Enterprise productivity has stalled despite decades of software spend. Knowledge workers spend most of their time gathering context, switching tools, and performing repetitive tasks rather than producing outcomes.
§01
The Problem
- Data is fragmented. Critical context is scattered across email, docs, tickets, CRMs, and data warehouses that don't share a common record.
- Tasks are manual. Routine workflows, reporting, reconciliation, routing, follow-up, still depend on humans moving data between systems.
- Insight is delayed. Decisions are made on stale dashboards because synthesis is slow and human-gated.
- Horizontal tools underperform. Generic copilots summarize but rarely execute the specific, high-value workflows of a function or industry.
§02
Why Now
- Agents can take actions, not just answer. Tool use lets systems update records, trigger workflows, and complete tasks end to end.
- Retrieval over enterprise data is reliable. Modern stacks can ground responses in a company's own systems with citations.
- Evaluation makes deployment safe. Teams can now measure and gate agent behavior before it touches production workflows.
§03
What We're Funding
- Vertical workflow agents. Systems that own a specific function, finance ops, revenue ops, support, end to end rather than assisting around the edges.
- Unified context layers. Infrastructure that assembles a live, permissioned view of enterprise data for agents to act on.
- Action and orchestration. Reliable execution across SaaS systems with human-in-the-loop controls and audit trails.
§04
gAI's Bet
We back founders building productivity systems that do the work, not just describe it, embedded deeply enough in a function or industry to create a durable wedge.
End of thesis · T6