Mark: My Journey Building an AI Business Analyst Assistant
Why I Started Mark
If you’ve ever been the “data person” at a growing company, you know the feeling: every department has questions, and every question feels urgent.
As the data scientist and analytics lead, I built and maintained the dashboards and data pipelines for the entire company. But with only one of me, time quickly became the bottleneck. I’d get hit with questions like:
- “How are we doing this month compared to last month?”
- “Why are sales down?”
- “What should we focus on right now?”
- “Is retention improving, or are we losing customers?”
These are exactly the kind of questions that dashboards are supposed to answer. But the reality? Most business users don’t want to hunt for answers in a dashboard — they just want a direct, actionable response, right when they need it.
That’s why I started building Mark.
The Vision: A Specialized Agentic System for Business Analysts
Mark is my attempt to “duplicate myself” — not just as a tool, but as an always-available assistant that understands business context and can engage in real, back-and-forth conversations.
Mark isn’t just a chatbot. Under the hood, it’s an AI-powered business analyst that connects directly to your BigQuery data warehouse, performs complex analysis, generates visualizations, and crafts clear, data-driven narratives.
The goal: turn raw data into direct business insight — with zero friction for the user.
What Makes Mark Different
- Conversational Interface: Ask questions in plain language. Mark handles the translation to data queries, even for nuanced or ambiguous business problems.
- Agentic AI: Mark isn’t just reactive. It takes initiative to ask clarifying questions, explore root causes, and pull in relevant context from your data — just like a real analyst would.
- Direct-to-Insight Workflow: No more dashboard digging. Mark delivers answers, explanations, and charts right in the conversation.
- Built for Busy Teams: Mark is designed for companies where the “data person” is overextended, and business users need fast, clear answers — not a new dashboard to learn.
The Problem I’m Solving
At the heart of it: scaling myself. I want to empower every business user to get the answers they need, when they need them — without waiting in line or learning SQL.
With Mark, the experience should feel like having your own personal business analyst, 24/7.
Where I’m At Today
Mark is live and integrated with our BigQuery data warehouse. I’m continuing to iterate, improve the agent’s reasoning, and expand the types of analysis it can deliver. My focus right now is on making the experience as natural and trustworthy as possible, so users feel like they’re working with a real collaborator — not just another tool.