EveryPoint puts verified stockpile data inside AI assistants

EveryPoint launched the Stockpile Reports MCP in beta for Business customers, letting AI assistants such as Claude, ChatGPT, Gemini and Copilot query verified stockpile measurements directly. The move gives construction, aggregates and mining operators a way to ask natural-language questions about their piles and get answers from their own data, not estimates.

Why it matters: - EveryPoint is trying to make stockpile data usable by AI assistants without losing verification or customer control. - The update targets bulk-material operators in construction, aggregates and mining, where pile data has traditionally been hard to track and hard to query. - EveryPoint says the approach gives businesses a single source of truth for stockpiles as AI tools take on more operational work.

What happened: - EveryPoint introduced the Stockpile Reports MCP, a new interface that makes verified bulk-material data callable by AI assistants. - The beta is available now to Stockpile Reports Business customers. - The MCP works with assistants including Claude, ChatGPT, Gemini and Copilot. - The company framed the launch as part of an emerging agentic supply chain, where operators can ask AI questions in plain language and get answers grounded in verified measurements.

The details: - The Model Context Protocol, or MCP, is an open standard introduced by Anthropic that lets AI assistants connect securely to outside tools and data. - Through the Stockpile Reports MCP, an assistant can retrieve a pile’s verified volume, tonnage, location, product type, change over time, underlying report imagery and coverage-confidence scores. - Example questions include asking for total tonnage of #57 stone across all yards as of yesterday or estimating when Yard A hits its reorder point based on this week’s sell-through. - Stockpile Reports has turned images of the physical world into measurements for more than a decade, according to EveryPoint. - EveryPoint has built computer vision and machine learning models since 2011. - The Stockpile Reports data repository dates to 2012 and includes hundreds of thousands of distinct stockpiles, more than 5 million measurements and more than 14 billion tons of material measured. - The technology is protected by 13 granted patents. - The company says customers decide who can connect to their stockpile data and who benefits from access. - Stockpile Reports is built by EveryPoint and can turn a phone, drone or camera image into verified volume and tonnage for every pile on every site.

Between the lines: - The launch is a bid to position verified industrial data as infrastructure for AI agents, not just a dashboard or reporting tool. - EveryPoint is contrasting verified physical measurements with general AI models that may not have access to customer-specific operational data. - The company is also signaling that data control could become a selling point as more businesses adopt AI assistants for day-to-day decisions.

What's next: - The Stockpile Reports MCP is in beta and Business customers get first access. - Existing customers can opt in from within their Stockpile Reports account. - Interested users can join the beta waitlist at more information. - EveryPoint also points readers to learn more about Physical AI.

The bottom line: - EveryPoint is turning verified stockpile measurements into AI-accessible operational data, with customer-controlled access at the center of the pitch.