Mantu x Dust partnership: turning AI adoption into business value. Mantu and Dust are partnering to help organizations move from AI ambition to operational use. Dust provides the platform to build, deploy and improve AI agents. Mantu brings the expertise to integrate them into business processes, transform workflows and support adoption, with the governance needed to manage usage, costs and outcomes. The partnership builds on Mantu's own experience deploying Dust internally. In just three months, more than 1,600 employees adopted the platform and created over 3,000 AI agents. Across 12,000 people in more than 60 countries, Mantu is testing what it takes to integrate AI into everyday work, give teams ownership of their agents and establish the right governance. Mantu is now bringing this combination of technology and operational experience to its clients. By combining Dust's platform with Mantu's knowledge of business processes, technology environments and organizational constraints, the two partners help organizations identify relevant use cases, integrate AI agents into existing workflows and support teams in adopting new ways of working. AI adoption at scale in only three months. Mantu launched its Dust Sprint in May 2026 to bring AI closer to the daily work of its teams... Three months later, already more than 1,600 employees had joined the platform, nearly 1,200 were using it each month, and teams were exchanging more than 30,000 messages per week on Dust. The rollout reaches more than 60 offices and spans several Mantu businesses, including Amaris Consulting, LittleBig Connection and Agence Mantu. Teams are using AI to access internal knowledge, reduce repetitive tasks and collaborate across functions. To guide this transformation, Mantu's leadership developed 40 action plans that identified more than 400 use cases and set priorities for different business activities. Ten dedicated AI Transformation Leaders and a community of AI Champions support teams as they put these use cases into practice. This organization helps connect AI adoption with business needs. Teams can identify where agents can support their work, build solutions around specific tasks and share useful applications with colleagues. The objective is to help teams spend less time on repetitive work and more time on activities that rely on their expertise, from analysis and strategic recommendations to decision-making and client relationships. AI agents support teams in every day work. Dust provides a platform where teams can build, deploy and improve custom AI agents. These agents can support specific tasks and be shared across departments when other teams face similar needs. At Mantu, commercial teams use agents to find internal expertise and references, analyze markets and prepare responses to requests for proposals. Easier access to knowledge helps teams mobilize capabilities across an international organization and prepare more relevant responses for clients. AI is also changing internal processes. At Amaris Consulting, an agent supports manager onboarding by handling several tasks previously completed manually, including appointment coordination, equipment requests and the transmission of training information. The estimated time saved is 10 hours per onboarding. These examples show how AI agents can support different business activities, provided they address clearly identified needs. Their value depends on how well they fit into existing workflows, how easily teams can use them and whether the solutions can be shared or adapted elsewhere. As adoption grows, organizations also need to keep track of the agents they create, their owners, their access permissions and the resources they consume. A transformation structured across Mantu. Mantu has also established a governance framework to support this development. Each agent has an identified owner and belongs to one of three categories, depending on its intended use: "Corporate" agents are available across Mantu, "Operational" agents are restricted to specific teams, and "Personal / Unpublished" agents remain in private spaces during development, testing or evaluation. This framework gives teams room to develop solutions while maintaining oversight of data access, usage and ownership. Mantu also monitors token consumption across AI models and applies a FinOps approach to manage AI costs. Building and deploying agents requires people who understand both the technology and the work it needs to support. At Mantu, Hybrid Builders bring together business expertise and technological skills to help teams identify opportunities, design relevant solutions and integrate them into everyday operations. Working close to business teams, Hybrid Builders identify processes that can be improved, build agents around operational needs and help useful solutions reach other teams. They also contribute to agent reliability, security, compliance and readiness to scale. Their role extends across the AI transformation process. They connect business priorities with technical capabilities, while helping teams adapt their workflows and adopt new ways of working. From internal deployment to client transformation. Mantu's work with Dust follows its Customer Zero approach: testing technologies within its own organization, learning