Sandstone is a New York-based legal AI company building an intelligence layer for in-house legal departments, aiming to usher in what it calls the era of AI-native legal teams. Corporate legal functions sit on a deep well of institutional knowledge — past advice, negotiated positions, playbooks and precedents — that is usually scattered across people, inboxes and documents. Sandstone's premise is that this knowledge can be captured and turned into agentic workflows that automate the high-volume, repetitive work flowing into legal from the rest of the business.
The platform lets in-house teams deploy their own legal agents in under ten minutes, automating intake, triage and routine workflows directly across the systems business teams already use — Slack, Salesforce and email — rather than forcing employees into a separate legal portal. Critically, Sandstone is designed to be continuously learning from every interaction, building dynamic playbooks so that the system's institutional context compounds over time and turnaround times shrink as adoption grows.
Sandstone's founding and core team blends deep legal and engineering experience, including veterans such as Jennifer Poon (NetDocuments, Akin Gump), Bo Xiang (a Paul Hastings GC-turned-engineer) and Devon Willitts (Davis Polk, formerly lead legal engineer at Robin), alongside legal engineers drawn from Microsoft, Google, Amazon and AI research labs. That mix reflects the company's bet that AI-native legal tooling has to be built by people who understand both the law and modern software.
The company launched publicly with a $10 million seed round led by Sequoia Capital, with participation from SV Angel, Kearny Jackson and more than 20 general counsel as angel investors — an unusually GC-heavy cap table that signals strong demand-side validation. Sandstone reported a couple dozen paying customers, a mix of smaller legal departments and a few Fortune 500s, with early users citing dramatic reductions in turnaround time and measurable ROI as context compounds.
Sandstone competes in the increasingly crowded in-house legal AI space, but differentiates on agentic automation embedded in business tools and a learning loop that turns scattered institutional knowledge into live, reusable workflows.