Driven by curiosity. Guided by intelligence.
A consulting and premium app development firm. We work with a small number of industry-leading companies, and we are as selective as the firms who work with us.
Six intelligence-powered apps. One platform underneath.
Vitruvious
The platformThe first Learning Management and Intelligence System: the top 25 education products on one platform, one record and one log-in.
See it →Agentic IEP
For districtsTurns evaluation data into complete, compliant, individualized IEPs in minutes, and tracks every protected student.
See it →SammySchoolhouse
For kids, families and schoolsA learning companion that teaches every standard, supports wellbeing and turns every answer into progress evidence.
See it →SammyTutor
For familiesVoice-first, one-to-one tutoring that remembers each child and adapts every session.
See it →Language Learner Companion
For emergent bilingual programsManages the full spectrum of English learner requirements under federal and state law, for every student.
See it →Next Commit
For athletesA recruiting platform for high school, college and professional athletes.
See it →We build the apps. You own the market.
We partner with established companies that have strong sales and distribution. We handle the apps and the content; you handle the relationships and the sales.
We build
A finished, industry-leading app, designed and engineered to exceed the top product in your market.
We make it yours
Your brand, your colors and your workflows, delivered turnkey.
You take it to market
Your sales organization, your clients. We keep the app improving.
Why have 25 log-ins when you can have one?
Vitruvious is the first Learning Management and Intelligence System: the full functionality of the top 25 products in education technology today, on one platform, one student record and one log-in. Agentic IEP, SammySchoolhouse, SammyTutor and Language Learner Companion all run on it.
Student information system
Enrollment, demographics, attendance, scheduling, grades, transcripts
In place of PowerSchool SISLearning management system
Courses, content, assignments, grading, gradebook
In place of CanvasBenchmark & interim assessment
Adaptive benchmarks with RIT-compatible scale scores
In place of MAP GrowthUniversal screening
Fall, winter and spring screeners with risk flags
In place of Star AssessmentsCurriculum-based measurement
Probes, aimlines and decision rules
In place of aimswebPlusProgress monitoring
Real-time goal graphs for every student
In place of FastBridgeAdaptive practice
Personalized practice in reading and math
In place of IXLAI tutoring
Voice-first tutoring through Sammy
In place of KhanmigoSpecial education management
Agentic IEP: evaluations, ARDs, IEPs, timelines
In place of Frontline IEPSection 504 management
504 plans, reviews and accommodation logs
In place of SuccessEd 504IEP goals & SDI library
Goals, objectives and specially designed instruction
In place of GoalbookEnglish learner program
Language Learner Companion: identification, LPAC, monitoring
In place of EllevationDyslexia services
Screening, identification and services tracking
In place of LexiaGifted & talented
Identification and service tracking
In place of CogATMTSS / RTI
Tiers, interventions and decision rules
In place of Branching MindsBehavior & discipline
Incidents, interventions and reporting
In place of PBIS RewardsSocial-emotional learning
Check-ins, self-regulation and wellbeing
In place of PanoramaFamily communication
Messaging, digests and home-language updates
In place of ParentSquareCompliance & state reporting
Audit-grade timelines, Texas reporting
In place of EduphoriaMedicaid (SHARS) billing
Service logging, claims and denial prevention
In place of PCG EDPlanAnalytics & dashboards
District, campus and teacher intelligence
In place of SchoolzillaProfessional development
On-demand training and CPE tracking
In place of Frontline PLData integration
LTI 1.3, OneRoster and Ed-Fi connections
In place of CleverRecords & documents
Secure student files and evidence trails
In place of LaserficheSecurity & privacy
Role-based access, consent and full audit trail
In place of ClassLinkEvery area of need in education RFPs, on one platform.
We analyzed 41 open district and statewide solicitations from October 2026 and grouped every requirement into 27 areas of need. Vitruvious, with AgenticSchool, Agentic IEP, Language Learner Companion and SammySchoolhouse on one record, covers all 27 in its catalog of 292 functions across 24 modules.
