AI for Founders with Ryan Estes The AI Analyst That Never Sleeps: Burak Karakan of Bruin
Say Hello to Your AI Data Analyst: How Bruin Is Replacing Headcount, Not Just Dashboards
There is a question Burak Karakan wants every founder to ask themselves right now: Do you know where your agents are?
Burak is the co-founder of Bruin, an AI data analyst that connects directly to your data warehouse and answers any question in under 90 seconds, right inside Slack, Microsoft Teams, or a clean web UI. No dashboards to wrangle. No tickets to the data team. No waiting two days for a report that is already out of date by the time it lands in your inbox.
Calling in from Istanbul, one of the world's oldest crossroads of culture and commerce, Burak brings the kind of perspective that only comes from years of building data infrastructure inside big enterprises and scrappy startups alike. That experience is the foundation Bruin was built on.
The Framework: Data as the Operating System of Agentic AI
Burak lays out what he calls the Virtual Data Team model. As companies begin spinning up multiple AI agents across marketing, sales, operations, and support, those agents will need to collaborate, just like humans do. The missing piece is not more agents. It is a centralized, governed, trustworthy data layer that all of them can query reliably.
Bruin fills that role. Think of it as the data team member that every agent in your org chart can ping before making a decision.
Key principles of the framework:
- Data still lives in your warehouse (Snowflake, BigQuery, etc.). Bruin does not move or copy it.
- Every action the agent takes is traceable. You can click through and see exactly how it arrived at any answer.
- Granular access control means marketing agents only see marketing data, while executive channels get broader access.
- Multiple deployment models are available: fully managed cloud, hybrid, or fully on-premise with your own LLMs.
- An MCP server exposes Bruin's full capabilities so other agents can query it programmatically.
The Experiment: From Reactive to Proactive Intelligence
Burak draws a sharp distinction between a reporting tool and a reasoning system. Bruin starts as the former and is actively evolving into the latter. Current customers are already using it to route incoming support tickets through the AI analyst before the support agent even sees them, pulling customer purchase history, validating claims, and generating accurate responses. What used to take two to three hours per ticket now takes about 40 seconds.
The next frontier Burak is building toward: agents that proactively surface problems you have not thought to ask about yet. Upcoming capacity shortfalls. Campaign spend misaligned with available sales bandwidth. Churn patterns hiding in plain sight. The data already knows. Bruin is learning to tell you before you ask.
The Wild West Warning: A Framework for Agent Governance
Burak introduces what might be called the Do You Know Where Your Agents Are test. As organizations deploy more and more autonomous agents, the risks compound fast if data access is uncontrolled.
His governance framework:
- Run data quality checks at onboarding before the agent ever touches live data.
- Assign read-only permissions scoped to exactly what each agent needs.
- Use agent-controlled outputs (one agent checks another agent's answers before they surface to users).
- Set hard spending limits per query so no agent can run a runaway Snowflake job.
- Control internet access permissions per agent, per channel, per use case.
The punchline: if your agent only has read access to two marketing tables, the blast radius of any mistake is tiny. Structure the permissions right and you can let the agents run free.
https://www.linkedin.com/in/burakkarakan/
https://www.linkedin.com/in/estesryan/
https://kitcaster.com/application