You are the one who signs. Here is what you are signing for.
One of your psychologists probably sent you this. AUTOMATA is report-writing software built by a practicing California school psychologist. It does not decide anything. It reads the score reports your team already produces, builds the tables and graphs, drafts the narrative in your district's own template, and checks every number in the prose against the table it came from, before a psychologist reviews and signs.
District and group subscription prices are negotiable · See all plans
A weak report is the expensive one.
The cost of a report is not the hours it took. It is what happens when it is challenged: an independent educational evaluation at district expense, a compliance complaint, a compensatory services claim, or a due process hearing with counsel and an outside expert on the clock. Those cases usually begin with the same three things — a score in the narrative that does not match its table, a conclusion the data does not support, or a required section that is missing. AUTOMATA checks all three before the report is signed, on every report, for every evaluator on the team. One avoided hearing pays for the district plan many times over.
5,400 hrs ÷ 8 hrs a day ÷ 190 school days ≈ 3.5
1,200 reports × 4.5 hours saved = 5,400 hours
District and group subscription prices are negotiable.
The time claim behind those hours is Dr. Valentine's own caseload: a report that took about five hours takes about thirty minutes. Your team's numbers will be their own, which is why the first step is a walkthrough on one of your reports rather than a contract.
One standard, every evaluator.
Your template, not ours
District report templates, eligibility criteria, required headings and required wording built in, so a report reads the same whoever on your team wrote it.
Timelines met
Evaluations finish without overtime, weekend work, or emergency contract evaluators to clear a backlog.
Capacity without hiring
Four and a half hours back on every report is assessment capacity you already pay for and cannot currently reach.
Interns to standard immediately
New and intern psychologists produce district-standard reports from their first evaluation instead of their fiftieth.
An audit trail
A log of every automated check and every human approval, so you can show how a report was built if anyone asks.
Retention
Paperwork load is the most common reason school psychologists leave a district. This is the part of the job they leave over.
Answered before you ask them.
Where does student data go?
Drafting runs on de-identified records. Names and identifiers stay in your district's own files, and nothing identifying is stored on our side. FERPA documentation, a data processing agreement and a security review are available on request.
Who is accountable?
Your psychologist, exactly as now. Every automated change is a proposal a person accepts or rejects, and nothing enters a report without approval. The signature is a human decision about a human child.
Does it replace judgment?
No. It does not determine eligibility, write findings your data does not support, or silently correct anything. Mismatches are flagged for the psychologist to resolve.
What does rollout take?
A walkthrough on one of your own reports, then we map your templates, criteria and forms, build them, and test on real reports with your team before launch.
The “Settled Findings” AI model: zero hallucinated scores, zero downtime.
Directors rightfully ask: how does the AI operate, and can it make up numbers? In AUTOMATA, the language model is never allowed to interpret raw scores or calculate cutoffs on its own. All clinical calculations and classifications are locked deterministically on your machine before any phrasing occurs.
Where it is controlled
District and evaluator controlled: Evaluated in main.js (composeNarrativesStep) and governed under Settings > AI.
When “Compose Test Narratives” runs (or during full automated runs), the app verifies your district AI policy. If enabled, it calls ai-writer.js (composeWithAi) using encrypted local keys (Anthropic Claude, OpenAI, or Groq). If disabled, it runs 100% offline.
The “Settled Findings” rule
Local classification first (narratives.js): The local deterministic engine computes every exact score, percentile, publisher descriptor band, relative strength/weakness, and classroom impact statement first.
Phrasing only (ai-writer.js): Only that structured packet of verified findings is sent to the LLM to format natural clinical prose.
Strict FERPA privacy: The student’s real name never leaves your Mac or district machine—it is anonymized as STUDENT before transmission and restored locally after the prose returns.
Verification & automatic local fallback
Deterministic number verification: Every single number, score, and percentile returned by the AI is cross-checked against the original raw test data. If any score or percentile is missing or altered, the AI output is thrown away.
Zero-downtime local fallback: If the district network drops, an API key limits, or AI is turned off, AUTOMATA seamlessly switches to narratives.js’s built-in local clinical text generator. You always get complete, publication-ready narratives with zero downtime.
From one report to the whole team.
Walkthrough
A live walkthrough on one of your own reports. No payment, no agreement.
Scope
We map your templates, eligibility criteria, forms and the systems AUTOMATA needs to reach.
Build and test
We build your district's version and test it on real reports with your team.
Train and launch
Onboarding from a practicing school psychologist who writes these reports every week.
Bring us one of your reports.
We will walk your team through AUTOMATA on a report your district has already written, and send whatever documentation your business office and counsel need to approve it.