Method: each solicitation's listed requirements were mapped to these areas; the full catalog covers every area. Live today means in production now; in build means in the catalog and on the build plan. With every function delivered, 14 of the 41 RFPs match at 100% and 22 at 90% or more (average 82%); adding the Evidence Package, survey builder and two further planned extensions raises that to 29 at 100%. Three requests sometimes bundled into the same solicitations, district finance and HR systems, online courses staffed by teachers, and world-language proficiency tests, fall outside the platform's purpose and are not counted.
Every federally protected student, tracked on one record.
Seventeen federal protections can apply to a single child. Vitruvious watches all of them at once: who qualifies, what each law requires, which clock is running and what proof is on file. When a student carries several protections, the obligations are reconciled in one place instead of across separate systems.
- Child FindReferral received; evaluation consent due in 15 school days!
- ELHome language survey complete; TELPAS window flagged✓
- ImmigrantNewcomer supports assigned✓
- MigrantCertificate of Eligibility on file; summer services offered✓
- NativeTitle VI eligibility form on file✓
- FERPASpanish-language consent captured✓
Example students for illustration.
- 01Special educationIDEA Part B
- 02Child Find & early childhoodIDEA Parts B and C
- 03Section 504Rehabilitation Act
- 04Disability accessADA Title II
- 05English learnersTitle III, EEOA
- 06Immigrant studentsTitle III
- 07Migrant studentsTitle I, Part C
- 08Homeless studentsMcKinney-Vento
- 09Students in foster careESSA Title I
- 10Low-income studentsTitle I, Part A
- 11Neglected or delinquentTitle I, Part D
- 12Military-connectedESSA
- 13American Indian & Alaska NativeESSA Title VI
- 14Race, color, national originCivil Rights Act Title VI
- 15Sex-based, pregnant & parentingTitle IX
- 16Disciplinary protectionsIDEA §615(k)
- 17Student privacyFERPA, PPRA
Workflow technology so your teachers can teach instead of doing paperwork.
Agentic IEP writes fully compliant IEPs, individualized and maximized for each child, with a level of granularity not seen in today's IEPs. Every IEP is expert created, checked for regulatory violations, paired with aligned SDI resources, and supported by AI toolkits with hundreds of individualized agents working on behalf of the educator and the child. It is the most advanced IEP intelligence and student management system on the market today.
- Aligned to all 50 states, national and state-specific standards
- Built-in compliance audit on every draft
- No boilerplate: every IEP's text is original to that child
- Prior Written Notice in English and Spanish
- Progress monitoring, caseload and daily tracking
- Agentic 504, resource kits and supports library
- District Command Center with compliance trace
- Zero-retention data architecture
From evaluation data to a complete Texas IEP.
Watch Agentic IEP read a Full Individual and Initial Evaluation, extract and cross-check every data point, then write the complete ARD document: eligibility, PLAAFP, TEKS-aligned goals, accommodations, modifications, STAAR designated supports, services and every required consideration, each one traced to the data behind it.
- 01 · Intake agentRead evaluation6 sections · 3 instruments
- 02 · Extraction agentExtract data14 of 14 extracted
- 03 · Validation agentCross-check4 of 4 checks passed
- 04 · IEP writerARD, eligibility & PLAAFP3 of 3 paragraphs
- 05 · Goal architectTEKS-aligned goals3 of 3 goals
- 06 · Services agentServices, supplements & PWNsupplements reviewed
- 07 · Compliance auditorTEA compliance audit18 of 18 passed
Cognitive
Academic achievement
Phonological processing
Curriculum-based measures
Observations and input
Eligibility determination
AARD meeting
| Student | Student A · Grade 4 · Sample Elementary |
|---|---|
| Meeting purpose | Initial ARD: eligibility and initial placement |
| Meeting date | 10/14/2026 (within 30 calendar days of the 09/22/2026 FIE report) |
| ARD committee | Parent · general education teacher · special education teacher · administrator (district representative) · M. Reyes, LSSP (interprets evaluation results) |
| Language | English; interpreter not needed |
BEligibility
The ARD committee reviewed the FIE dated 09/22/2026 (M. Reyes, LSSP) and determined that Student A meets eligibility criteria for a Specific Learning Disability in basic reading skills, reading fluency skills and written expression, with dyslexia identified. The disability adversely affects educational performance, and Student A needs specially designed instruction to make progress in the general curriculum.
