn from third-party sources that we believe to be reliable, including news reports, data providers and company announcements. We have not independently verified them, and some may be estimates, preliminary, reported rather than confirmed, or later revised. Cash Capital makes no representation or warranty as to their accuracy, completeness or timeliness and accepts no liability for any loss arising from reliance on them. Readers should consult the original sources and their own advisers before making any decision.
Predictions, outlooks and other forward-looking statements are opinions about uncertain future events and may not occur. We do not undertake to update this publication. Companies named are used for illustration only; unless stated otherwise they are not Cash Capital portfolio companies, and naming them does not imply any relationship, endorsement or recommendation. See our Terms & Disclosures.