Databricks launches Genie One, an AI agent that promises to bring intelligent automation to every area of the business
Artificial intelligence has already become routine for software engineering teams, but most companies are still operating in trial-and-error mode when it comes to automating processes in areas like sales, marketing, and finance.
The problem is not a lack of tools — it is a lack of context.
Traditional copilots try their best to help, but when business knowledge is scattered across outdated documents, disconnected systems, and even locked inside employees’ heads, AI just guesses. And guessing, depending on the department, can get expensive.
That is exactly the gap Databricks decided to tackle head-on with the launch of Genie One, announced during the Data + AI Summit 2026 in San Francisco. The pitch is ambitious: an AI agent that does not just answer questions but acts on behalf of employees, understands the real context of the organization, and can actually execute tasks.
Less guesswork, more results. 🎯
In the sections below, you will learn what makes Genie One different from everything we have seen so far, which features ship alongside it, and what this could mean for the future of enterprise automation.
What is Genie One and why it is different
Genie One is not just another generic AI assistant that answers questions based on public internet data. Databricks built this agent to operate inside a company’s own data ecosystem, connecting directly to internal sources such as Databricks SQL tables, data pipelines, catalogs, and even technical documentation stored in the corporate environment. That means when a sales analyst asks about last quarter’s performance by region, Genie One does not guess an answer based on generic patterns — it queries the organization’s actual data and delivers a precise, contextualized, and actionable response.
That difference might seem subtle, but in practice it changes everything because it eliminates the main problem with today’s copilots: the lack of grounding in business reality.
Beyond answering questions, Genie One was designed to execute tasks. That puts it in a different category from ordinary conversational assistants. It can trigger automated workflows, fire off notifications, update records, and coordinate actions across different systems — all from a natural-language instruction. Imagine a marketing manager asking the agent to compile campaign results, cross-reference them with conversion data, and send a summary to the team, without opening five different tools, without copying and pasting anything, without waiting for an available analyst. Genie One simply does it.
This level of operational autonomy is what Databricks calls agentic AI, and it is where the company is placing its bets for the next wave of artificial intelligence adoption in the enterprise.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
Genie Ontology: the context layer that changes the game
The big technical ace up Genie One’s sleeve is Genie Ontology, a context layer that continuously self-improves. Unlike solutions that rely on a fixed dataset to function, Genie Ontology actively sweeps across an organization’s entire body of knowledge. It scans structured and unstructured data, corporate documents, internal content, connected applications, and even learns from employees’ own interactions.
The result is a deep, constantly updated map of everything the company knows, where that knowledge lives, and how it all connects. It is as if Genie One had a complete mental map of the business rather than just loose fragments tossed into a database.
In practice, Genie Ontology pulls knowledge from every source it is granted access to, including the Databricks platform itself but also workplace applications, shared files, support tickets, chat tools, and even meeting recordings. With that solid foundation of contextual truth, Genie One can generate responses grounded in real, reliable business knowledge, which allows it to take the right action instead of simply guessing.
Ali Ghodsi, co-founder and CEO of Databricks, was blunt during the event: most of today’s enterprise AI systems operate with false confidence. According to him, if a CFO asks why margins shifted and the AI cannot explain, or if a sales leader wants to identify the next upsell opportunity and the system comes up empty, the problem is not artificial intelligence itself — it is context. Genie Ontology exists to solve exactly that, continuously learning the context of data everywhere to deliver faster answers and more accurate agents. 🧠
How it works in practice: SQL instead of hallucination
A technical detail that makes a huge difference in how Genie One operates is the way it retrieves information. Instead of trying to reason over fragmented documents and risking hallucination, the agent retrieves the context it needs through direct SQL queries. That is a much more robust approach because it relies on structured, verifiable data rather than trusting solely in the generative capabilities of the language model.
But Genie One goes beyond just answering questions accurately. It also offers visual interfaces, such as interactive charts, and lets users set up custom alerts. So, for example, if a financial analyst wants to be notified every time a metric crosses a certain threshold, they can configure that directly in the agent without building a separate dashboard or asking the data team for help.
Another key point is the integration with the Model Context Protocol (MCP), which allows Genie One to use third-party software and tools to execute actions within any business workflow. This opens up a massive range of possibilities because it means the agent is not locked into the Databricks ecosystem — it can interact with the rest of a company’s tech stack natively.
Genie Agents: turning conversations into reusable automations
Alongside Genie One, Databricks also made Genie Agents generally available. The concept is compelling: any conversation a user has with Genie, the company’s original AI chatbot, can be turned into a reusable agent that inherits the source data, instructions, and behavior from that interaction.
In practice, this means that if an analyst creates a specific query flow that works well for generating a weekly sales report, they can save that flow as a standalone agent. After that, any teammate can run that agent to replicate the same workflow without rebuilding everything from scratch. It is automation with reuse, something that makes a lot of sense for scaling AI adoption across large teams.
Employees can use these agents to run repeatable workflows and significantly speed up their day-to-day productivity. It is a smart way to democratize access to automation without requiring every individual to master the technical aspects behind the tool. 🔄
Genie App Builder: vibe coding for the business user
Another highlight announced at the event was the Genie App Builder, which offers a full vibe coding environment aimed at business professionals. The idea is that anyone can upload relevant context and generate a preview of an application connected to that data, all in a visual and intuitive way, with no need to write complex code.