1Present Levels of Academic Achievement and Functional Performance (PLAAFP)
Student A is a curious, verbal fourth grader who is most engaged in science and hands-on building. Overall cognitive ability is in the average range (WISC-V Full Scale IQ 98; administered 09/08/2026 by M. Reyes, LSSP), with relative weaknesses in working memory (82) and processing speed (85). Math problem solving is a strength (WIAT-4 standard score 96; administered 09/10/2026 by M. Reyes, LSSP), and listening comprehension is strong when text is read aloud.
The primary area of need is basic reading. Decoding is significantly below expectations (WIAT-4 Basic Reading 76), consistent with weak phonological awareness (CTOPP-2 composite 78). On grade 4 passages, Student A reads 52 words correct per minute with 91% accuracy and correctly decodes 8 of 20 multisyllabic words. Written expression is also an area of need (WIAT-4 79): a 3-minute writing sample produced 18 correct writing sequences, with limited organization and frequent spelling errors.
Effect on progress in the general curriculum: difficulty decoding grade-level text limits independent access to reading, science and social studies materials and to written assignments, even though content understanding is strong when text is accessible. Lower processing speed and working memory affect copying, multi-step directions and timed tasks.
2Measurable annual goals (TEKS-aligned)
Progress toward each goal is reported to the parent every nine weeks, at the same time as report cards.
3Instructional accommodations
| Oral administration or text-to-speech for content-area assignments and tests (not for reading-skill assessments) | Linked toBasic Reading 76 · strong listening comprehension |
| Extended time (1.5x) on tests and written assignments | Linked toProcessing Speed 85 |
| Directions given in short steps, with a check for understanding | Linked toWorking Memory 82 |
| Graphic organizers for all writing tasks | Linked toWritten Expression 79 |
| Speech-to-text for extended written responses | Linked toWriting CBM 18 CWS |
| Copy of class notes provided; reduced copying from the board | Linked toProcessing Speed 85 |
4Modifications
| Independent reading assigned at instructional level rather than grade-level text | Linked toOral reading fluency 52 WCPM |
| Spelling list limited to phonics patterns already taught | Linked toPhonological Awareness 78 |
| Written work graded on content and organization; spelling scored separately against taught patterns | Linked toWritten Expression 79 |
5STAAR participation and designated supports
Student A will take STAAR grade 4 Reading Language Arts and Mathematics (general assessment) with the designated supports below, each used routinely in instruction.
| Text-to-speech (Mathematics) | Linked toBasic Reading 76 |
| Speech-to-text (written responses) | Linked toWriting CBM 18 CWS |
| Spelling assistance | Linked toPhonological Awareness 78 |
| Individualized structured reminders | Linked toWorking Memory 82 |
6Schedule of services
| Service | Setting | Frequency | Provider |
|---|---|---|---|
| Specially designed reading instruction (evidence-based dyslexia instruction) | Special education resource | 45 min daily | Special education teacher |
| Written expression support | General education (inclusion) | 30 min, 3x weekly | Special education teacher |
| Related services | — | None required at this time | — |
| Supplementary aids and services | Text-to-speech and speech-to-text available in all classes; accessible digital versions of content-area texts |
|---|---|
| Supports for school personnel | Training for the general education teacher on accommodations and STAAR designated supports by 10/30/2026; 30-minute monthly collaboration between general and special education teachers |
| Dyslexia instruction provider | Special education teacher trained in evidence-based dyslexia instruction |
7Specially designed instruction by goal
| Goal | Specially designed instruction | Who and where | How often |
|---|---|---|---|
| Goal 1 · Reading fluency | Repeated reading of instructional-level text with corrective feedback; phrase-cued text for prosody | Special education teacher · resource | Daily, within the 45-minute reading block |
| Goal 2 · Decoding | Explicit, systematic structured-literacy instruction in syllable types and syllable division, with cumulative review | Special education teacher · resource | Daily, within the 45-minute reading block |
