Stop Copy‑Pasting Between BI, Notebooks, and Docs
Business users want one chat box where they can ask, "What was MRR last quarter and why did it dip in APAC?" Traditional SQL chatbots can't read PDFs; generic RAG chatbots can't calculate metrics. PipeIQ bridges both worlds inside your Databricks Lakehouse.
Agentic Orchestration Flow
1. Ask in Plain English
Users type questions in our chat UI or their favorite interface (Slack, Teams, embedded widget).
2. Supervisor Agent Routes
An LLM‑enhanced overseer decides whether the query needs numeric answers (Text‑to‑SQL) or narrative context (RAG)—and can blend both when necessary.
3. Answers + Citations
The response merges SQL results with document excerpts, delivering trusted narratives with clickable citations, ready for dashboards or chat.
Key Capabilities
- ✅Text‑to‑SQL agent trained on your schema & joins
- ✅Automatic query optimization & row‑level security enforcement
- ✅Vector‑indexed unstructured data (PDFs, PPTs, HTML, images)
- ✅LLM responses with delta table snapshots for auditability
- ✅Citations & SQL preview in every answer
- ✅Slack / Teams / Web embed connectors
- ✅Fine‑tuning hooks for domain vocabulary
- ✅Runs entirely inside Databricks—no data leaves your region
Reference Architecture
A single supervisor agent coordinates specialized children: a Text‑to‑SQLagent hitting the Lakehouse and a RAG Insight agent querying the vector store—then composits answers with full provenance.
Give Your Team a Universal Chat Box
Schedule a discovery call to see how PipeIQ can unlock instant insights from every corner of your Lakehouse—without moving data.
