Build is a prominent example of the "agentic services" trend, where AI agents are deployed to handle end-to-end professional workflows rather than just providing chat interfaces. Their platform is built on LangGraph, making them a key player in the practical application of multi-agent orchestration. They demonstrate how agentic systems can be used to navigate complex, unstructured data environments like municipal records and geospatial databases.
In the broader ecosystem, Build represents the shift toward vertical AI agents that incorporate human-in-the-loop verification for high-stakes industries. They are active in the agent application layer, specifically championing the use of agents for document-heavy industries like commercial real estate. Their success or failure will likely serve as a signal for the viability of the "outcome-based" pricing model for AI agents in professional services.
Commercial real estate development is a business defined by friction. For decades, the process of identifying sites and navigating due diligence has relied on a fragmented network of consultants, analysts, and lawyers. Build is a company attempting to replace these manual workflows with a multi-agent AI system they call Dougie. They describe themselves as an agentic development firm. This distinction reflects a shift in how vertical software is sold. Instead of providing a tool for human analysts to use, Build sells the outcomes those analysts typically produce.
The company was founded by James Stirrat-Ellis and Ben McClusky. Stirrat-Ellis brings a background in architecture and design from Harvard and Heatherwick Studio, while McClusky provides the technical foundation in multi-agent reinforcement learning. This combination allows the company to approach the "built world" with an understanding of both the physical constraints of development and the computational complexity of automating it. Based in New York, San Francisco, and London, the firm operates at the intersection of high-end real estate and agentic research.
At the center of their offering is Dougie. Technically, Dougie is a multi-agent system built on LangGraph, a framework designed for creating stateful, multi-actor applications. This technical choice is significant. Most language model applications struggle with the non-linear nature of real estate data, which spans municipal portals, geospatial records, and utility transcripts. By using an agentic framework, Build can trigger specific sub-tasks that mirror how a human team works. One agent might handle data retrieval from a zoning board while another performs financial modeling.
The user experience is intentionally low-friction. Clients interact with the system primarily through email. A developer can email a request to Dougie to start a site selection study or a due diligence report. The system then processes hundreds of data sources simultaneously to generate a professional-grade document. Crucially, the company maintains a human-in-the-loop model. Internal domain experts review and verify every deliverable before it reaches the client. This human oversight is a pragmatic acknowledgement that in institutional real estate, raw AI output is rarely sufficient.
Build's business model departs from industry norms. Traditional firms charge by the hour or through retainers, creating a misaligned incentive where slower work generates more revenue. Build uses fixed, deliverable-based pricing. This approach treats real estate intelligence as a product rather than a service. It is a bet that speed is the ultimate competitive advantage for developers. Delivering in days what usually takes weeks allows clients to move faster on high-value opportunities.
The company serves institutional clients such as Tishman Speyer and Stack Infrastructure. These are firms operating at a scale where incremental improvements in speed result in significant saved interest or early-mover advantages on prime land. By focusing on asset classes like data centers and logistics, Build is targeting the most capital-intensive sectors of the modern economy. They are not building another dashboard; they are building an automated pipeline for the physical world.
A multi-agent AI system built on LangGraph that automates commercial real estate development workflows.
Build is hiring