Building AI applications has never been easier. Making them reliable in production remains the real challenge.
That’s because agents need more than intelligence - they need the right context, access to real-time data, and the controls to act with confidence.
The next generation of agentic software will be built across an ecosystem—the systems businesses run on, the applications people use, and the tools developers already build with. That is why we’re actively building this next generation with partners from day one, rather than bolting an ecosystem on afterward.
Today at MongoDB.local NYC, we launched the next chapter of our intelligent data platform, with partners front and center.
Building the agentic ecosystem with Atlas Agent Engine
Atlas Agent Engine, announced this week in Public Preview, gives enterprises a single foundation for building, deploying, and governing AI agents. We’re building it with the partners who are already helping enterprises navigate their most important technology shifts.
Six of the leading global systems integrators are launching with us: Accenture, Capgemini, Cognizant, Infosys, Tata Consultancy Services (TCS), and Tech Mahindra. They are building Atlas Agent Engine into the AI offerings they already take to market, backed by the industry depth, delivery muscle, and client relationships they have spent decades earning.
Enterprise clients are focused on addressing a common challenge: agent pilots multiply faster than the governance and context needed to run them at scale. Cognizant designs, orchestrates, and operationalizes agentic systems across the AI lifecycle, and MongoDB's Atlas Agent Engine is designed to give that work a governed, context-aware foundation to build on. Together, these capabilities are built to help clients progress from pilot to production while supporting the accountability and domain depth that governed AI at enterprise scale requires.
”That delivery foundation is complemented by a broader ecosystem of technology partners. Atlas Agent Engine is neutral across models and frameworks, so enterprises can use the technologies they already trust and adapt as the market evolves. Customers can choose their preferred authentication and security tools to keep agents safe, while technology partners like Arize AI help make those agents production-ready with trace monitoring.
Agents are moving from demos into production workflows, where reliability depends on both the context they act on and the ability to understand what happened at runtime. MongoDB gives agents access to the data they need, while Arize traces each run and supports agent evaluation across both the outcome and the path taken. We’re excited to be a launch partner for Atlas Agent Engine, helping enterprises ship agents they can understand, evaluate, and improve with confidence.
”That combination matters because getting agents into production is rarely a technology problem alone. It takes people who know an industry well enough to know which decisions an agent should be trusted with, and who are already in the room when enterprises make them.
It also changes where MongoDB shows up. Our partners are building agentic practices, training delivery teams, developing reference architectures, and shaping enterprise offerings across key markets worldwide. When those teams build on Atlas Agent Engine, that capability travels with them, extending MongoDB’s reach to more customers and industries than we could reach alone.
MongoDB provides the intelligent data and agent platform. Partners bring the expertise, workflows, and reach that help put the platform to work with all the tools that enterprises know and love. Together, we are building a production path for enterprise agents alongside some of the strongest ecosystem partners out there.
The next wave of enterprise AI will be defined by how effectively organizations connect trusted data, contextual intelligence, and autonomous action. By combining Infosys Topaz and our experience in helping enterprises navigate large-scale AI transformation with Atlas Agent Engine, we are enabling clients to build intelligent systems that can reason, adapt, and deliver value in real-world business environments. Together, we are helping organizations accelerate modernization, unlock the full potential of their data, and establish a strong foundation for scalable, responsible adoption of agentic AI.
”Modernizing enterprise applications with Cognition’s Devin
Agents cannot act on data that is trapped behind legacy infrastructure. Before most enterprises can put an agent to work, they have to modernize the code tying that infrastructure together.
We’re taking on that challenge in partnership with Cognition. This week, we announced Devin for MongoDB Modernizations, a new offering leveraging Cognition’s AI software engineer to help customers modernize their systems and realize value from MongoDB faster and easier.
Now, Devin connects to MongoDB's Application Modernization Platform (AMP), giving it direct access to AMP's migration tooling. Those tools move and validate data into MongoDB Atlas, giving joint customers a faster, more automated path from legacy code to production on Atlas.
Modernization has two parts: rewriting business logic and data-access layers, and moving the underlying data. Devin handles the code, planning, and rewriting the business logic and data access layers case by case, while AMP’s deterministic tooling handles the data itself, moving and validating each record into MongoDB. Devin orchestrates the full process end-to-end, so customers get one coordinated migration instead of a patchwork of scripts and manual handoffs. In early joint testing, work that had taken five to six hours finished in a little over one.
What stays with the human engineers are the decisions that shape the outcome: how to model the data on the other side, and when to cut over. Teams get to production on Atlas in months rather than years, with their engineers freed up for what comes next.
Even the most ambitious engineering teams spend much of their capacity keeping legacy systems alive. Devin for MongoDB Modernizations solves that pain. Devin does the heavy lifting, rewriting code across the codebase, so customers reach production on MongoDB in weeks instead of years. Engineers can get back to building great software and solving hard problems.
”Meeting developers where they build
AI is changing not only what developers build, but how they build it. Increasingly, an application starts as a prompt. We want MongoDB to be there from that first prompt all the way through production, which is why we’re excited to announce three new partnerships that extend that reach.
Emergent helps builders turn a natural-language idea into a working AI application, now with Atlas as their default database. Through this partnership, builders on Emergent will be able to connect their Atlas account and ship with MongoDB through Emergent from the first prompt, rather than stumbling into the database decision after the fact.
Our users come in, describe an app, and our agents build it. But what they ask for keeps changing mid-build. The schema evolves the same way the conversation does. We believe that MongoDB is the best database for agentic coding. Nothing else we tested kept up with how our users build.
”Monk.io is a universal software operator, available as a plugin inside popular coding agents with MongoDB Atlas now as the target data layer. Customers can leverage coding agents such as Claude Code, Codex, and SpaceXAI to write the code, while Monk deploys and operates the application—including the Atlas data layer—in customers’ accounts, taking care of provisioning and networking to backups and recovery. Monk maintains a live map of the application and its infrastructure, executes changes deterministically, keeps credentials hidden from agents, enforces organizational permissions for every action, and requires human approval outside the agent’s chat for costly or destructive changes—giving joint customers more ways to run applications safely.
Retool gives teams a secure, governed way to use AI to build internal tools on the data their business already runs on. MongoDB has been a first-class Retool integration for years, and now Retool Cloud customers can connect directly to Atlas and provision scoped access without manually managing credentials. Together, we’re making it easier to ship apps on real business data without trading away the security and control enterprises need.
Emergent, Monk.io, and Retool join a much broader ecosystem extending MongoDB across the models, frameworks, tools, and services that make AI applications useful in the real world.
A vision for the ecosystem
None of this is a collection of one-off integrations. It is a deliberate recognition that the agentic era is the ecosystem era.
The enterprises that get agents into production over the next few years will not do it alone, and neither will we. They will succeed alongside partners who already understand their industry, their systems, and their data—on a foundation that does not force them to bet everything on a single model, cloud, or framework.
To every partner and sponsor with us at MongoDB.local NYC today: thank you. You are the reason this is an ecosystem story and not just a product story. We are only getting started.
Next Steps
Explore Atlas Agent Engine in Public Preview and see what you can build. If you're a partner interested in building with us, learn more about the MongoDB Partner Ecosystem.