The key point here is that applications generated by the App Builder are fully governed by the Unity Catalog, Databricks’ data governance layer. This ensures that access permissions, traceability, and data lineage are preserved even when the person building the application is not a data engineer. It is democratization with governance, a balance many platforms try to achieve and few actually pull off.
This deep integration with Unity Catalog solves one of the classic problems with enterprise AI tools: security. Instead of creating a parallel layer of access control, Genie One and the entire Genie ecosystem inherit the policies already defined by the data team, which makes life much easier for IT and reduces the risk of sensitive information leaks. It is artificial intelligence that plays by the house rules. 🏠
Updates for data engineers: Genie Code and Genie ZeroOps
Databricks did not forget about technical professionals in this launch cycle. Genie Code, the tool that helps data engineers plan, build, and run data engineering and analytics workflows, received significant updates. It now includes the ability to track progress and review each step across different projects, making the management of complex pipelines much more transparent and organized.
Genie ZeroOps is a heavyweight addition: it is a background agent designed to monitor, investigate, and propose fixes autonomously for data pipelines, tables, machine learning models, and other components of the data infrastructure. In practice, it is like having an on-call engineer who never sleeps, always watching for potential issues and ready to suggest solutions before they impact operations. For anyone working in data engineering, this could drastically cut the time spent on troubleshooting and reactive maintenance. ⚙️
Pricing model: pay-as-you-go instead of traditional SaaS
In a move that could attract plenty of attention in the market, Databricks chose not to follow the traditional software-as-a-service licensing model for Genie One. Instead, the company adopted a more straightforward pay-as-you-go model where customers pay for the tokens they consume.
This approach eliminates the fixed cost of licenses and allows companies of different sizes to experiment with the agent without a heavy financial commitment right out of the gate. It is a strategy aligned with how the AI market has been moving, with organizations preferring more flexible models that scale according to actual usage of the tool.
AI-powered workflows: automation that finally makes sense
One of the most talked-about aspects of the launch was Genie One’s ability to orchestrate complex workflows autonomously. In practice, this means the agent can chain a sequence of actions, validate intermediate conditions, make decisions based on the data it encounters along the way, and only then complete the requested task. This behavior is much closer to what a human analyst does than any rule-based automation we have seen to date.
A concrete example: a finance team can ask Genie One to identify overdue invoices above a certain amount, cross-reference them with the customer relationship history, and suggest the most appropriate approach for each case — all in seconds, based on real data from the company’s CRM and ERP.
Databricks also highlighted that Genie One integrates natively with external tools through connectors, which opens the door for these workflows to extend beyond the Databricks ecosystem and reach platforms widely used across large organizations. This interoperability is strategic because it eliminates the need to replace an entire existing infrastructure before using the agent. Companies can start right where they are, connect Genie One to the systems they already use, and expand adoption as confidence in the agent grows. It is a much more realistic adoption curve than solutions that demand a full digital transformation before delivering any results. 🚀
What makes this approach to automation even more interesting is the human-in-the-loop concept that Databricks built into Genie One’s design. For high-impact tasks such as discount approvals, customer communications, or significant financial transactions, the agent can be configured to pause the workflow and request human confirmation before proceeding. This is not a limitation — it is intelligence. The goal is not to remove human judgment from the most critical decisions but to free people from repetitive, mechanical tasks so they can focus on what truly requires discernment and creativity. Automation with responsibility, as it should have been from the start.
The market perspective: Databricks moving up the stack
Holger Mueller of Constellation Research noted that Genie One is a clear example of how Databricks is moving up the technology stack, going beyond the data layer the company is known for. According to him, Genie One is essential for exposing, finding, and making sense of much of the data stored on the platform autonomously, which makes it vital for building AI agents.
This move makes total sense when you look at the competitive landscape. Companies like Snowflake are also vying for the agentic AI space and trying to position themselves as the preferred platform for businesses that want to move beyond traditional data analytics. With Genie One and the full family of products announced at the Data + AI Summit, Databricks is clearly signaling that it does not intend to be just behind-the-scenes infrastructure — it wants to be the intelligence layer that employees use directly in their day-to-day work.
What this means for the future of enterprise AI
The launch of Genie One arrives at a time when the enterprise artificial intelligence market is going through a kind of maturation. The initial hype phase, where any tool with an embedded chatbot was sold as a revolution, is giving way to far more discerning demand. Companies want to know whether AI actually solves business problems, whether it integrates with what already exists, whether it is secure and auditable, and whether the ROI is tangible. Genie One appears to have been developed to answer all four of those questions convincingly, and that puts Databricks in a very strong position in this new phase of the market.
Another important aspect is the impact this type of solution can have on how data teams are valued within organizations. For a long time, data engineering teams worked behind the scenes building pipelines and maintaining infrastructure without business teams directly feeling the value of that work. With Genie One, the data these teams organize and govern becomes the direct fuel for an agent that any employee can use day to day. This creates a much more visible connection between technical work and business outcomes, which is likely to raise the recognition and strategic influence of these teams within companies. It is a virtuous cycle that benefits everyone. ✅
Finally, it is worth noting that Genie One also signals a clear direction for where Databricks wants to take its platform in the coming years. The company started as a data engineering solution for technical teams and gradually expanded its scope toward the business user. With Genie One, that shift becomes fully realized. The platform is no longer an exclusive environment for engineers and data scientists — it becomes an artificial intelligence layer accessible to virtually any role within an organization, from the financial analyst to the sales manager.
If this vision plays out in practice, the impact on how companies use their data could be enormous — and we might be watching, right now, the first steps of a shift that will unfold over the next decade. 🌐
