What it means
A data pipeline is an automated sequence of steps that collects data from sources, cleans and transforms it, and delivers it to where it is needed, such as a database, analytics tool or AI system. Pipelines run repeatedly so that data stays current.
Real-world example
A library network's AI search assistant depends on a nightly data pipeline. It exports new catalogue records, converts them into a common format, removes duplicate entries, splits descriptions into chunks, creates embeddings and updates the vector database. If the pipeline fails one night, the assistant does not know about that day's new acquisitions.
Why it matters for AI adoption
AI systems are only as current and reliable as the data that reaches them. Pipelines need owners, monitoring and error handling. When AI answers seem wrong, the cause is often an outdated or broken pipeline rather than the model.
Questions to ask
- How does data get from our source systems into the AI system, and how often?
- Who is alerted if the pipeline fails, and how quickly?