4 Trends Shaping Data Management and AI in 2026
Here at Data Tribes, we encountered this article by Eric Avidon on TechTarget and thought it was worth bringing to the tribe. It breaks down four trends set to shape data management and AI through 2026 and honestly, it's a solid read.
1. Agents Need Context Semantic Models Are Step One
The article points out that connecting an AI agent to data is only half the battle. The agent also needs to understand what that data means in a business context and that's where semantic models come in.
Think of a semantic model as a shared dictionary for your organization's data: it defines metrics, relationships, and business logic so agents don't just retrieve data, they interpret it correctly. In multi-agent environments, this shared understanding is what keeps outputs consistent across teams.
Vendors like AtScale, DBT Labs, Google Looker, and ThoughtSpot have been offering semantic layers for years, but they were largely optional when organizations were building static dashboards. With agents now making autonomous decisions, that's no longer the case.
One gap is still worth watching: most semantic models express business logic as SQL, which isn't expressive enough to capture the full complexity of real business metrics. That limitation hasn't been fully addressed yet.
2. The Race to Standardize Agent Communication
When agents need to collaborate, think separate agents managing inventory, warehouse ops, and delivery routing they need a shared protocol. Google Cloud's Agent2Agent (A2A), launched in April 2025, is the frontrunner, with backing from AWS, Microsoft, Oracle, Databricks, and Snowflake. It also merged with IBM's Agent Communication Protocol in September 2025, consolidating two competing frameworks.
Whether A2A becomes the standard, though, is still an open question. The article raises a compelling counterpoint: MCP, already the default for connecting agents to data sources, may simply evolve to absorb what A2A does rather than the two coexisting. The market tends to crown one winner.
The need is clear. The standard that fills it is still being decided.
3. Automation Is Moving Up the Stack
This is the trend with the most far-reaching implications. Agents are no longer just supporting data teams, they're beginning to run business operations: customer support, invoice processing, payment approvals, supply chain optimization.
Within the data world, tasks like pipeline building, quality monitoring, observability, and governance are increasingly being handed to agents, with companies like Monte Carlo, Informatica, and ThoughtSpot already shipping in this space.
The article also raises a bigger question that we think will define a lot of conversations this year: as agents take over tasks previously done by people, where does the AI end and the team begin? That's as much a cultural challenge as a technical one.
4. Consolidation Is Already Happening
The late 2025 M&A activity wasn't random. Salesforce acquired Informatica. Fivetran and DBT Labs agreed to merge. IBM acquired Confluent. The article argues this is the start of a broader wave driven by the high cost of AI development and enterprises wanting simpler, more integrated platforms over a sprawling stack of specialist tools.
Independent vendors in data cataloging, observability, and ETL are most exposed; their buyer bases overlap heavily with larger platforms that are increasingly bundling similar capabilities.
That said, the article notes there's still room for independents that are genuinely cloud-neutral, model-flexible, and architecture-agnostic. Lock-in concerns aren't going away, and sophisticated data teams still value best-in-class tools.