| Goal 3 · Written expression | Explicit modeling of paragraph structure with graphic organizer, sentence frames and a self-monitoring checklist | Special education teacher · general education class | 30 minutes, 3 times weekly |
8Instructional arrangement and least restrictive environment
| Instructional arrangement | 41: resource, less than 50% of the school day |
|---|---|
| Least restrictive environment | Student A remains in general education for math, science, social studies and specials. Removal is limited to targeted reading instruction that cannot be delivered with fidelity in the general classroom. |
| Nonacademic and extracurricular | Participates with peers in lunch, recess, field trips, clubs and school events |
|---|---|
| Physical education | General physical education |
| Transportation | Regular transportation; special transportation not needed |
9Special factors
| Behavior that impedes learning | No |
|---|---|
| Limited English proficiency | No. English is the home language; LPAC participation not required |
| Blind or visually impaired | No |
| Deaf or hard of hearing | No |
| Communication needs | No needs beyond those addressed in this IEP |
| Assistive technology | Yes. Text-to-speech and speech-to-text tools (see accommodations) |
10Texas ARD supplements
| Autism supplement (19 TAC §89.1055(e)) | Reviewed. Not applicable: eligibility is a Specific Learning Disability, not autism. The 11 required strategies are addressed only for students with autism. |
|---|---|
| Visual impairment supplement | Not applicable |
| Communication plan (deaf or hard of hearing) | Not applicable |
| Transition services and graduation planning | Not required at this time (student is under age 14) |
| Behavior intervention plan | Not needed |
| Dyslexia services | Applicable. Provided through special education (see services) |
11Other ARD considerations
| Assistive technology | Considered. Text-to-speech and speech-to-text tools provided (see accommodations). |
|---|---|
| Extended school year (ESY) | Considered. Not needed at this time; no evidence of severe or substantial regression. |
| Behavior | No behavior intervention plan needed. |
| Transition | Not required at this time (student is under age 14). |
| Parent concerns | Reading homework is frustrating. Addressed through instructional-level independent reading and family progress updates. |
12Prior Written Notice
| Action proposed | Begin special education services for a Specific Learning Disability with dyslexia, as described in this IEP |
|---|---|
| Reasons | The FIE shows needs in basic reading, reading fluency and written expression that require specially designed instruction |
| Data used | FIE dated 09/22/2026: WISC-V, WIAT-4, CTOPP-2, curriculum-based measures, teacher observation and parent input |
| Options considered and rejected | Section 504 accommodations only: rejected, because accommodations alone do not provide the explicit instruction Student A needs |
| Other factors | Parent input on homework frustration and interest in science |
| Implementation | Parent agreed to waive the five-school-day notice period; services begin 10/15/2026 |
13Parent participation and procedural requirements
| Provided to parent | Notice of Procedural Safeguards · A Guide to the Admission, Review, and Dismissal Process · Prior Written Notice (English) |
|---|---|
| Consent | Parent consent for initial provision of special education services obtained |
| Committee decision | ARD committee in agreement |
14Deliberations
The committee reviewed the FIE and the parent’s input. The parent agreed with eligibility, the goals and the services, and asked for regular updates on reading progress at home. The general education teacher will implement the accommodations and designated supports in all classes, and the special education teacher will share progress every nine weeks. All members agreed.
15ARD committee signatures
| Role | Name | Decision |
|---|---|---|
| Parent | Ms. A. (parent) | ✓ Agree |
| General education teacher | J. Carter | ✓ Agree |
| Special education teacher | R. Patel | ✓ Agree |
| District representative | L. Nguyen, assistant principal | ✓ Agree |
| Evaluation results | M. Reyes, LSSP | ✓ Agree |
16Key dates
| Services begin | 10/15/2026 |
|---|---|
| Progress reports | Every nine weeks, with report cards |
| Next annual ARD | On or before 10/13/2027 |
| Reevaluation due | On or before 09/22/2029 |
✓Compliance audit
Simulation with a fictional student and sample data, shown at reading speed. Agentic IEP produces a complete IEP draft in less than four minutes.
