Data to decisions. Automated.
Reliable answers from your data.
What I build
AI and agentic systems development
End-to-end: architecture, development, deployment. AI agents, RAG systems, connectors (MCP servers), Claude CoWork plugins, natural-language data querying. The AI layer that lets your team ask questions in plain words and get correct, data-grounded answers.
Knowledge graph and ontologies development
Graph databases (Neo4j), ontology design. Decision systems where business logic and audit trails are explicit and every answer is traceable.
Automated analytics and unified BI
A single source of truth across your source systems. Unified analytics over finance, ops, HR, and product. Dashboards and tools that let leadership get answers directly.
Includes: Data pipelines (ETL/ELT), data modeling (semantic modeling), visualization, UI/UX
Financial modeling and analysis
FP&A and budget consolidation. SaaS metrics and unit economics. Funnel analysis across marketing, sales, and recruitment. Metric decomposition (what changed, why, and by how much). Actual vs. budget and prior year. Resource planning with sensitivity analysis.
AI literacy training and support
Hands-on workshops. Help your leadership and teams understand what AI can (and can't) do for their workflows.
Data literacy and BI self-service training
Teach your team to explore the data and answer their own questions. Fewer requests funneled through analysts, more of the company working with data directly.
Process automation and documentation
Automate manual reporting and data processes. Document for handover so they keep running without ongoing support.
How I work
Map it
Requirements gathering first: the decisions you need to make, the people who make them, and the sources that should support them.
Model it
One trustworthy model: extraction, transformation, and a layer that matches how your team actually thinks.
Make it usable
Applications your team can drive directly. Built in short, agile iterations: you work with a live version early and steer what comes next.
Hand it over
Automated and documented, so it runs without me.
Sample work
Power BI MCP connector
Privately query your published Power BI models from any MCP client, with row-level security enforced on every query, no need for Fabric license. The foundation for the natural-language layer I build on top.
Get the connector → Deployed BIArabtech Data
Deployed BI dashboard for a public-data analytics project. Full data pipeline and interactive visualization.
Visit the dashboard ↗
Querist is my studio. I'm Effrat Katz.
AI can write a query in seconds. Whether the answer is right depends on everything around the model: how the data is structured and what the metrics actually mean. Engineering that context is what I do.
I built that context by hand for six-plus years before AI could use it, as a BI consultant for B2B SaaS companies: semantic models that make numbers trustworthy, one definition of revenue, one source of truth.
That's the layer AI needs now. Most AI consultants haven't done the data work. Most BI consultants don't build AI systems. I do both, so the answers hold up.
Certified in Agentic Data Analysis (UpScale Analytics, 2026) and Neo4j GraphAcademy (Neo4j, Cypher, and Graph Data Modeling, 2026).
I work in English and Hebrew. Find me on LinkedIn.
Let's talk
Tell me what you're trying to solve with your data.