AI is entering a new phase of enterprise transformation. The first wave of adoption focused on copilots—tools that helped humans with drafting, summarizing, or speeding up knowledge work. But the next and more important shift is already happening: the rise of agent-first systems of action.
At BGV, we believe that the most significant outcomes in AI will not come from companies creating better assistants. Instead, they will result from companies developing agents that carry out tasks. This distinction is important. Copilots boost productivity, but agents fundamentally shift who—or what—operates within the workflow. In an agent-first world, humans don’t just get recommendations; they delegate tasks from start to finish, and software becomes the default executor.
From Copilots to Agents: The Next AI Layer
Agent-first systems mark the shift from “AI helps you do the work” to “AI does the work for you.” These agents are not just simple chat interfaces or productivity tools. They are embedded operators that can start actions, coordinate across systems, make decisions within set boundaries, and keep improving through feedback loops.
This shift introduces a new system layer: the system of action. Historically, systems of record stored information, and systems of engagement enhanced user interaction. The next dominant layer will be systems of action—platforms where agents are responsible for execution across high-friction workflows.
The companies owning this layer will shape the future of enterprise software.
Why High-Friction Industries Create the Biggest Opportunities
The most valuable agent-first businesses will arise in industries where workflows are complicated, repetitive, and limited by labor constraints. These are sectors where humans are overwhelmed, processes are disjointed, and execution becomes the main bottleneck.
BGV focuses on industries with four shared characteristics:
- Labor is structurally constrained
Many sectors face persistent labor shortages, rising wage pressure, or talent gaps that cannot be solved through hiring alone. - Workflows are repetitive and operationally dense
These industries involve high volumes of standardized tasks—coordination, scheduling, documentation, compliance, and decision-making. - Agents can become the default execution layer
Once embedded, agents are not optional tools—they become the operating fabric of the business. - Industries are deeply data-intensive, with fragmented information flows
The most fertile ground for agent-first systems also lies in sectors where massive volumes of structured and unstructured data are generated every day, but remain trapped across disconnected systems. In these environments, agents gain a compounding advantage: every action taken produces more proprietary workflow data, improving decision-making, personalization, and automation over time. Data-rich industries enable agents not only to execute tasks, but to continuously learn, adapt, and build durable intelligence moats that strengthen with scale.
Retail, healthcare, and energy management are prime examples.
Retail: Agents as the New Operational Backbone
Retail faces complex operations: managing inventory, forecasting demand, scheduling staff, supporting customers, optimizing prices, and coordinating supply chains. These tasks are repetitive, margin-sensitive, and often handled by separate systems.
Agent-first platforms can independently handle replenishment cycles, improve store operations, and make real-time decisions at scale. The winning company won’t be the one that provides store managers with better dashboards—it will be the one whose agents manage the workflow.
Healthcare: Execution in a Labor-Constrained System
Healthcare faces one of the most severe labor shortages globally. Administrative burdens keep increasing, clinician burnout is widespread, and workflows stay fragmented across payers, providers, and legacy systems.
Agent-first systems can automate documentation, claims processing, patient coordination, and clinical operations. Over time, these agents become deeply integrated into care delivery infrastructure, creating significant distribution and data moats.
Healthcare is not just a vertical—it is a frontier for durable systems of action.
Energy Management: Toward Autonomous Infrastructure
Energy is increasingly decentralized, dynamic, and software-driven. Grid optimization, load balancing, predictive maintenance, and energy storage coordination demand ongoing decision-making and implementation.
Agent-first systems can enhance autonomy across energy infrastructure, promoting more resilient operations and less human intervention. This is where software starts to merge with energy autonomy—agents embedded directly into physical systems.
Durable Moats in the Agentic Economy
BGV invests in system-of-action startups that can develop durable moats in the agentic economy. In this new era, defensibility comes not from features, but from workflow ownership and execution depth.
The strongest companies will compound advantages through:
- Deep workflow embedding and switching costs
- Proprietary execution data generated through agent actions
- Distribution leverage inside operational ecosystems
- Sustainable economic scaling models where automation drives margins
Agents that execute repeatedly inside mission-critical workflows become harder to displace over time.
Sustainable Scaling Models: Economic Leverage, Not Just Technical Scale
A key trait of lasting agent-first winners will be their ability to grow economically, not just technically. In the agentic economy, it’s not enough for systems to perform impressive tasks—businesses need to improve efficiency and increase margins as adoption expands. At BGV, we support companies building sustainable growth models through disciplined architectural choices: using specialized language models (SLMs) tuned for specific workflows, relying on open-source foundations to lessen hyperscaler dependence, and focusing on agent orchestration rather than brute-force inference-heavy compute. The strongest platforms will align pricing with the outcomes they deliver, charging based on the value created by workflows rather than just seat-based licensing. Our main test is simple: as the agent scales to handle more tasks, customers, and environments, do margins get better? The companies that answer yes will shape the durable systems of action for the next decade.
Beyond Workflows: Frontier Execution Layers
We also believe that the most ambitious outcomes go beyond digital workflows and into frontier execution layers—where agents act in the physical world.
This includes:
- Robotics plus embedded AI agents
- Autonomous energy systems
- Industrial automation driven by agentic coordination
- Real-world execution environments where software meets hardware
The future system of action is not confined to the browser. It will extend into factories, hospitals, warehouses, grids, and autonomous infrastructure.
The Agent-First Future
The next generation of industry-leading companies won’t be copilots. Instead, they’ll be agents that manage workflows, make decisions, and become the default action layer in high-friction industries.
At BGV, we invest where AI shifts from assisting humans to operating repetitive work. The agent-first era is here—and the systems of action built today will define the venture-scale outcomes of tomorrow.