Every protected student, always in compliance.
Once the IEP is final, Agentic IEP keeps working. It watches every timeline for every student with an IEP, 504 plan or dyslexia services, prepares each required document before it is due, and turns each finalized IEP into specially designed instruction that teachers can deliver the next morning.
| Student | Program | Requirement | Timeline | Status |
|---|---|---|---|---|
| J.M. · Gr 3 | IDEA · SLD | Progress report (every 9 weeks) | due in 3 days | Done |
| M.G. · Gr 1 | Referral | FIE within 45 school days of consent | day 31 of 45 | Done |
| Student A · Gr 4 | IDEA · SLD | Prior Written Notice (5 school days before change) | due in 5 school days | Done |
| L.P. · Gr 5 | Section 504 | Annual 504 plan review | due in 9 days | Done |
| K.T. · Gr 10 | IDEA · SLD | Annual ARD (within 12 months) | due in 14 days | Done |
| D.S. · Gr 2 | Dyslexia · 504 | Accommodation implementation | logged daily | Done |
| A.R. · Gr 7 | IDEA · OHI | Reevaluation (every 3 years) | window opens in 20 days | Done |
- REED drafted for A.R.; consent request prepared
- Today’s accommodations logged for D.S. (4 of 4)
- Annual ARD scheduled and notice sent for K.T.
- 504 committee review scheduled for L.P.
- Prior Written Notice delivered on time for Student A
- FIE on track for M.G.: report due 11/18
- Progress report generated and sent to parent for J.M.
- REED drafted for A.R.; consent request prepared
- Today’s accommodations logged for D.S. (4 of 4)
Decode multisyllabic words with 85% accuracy
Given 20 unfamiliar multisyllabic words, apply syllable-division and phonics rules on 4 of 5 consecutive weekly probes.
A tutoring session with SammyTutor.
Same student, same IEP goals. Sammy greets Avery, builds the lesson around her love of science, listens as she reads aloud, and coaches her in the moment. Every answer flows straight into Sammy's memory, updates her IEP progress monitoring, and reaches her family with one tap.
- 01Greet & plan
- 02Word warm-up
- 03Read aloud
- 04Think it through
- 05Celebrate
- 06Update IEP progress
- 07Notify family
Hi, Avery!
Sammy remembers every step of your learning.
Volcano explorers
- Word warm-upIEP Goal 2
- Read aloud: Rivers of RockIEP Goal 1
- Think it throughComprehension
Volcano Reader badge earned
Avery finished today's mission.
Simulation with a fictional student and sample data. Turn on Sammy's voice to hear her speak (uses your device's built-in voice).
A learning companion who knows every child, teaches every standard, and remembers every step.
SammySchoolhouse is a companion platform for kids and grown ups. She tutors interactively, teaches every standard and unit, and keeps each child's goals front and center until mastery. Every piece of student work flows back into Sammy's memory, updating progress-monitoring graphs and documentation automatically. With one-tap actions throughout and "Speak with Sammy" on every page, learning is always one conversation away.
- Interactive tutoring across every standard and unit
- Family Dashboard, Learning Wall and IEP Parent Coach
- Quick checks, quizzes and thinking-skills assessments
- Wellbeing: feelings, calm space, diary and goals
- School hub with schedule, assignments and announcements
- Mastery Map and live progress insights
One child. One day. Every minute counts toward a goal.
Watch an example school day unfold. Every check-in, lesson and quick check flows into one learner record, updates the IEP goal graph on its own and reaches the family before dinner.
- Daily check-inWellbeing
- Math lessonLearn · TEKS 4.3C
- Quick checkAssessment
- Work becomes evidenceProgress monitoring
- Calm spaceWellbeing · Focus
- Family digestFamily
- Mood logged: Ready
- Fractions lesson, IEP goal 2
- Quick check: 4 of 5
Example student for illustration.
