a personalized learning companion, named after Lilavati, the 12th-century mathematical treatise by Bhāskarāchārya, the Indian mathematician who developed calculus 500 years before Newton.
LeeLaa reads whatever assessment your child already takes, your state benchmark (FAST, iReady, MAP, STAAR, CAASPP, MCAS, NWEA), tutor reports (Kumon, RSM, Mathnasium), or our own quick benchmark if you have none yet, and personalizes from minute one. No forty-minute placement test. No re-diagnosing what the school already knows.
Upload your child's last benchmark report (state assessment, tutor progress sheet, or classroom report). We extract item-level performance, map it to a 95-skill graph, and seed mastery priors. Cold start: 5–15 minutes, not 40. No report? We run a short benchmark.
State benchmark, iReady, MAP, Kumon worksheet level, RSM lesson, classroom report card, each source has a confidence weight and time decay. The profile compiler fuses them into a single posterior per skill.
Each standard or benchmark has its own LLM prompt, FL B.E.S.T. ELA.4.R.2.4,
Common Core CCSS.RL.4.5, TEKS, NGSS, and beyond. Items are aligned to the
actual standard at controllable difficulty. Coverage is auditable at the file system.
How it works
Drop your child's last benchmark PDF (state assessment, iReady, MAP, Kumon, RSM, or a classroom report) into the parent app. We redact the PII before it touches any LLM, then extract the item-level performance. No report? We run a short starter benchmark.
Each evidence source carries an explicit confidence weight (state benchmark = 0.90, tutor worksheet = 0.50) and time-decays at a 180-day half-life. The result is a per-skill posterior the parent can audit.
A 30-question arc, 4 warm-up + 18 core + 5 stretch + 2 trap + 1 review. Selector picks skills that close the child's documented gaps. Every question is three LLM calls: GENERATE the item, DIAGNOSE the answer, TEACH via Socratic explanation matched to the error type.
When your uploaded benchmark and the in-app sessions disagree about mastery, the disagreement is shown in the parent dashboard. You decide which to trust. No silent miscalibration.
For whom
You have a child in K–8 with a benchmark report in your inbox (FAST, MAP, i‑Ready, STAAR, CAASPP, MCAS, NJSLA‑ELA, SC READY, Georgia Milestones, CMAS, IAR, NJSLA, NY State Tests, Regents, TCAP, SOL, or your district's annual). You're already paying a premium for enrichment. You want what those programs do, but anchored to the standards your child is actually tested on.
Your teachers are told to "differentiate by standard" but the adaptive software you bought ignores the benchmark data the district already collects. We bridge them.
The thesis
Every state in America runs a benchmark assessment three times a year. The results sit in PDFs in parents' inboxes. LeeLaa is the layer that finally does something useful with them.
We're starting with 50 families, anywhere your district runs a benchmark assessment (FAST, iReady, MAP, STAAR, CAASPP, MCAS, or your state equivalent). Send us a note and we'll reach out as the pilot opens.
leelaa@sciencephalon.com →