AI agents can now connect to data sources, build pipelines, manage credentials, and run production data work end-to-end — with humans watching, not driving
Datris today announced a major expansion of its agent-native data platform that makes AI agents true first-class operators of data infrastructure. Agents working through Datris can now connect to and continuously pull from data sources — S3, databases, internal APIs, and enterprise systems like Workday, Salesforce, and ServiceNow — build pipelines from scratch, generate validation rules and transformations in plain English, manage their own credentials, and have every action observed in real time, without a human writing glue code or sitting in front of a console.
“The data industry spent twenty years building tools for human engineers, and the last two trying to retrofit them for AI,” said Todd Fearn, founder. “We started over — without a human in the loop.”
— Todd Fearn
While the rest of the industry has spent two years bolting chat interfaces onto traditional data tools, Datris took the opposite approach: it rebuilt the data platform around the AI agent. Every capability is exposed through Model Context Protocol (MCP).
What’s new
Agents stand up their own data feeds. Datris introduces “taps” — recurring or on-demand pulls from a source that land data where the rest of the platform can use it. An agent describes the source in plain English; the platform owns the connection, the schedule, and the execution.
Marketing Technology News: MarTech Interview with Lee McCance, Chief Product Officer @ Adverity
Agents build pipelines, not just query them. An agent describes the work it wants done — ingest a CSV every hour, drop malformed rows, normalize timezones, land it in a warehouse — and the platform generates the schema, writes the data quality rules, produces the transformation logic, and stands up the pipeline as a single atomic operation.
Agents own their own credentials. An agent can request, store, rotate, and delete the API keys it depends on, scoped to credentials it created. Human-owned credentials remain protected and untouchable by agents — the platform enforces the line.
Every agent action is observed. A live operations view shows which agent invoked which capability, against which pipeline, with what result, as it happens. When something goes wrong, the platform returns errors in language the agent can act on, not a stack trace a human has to translate.
Marketing Technology News: What is a Full Stack Marketer; What MarTech Matters Most to Full Stack Marketers?
Open source, self-hostable
Datris is open source under AGPL-3.0. The full platform is on GitHub at github.com/datris/datris-platform-oss and runs on a single machine with Docker. Teams self-host the stack, inspect every line of code that touches their data, and extend the platform with their own MCP tools. There is no enterprise edition, no feature gating, and no telemetry. A hosted version is available at datris.ai for teams that prefer not to run the infrastructure themselves.