A complete school, in one companion.
Ten connected areas, one learner record and one log-in: instruction, assessment, wellbeing, family, school operations and progress monitoring, all guided by Sammy.
My Companion
Sammy, always one tap away- Speak with Sammy on every page
- Persistent memory of each learner
- Interest-built conversation
Learn
Every standard, every unit- Home, Explore and Path
- Tutor and Materials
- Learning Lab and Challenge Arena
- Standards and Thinking Areas
Assessments
Evidence, not opinion- Quick Checks and Quiz Me
- Mind Measure
- Thinking Skills
- Daily Check-In
Wellbeing
The whole child- Feelings and Calm Space
- Sensory and Focus
- Diary and Goals
- Reminders
Family
Home and school in sync- Family Dashboard
- Learning Wall
- IEP Parent Coach
School
A full school experience- School Hub
- Schedule and Assignments
- Announcements
Intelligence
What to do next- Mastery Map
- Action Board
- Next best actions
Progress
Growth you can see- Insights
- IEP goal progress monitoring
- Learning Profile
Records
One secure file- School File
- Devices
- Profile
System
The intelligence underneath- Memory and Kernel of Truth
- Architecture and Agentic Universe
- Innovation Flywheel
The intelligence behind SammySchoolhouse and SammyTutor.
Sammy is not a single tool. Every conversation, quiz and check-in runs through the full Agentic Learning platform: persistent memory, forecasting, a compounding growth flywheel and a person in command of every decision that matters.
Every learner, every need, every day.
How Sammy works
The end-to-end journey of a single learner signal. A child's action enters the intelligence cycle (collect, analyze, act) and returns as human-approved support. An intelligence cycle that keeps people in command.
Conversation · quiz answer · mood check-in
Learner signals · interactions · telemetry
Deterministic models + AI · mastery tracking · HDAIR diagnostics · forecast synthesis
Directives · next best actions · suggestions
Sammy prepares, a person decides
Persistent memory
How Sammy remembers a child over time. Interactions distill into durable memory: append-only signals feed a Bayesian belief model that shapes personalization. Nothing is a one-off; everything compounds.
Learner signals · conversations · lab telemetry · assessments · mood check-ins
Updating probabilities
Profile · preferences · shapes every future interaction
How Sammy thinks and forecasts
The living intelligence pipeline. New signals trigger pattern matching across sources; Sammy links them to memory and produces forecasts, surfacing next actions with confidence markers.
Trigger the pipeline
Match across sources · link to memory · filter and deduplicate
Forecast direction · leading strength · growth frontier
Digest and next actions
How Sammy grows, learns and improves
The growth flywheel: not a static loop but a compounding upward spiral. More time with Sammy produces a sharper model of the child, more targeted instruction and steadily advancing baselines.
Each Model, alone and together
The proprietary Models work as individually legible units, but their outputs interlock inside the AIMS core to form a single cohesive intelligence: deterministic rules combined with generative capacity.
HDAIR Model
Decision and intervention: five cognitive domains measured by Mind Measure
H-RSI
Recursive spiral intelligence: relational and emotional signals compounding in depth
HAICAM
Human-in-command: Sammy prepares, a person decides. Accountability and alignment
H-Theory SDI
Specially designed instruction generated directly from Learning Lab telemetry
AIMS
Tier semantics · provenance · concordance · consent
One truth engine. One memory system. One orchestration layer.
Sammy runs on an AI-native workflow architecture: a Master Knowledge Base of twelve specialized Knowledge Bases working as one. Tap any of them to explore what she is made of. The verb ring shows how Sammy thinks, end to end.
Memory Knowledge Base
Permissioned, long-term understanding of the learner: the story so far, told responsibly.
- Profile memory
- Milestones and timeline
- Preferences and values
- Evidence trails
The full spectrum of emergent bilingual needs, managed in one place and in full compliance.
Language Learner Companion manages every phase of the emergent bilingual and English learner process, from identification and placement through annual assessment decisions to reclassification and monitoring. It keeps every federal requirement and every agency rule on the clock, every student visible by name, and every teacher equipped with the instructional strategies each child needs.
