Agentic Data Quality: Why Your Data Needs It Twice
Agentic data quality means two things: quality for the data agents use, and agents that fix data issues. Here's why you need both, and how to start.
17 of 800 posts
Agentic data quality means two things: quality for the data agents use, and agents that fix data issues. Here's why you need both, and how to start.
A step-by-step agent observability example: how one bad answer was traced across context, performance, behavior, and output to its root cause, and fixed.
A large language model (LLM) is a neural network trained on large amounts of text to understand and generate human language.
An AI model is a system trained on data to recognize patterns and make predictions or generate content. A model takes an input, like a transaction, an image, or a question, and produces an output, like a fraud score, a label, or a written...
Tool calling, also called function calling, is how a large language model asks an application to run a function, such as querying a database, calling an API, or searching documents, and then uses the result in its answer.
A vector database is a database built to store embeddings, which are numeric representations of the meaning of text, images, or other data, and to quickly find the embeddings most similar to a query.
A schema change is any change to the structure of a dataset, such as a column being added, removed, renamed, or given a new data type.
An agent tells a customer their order shipped. It didn’t. Every service returned 200 OK, latency was normal, and the logs show a request in and a response out.
Your LLM app has been live for a month. Latency is under a second, the error rate is zero, and uptime is perfect. Meanwhile, a product manager forwards a screenshot: the bot told a customer about a discount that ended in March.
A customer asks whether they can still return a jacket. A chatbot would answer from what it was trained on. An AI agent looks up the order, checks the return policy table, sees the purchase was 34 days ago, notices the customer is a...
Two teams build the same support agent on the same model with nearly identical prompts. One answers correctly nine times out of ten. The other confidently quotes a shipping policy that changed in the spring.
A customer types “can I send this back?” Your help center never uses that phrase. It talks about the “return window” and the “refund policy.” Keyword search finds nothing.
A support agent is about to paste a customer’s full account number into a chat reply. An output check catches the pattern, redacts it, and logs the event. The customer never sees it. That’s a guardrail doing its job. Now a different run.
An agent tells a customer that refunds are available for 60 days. The answer is well written, confident, and even links to the refund policy. The policy says 30 days. Nobody hacked anything and nothing crashed.
An LLM firewall has a simple job. It sits between the people using your AI agent and the large language model (LLM) behind it, checking every message going in and every reply coming out so that it can block anything that looks like an...
Imagine you hire a bank teller. You check their references, confirm they have the right keys, and make sure they can’t touch the safe-deposit boxes. Then you turn off the cameras and never count the cash.
A support agent tells a customer they qualify for a refund they don’t. The trace exists. The source table exists. In the post-incident review, three teams point at each other: the AI engineers who built the agent, the data team that owns...