from their use and applying that experience to client projects. The partnership combines Dust's platform for building and deploying AI agents with Mantu's expertise in consulting, technology integration and organizational transformation. Mantu supports clients throughout the process, from identifying relevant use cases and connecting agents to data to training employees, managing change and establishing governance. The approach also covers the measurement of value and the management of AI consumption and costs. Through the AI Operator Sprint, Mantu aims to help organizations move towards operational use of agentic AI, drawing on its own deployment experience and international presence in more than 60 countries. "Our ambition is to transform the way an organization of 12,000 talents works, mobilizes its knowledge and creates value. Our Hybrid Builders give business teams the ability to build this transformation. By being our own first client, we learn quickly, adjust our methods and bring our clients an approach tested against operational reality." As Alizé Santini, Strategy and AI Transformation Director at Mantu Mantu also maintains its technological independence, selecting solutions according to each client's business needs, IT environment, security requirements and sovereignty constraints. Making agentic AI work in practice. Scaling AI adoption requires clear business priorities, teams with the skills to build relevant solutions and governance that supports responsible use. Mantu's deployment of Dust provides direct experience of how these elements can work together across an international organization. Through its partnership with Dust, Mantu is bringing this experience to clients. The focus is on integrating AI agents into business processes, supporting the people who use them and managing their performance over time. As AI becomes part of everyday work, organizations will need to keep improving how they identify use cases, share solutions and measure results. The aim is to make AI agents a useful resource for teams and a practical driver of business value.

Snowflake Ventures enterprise AI funds trust layer. Snowflake Ventures says enterprise AI scale fails on infrastructure, not models, and points to Dust and Gray Swan as its trust layer bets. StartupHub.ai Staff Snowflake frames Snowflake Ventures enterprise AI as a bet on the layer between models and apps, where pilots either become products or stall. Head of Snowflake Ventures Harsha Kapre named Dust and Gray Swan on Sep 8 as examples of that layer in production. How governance failures turn into security failures. Kapre argues most agent programs do not fail on model quality but on missing identity-aware access, policy guardrails and a single source of truth. Without that foundation even capable models cannot safely act across business workflows, so governance gaps become security gaps. Think of it like giving interns master keys to file paperwork, fast work but no audit trail until something leaks. Partner Paid placement Dust targets the workflow side, giving teams shared knowledge, search and Model Context Protocol connectors to run cross-functional agents on any model. Gray Swan targets the runtime side, scanning for vulnerabilities and enforcing protections before agents touch production data. Together they map to the three capabilities Snowflake calls foundational: model choice, governance at scale, and human-agent collaboration. Why production trust is still unfinished. Snowflake already pushed this thesis days earlier with CoCo, its coding agent that keeps data inside the perimeter, so Dust and Gray Swan extend the same inside-the-governance-boundary pitch. The overlap matters because Dust is integrating with Cortex Agents to ground agents in Snowflake data, which tightens Snowflake as the source of truth but also concentrates lock-in risk if governance lives there. Gray Swan fills a hole Snowflake cannot credibly fill alone, runtime defense for nondeterministic agents that traditional data controls were never built to watch. Databricks has made parallel governance bets, so Snowflake is signaling it will buy distribution for this layer rather than concede it. Neither announcement discloses investment size, terms or product milestones, so there is no way to judge traction beyond described GTM and ops usage at startups and enterprises. Procurement will now ask how agents inherit identity, log actions and revert bad writes, not which model they use. Until vendors publish third-party tests for those controls and clear rollback paths, the agentic enterprise stays a governed pilot for most shops. Snowflake is betting outsiders solve its hardest trust problem faster than it can ship natively. (C) 2026 StartupHub.ai. All rights reserved. You may not republish this article in full without a license. Search engines and AI research tools may crawl and summarize for reference. Bulk reproduction or model training requires a license. See our terms. Discussion. Join the conversation below StartupHub.ai Staff Editorial team The staff writers of StartupHub.ai, ranging from investment analysts to avid AI tool users, early adopters and critical enthusiasts. Backgrounds span engineering, business and the arts. We hold every piece to rigorous standards of research and review.