From the K12 Analytics Engineering team

Kai

A conversational analytics agent built on governed metrics, deterministic SQL, and your data team's expertise.

Ask in your own words

No training required. Just type what you're wondering about — the way you'd ask a colleague in the hallway.

Get answers with meaning

Kai doesn't just return a number — it explains what it means for your context, your schools, your students.

Go deeper naturally

Follow-up questions flow like a real conversation — drill into a school, compare years, explore what's driving a trend.

The problem with most AI

Hope is not an architecture

Most AI analytics tools take a shortcut: connect a language model to a database and let it generate SQL. Sometimes the answer is right. Sometimes it invents a column that doesn't exist. In education, “sometimes right” isn't good enough.

A wrong chronic absenteeism number could misallocate resources, trigger unnecessary interventions, or erode the trust your data team has spent years building. Kai takes a fundamentally different approach.

Three pillars

Built for reliability, not just capability

Kai was designed around a simple principle: an AI tool for education must be at least as trustworthy as the dashboards it supplements.

Deterministic, not probabilistic

The SQL behind every answer is generated by the semantic layer, not by a language model. There are no hallucinated joins, no invented columns, no approximate logic. The math is the math your team already validated.

Your ontology, encoded

Kai carries your organizational knowledge: what each metric means, what thresholds matter, what’s normal for your district at this time of year. This is context that turns a correct number into a useful insight.

Zero cross-district exposure

Every client operates in an isolated BigQuery environment. Kai accesses only your data. There is no shared model, no aggregated training set, no data leaving your environment.

A partnership in trust

Your expertise is the engine

We built the infrastructure — the warehouse, the semantic layer, the agent architecture, and the reliability patterns tested across 20+ school networks. But the real value comes from your data team: the metric definitions they've validated, the thresholds they've set, the context about what matters in your schools.

Kai doesn't replace that expertise. It distributes it. Every educator in your district gets access to the same rigor and insight that your data team brings to their work.

Common questions

What you might be wondering

What is Kai?

Kai is a conversational analytics agent for K–12 school districts. Instead of clicking through dashboards, you ask questions in plain English, like “What’s our chronic absenteeism rate this year compared to last?” Kai returns a sourced answer with context about what the numbers mean for your schools.

How does Kai generate answers?

Kai does not write SQL with a language model. It translates your question into a query against the semantic layer your data team has already validated. The language model handles the conversation (understanding your intent, explaining results, suggesting follow-ups), but the math comes from deterministic, governed metric definitions. No hallucinated joins, no invented columns.

What are the usage limits?

Each organization receives 50 million tokens per month. That’s enough for hundreds of in-depth conversations across your team. Usage resets monthly and is tracked at the organization level, not per user, so your entire staff shares a pooled allowance.

What data can Kai access?

Kai accesses the metrics and dimensions defined in your district’s semantic layer. That’s the same governed data that powers your existing dashboards: enrollment, attendance, chronic absenteeism, assessment results, and other metrics your data team has configured. Kai cannot access raw tables directly. It works only through the validated metric definitions your team controls.

Trusted data. Natural conversation. Every educator.

Kai is included in the K12 Analytics Engineering platform.

Sign In to Kai