As the AI gold rush continues, we’re taking a measured approach to identifying genuine opportunities in the space. While headlines focus on breakthrough models and massive funding rounds, the reality for most investors – particularly smaller funds – requires a more nuanced strategy. Here’s our current view of the AI investment landscape and where we see the most promising opportunities.
The Infrastructure Challenge
Despite the optimistic dreaming about two founders in a garage building the next great enterprise AI platform, the reality is that infrastructure plays require massive capital and established distribution networks. However, we do see selective opportunities in the infrastructure layer, particularly in data preparation and management. The unsexy but critical work of making enterprise data “AI-ready” through cleansing, aggregation, and standardization remains a significant challenge that smaller, focused teams can tackle effectively.
The Model Layer Evolution
The quality of off-the-shelf (OTS) models has reached such a high level that the traditional approach of building models from scratch is rarely the best path forward. Instead, we’re seeing two distinct opportunities emerge. First, there’s significant value in developing sophisticated ensemble approaches and adversarial frameworks that effectively put guardrails around existing OTS models, making them enterprise-ready and safer to deploy. Second, we’re seeing companies with access to massive proprietary datasets – particularly in healthcare and industrial sectors – build specialized models that address use cases no general-purpose LLM could handle. These datasets, often representing decades of industry-specific knowledge and experiences, enable the creation of highly specialized AI solutions that would be impossible to replicate using public training data alone.
The Application Layer: Where the Action Is
The most compelling opportunities we’re seeing are at the application layer, where several key themes are emerging:
Workflow Integration
Major opportunity lies in redesigning workflows to seamlessly integrate AI capabilities with human decision-making. The end goal is never to have AI spit out “x report,” but rather for AI to know when to automate and when to engage to make decisions. This requires deep domain expertise and a nuanced understanding of when to automate versus when to engage human judgment.
Expanding Operational Scope
We’re seeing companies leverage AI to tackle previously unfeasible tasks. For instance, in our own industry of investing, we’re seeing work being done around deal screening that could surface investment opportunities we’d typically miss due to bandwidth constraints. This isn’t about replacing our core functions, but extending our reach into areas that were previously cost-prohibitive to do fully.
Fundamental Business Process Reimagining
The most exciting opportunities come from completely rethinking traditional business processes through an AI-first lens. This process of “rethinking” versus “modifying” allows almost a clean slate to build what needs to be built from ground zero. Take CRM systems: instead of bolting AI onto existing platforms, innovative companies are reimagining sales processes with infinite recall, automated research, and intelligent reporting built into their core. While established players like Salesforce could theoretically build these capabilities, the innovator’s dilemma suggests they’ll struggle to move quickly enough.
Analog-to-Digital Transformation
Some of the most promising opportunities exist in traditionally resistant sectors like construction, where conventional digital solutions have failed to gain traction. By leveraging computer vision and voice interfaces, new solutions can digitize processes without disrupting existing workflows – think of the drywall installer who can’t stop to type but could benefit from real-time voice-based project management. The biggest opportunities in AI are still rooted in focusing on the small, pesky problems real people face day in and day out.
Optimizing the Mundane
We see significant value in addressing routine business processes and these are where we focus our efforts. A 10% improvement in core operations often delivers more value than a 200% improvement in edge cases. Simple innovations like adaptive UIs and automated data processing can deliver outsized returns by focusing on the 95% of business activities that don’t make headlines.
As the AI landscape continues to evolve, we remain focused on finding applications that deliver immediate value while positioning companies for long-term success. The winners in this space won’t necessarily be the ones with the most advanced technology, but those who best understand how to apply AI to solve real business problems effectively and sustainably.