- LPAC Builder for identification, placement and parental notification
- PreLas Companion and LAS Links Companion for assessment decisions
- Reclassification, program exit and post-exit monitoring
- Program Builder with fall, winter and spring cycles
- Status meters showing every student and every deadline
- Instructional strategies matched to each proficiency level
- Leader dashboards from campus to district
- Agency-specific Knowledge Bases for every edition
The full spectrum of language learner needs, in one place.
Every phase of the emergent bilingual and English learner process, every student group that needs protection, and every law that governs them, managed on one record with every deadline in view.
Beginning of year
Identify and place- Home language survey triggers
- PreLas and LAS Links screening
- LPAC placement decision
- Parental notification and approval
Throughout the year
Serve and monitor- Program and service delivery
- Progress monitoring by domain
- Instructional strategies by level
- Accommodations in every class
Middle of year
Decide and adjust- Assessment participation decisions
- Designated supports for testing
- Program adjustments
- Committee documentation
End of year
Review and exit- Annual proficiency review
- Reclassification criteria checked
- Program exit process
- Parent notification of exit
After exit
Monitor and support- Post-exit monitoring years
- Re-entry triggers
- Academic progress checks
- Longitudinal record
Every athlete's recruiting journey, organized, measured and moving forward.
Next Commit is a recruiting platform for high school, college and professional athletes. It brings an athlete's profile, stats, film and academics into one place, keeps every coach conversation, visit and offer on a single timeline, and shows the athlete and family what to do next to stay on track.
- Athlete profile with stats, measurables and highlight film
- Recruiting timeline with every contact, visit and offer
- Coach outreach tracking and follow-up reminders
- Academic and eligibility tracking alongside athletics
- Recruiting readiness score with next steps
- Family dashboard to follow progress together
44 waypoints. Every rule visible.
Follow one real pass through AIMS, the Agentic Intelligence Management System: four beats, five gates and one signal travelling the whole path. Nothing is hidden, and nothing skips a gate.
Reinvent and redefine your industry.
Is your company better equipped to sell products than build them?
We build world-class, AI workflow-driven apps across multiple industries for established companies with robust distribution verticals. Engagements are by selection.
Built by us
AI workflow and intelligence, engineered to a premium standard.
Made yours
Your brand and your workflows, delivered turnkey.
Taken to market by you
Your sales organization, your clients.
Selective
A limited number of engagements.
Proprietary
Built on our own models.
Turnkey
Delivered complete.
In market
Products in daily use.
The product that redefines your category.
We build for industry-leading companies that intend to reinvent and redefine their industry. Each product is created for turnkey lease or purchase.
AI workflow at the core
The product does the sequencing, drafting and checking that used to take your customers weeks.
Intelligence-powered
Learning intelligence, machine learning and prediction built in from the first screen.
Turnkey
Delivered complete, with the content inside, ready for your sales organization.
Under your name
Customized to your company and your market, and carried by your brand.
How an engagement runs.
Three stages, from the first working session to a finished product.
Discovery
We sit down with your essential company personnel to gather every requirement and the functions, features and outcomes you want.
Design
We work together on the branding, the colors and the workflows, until the product is unmistakably yours.
Standard
We ensure your app not only meets, but measurably exceeds, the top-functioning, market-owning product in your industry.
Consulting and development.
The full scope of an engagement.
Turnkey product builds
AI workflow and intelligence-powered apps, designed, engineered and delivered finished.
Requirements and discovery
Working sessions with your essential people to capture every function, feature and outcome.
Branding and customization
Your name, your colors and your workflows, built into the product.
Content development
The content that goes inside the app, so it arrives complete.
Competitive benchmarking
Measurement against the leading product in your market.
Visioning and consulting
Product-line strategy for firms deciding what to build next.
Advanced intelligence, put to work.
We apply the most advanced intelligence software available today, and every piece of it has a job to do inside the product.
Symbiotic learning apps
AI-driven apps that adapt to each user and sharpen with every interaction.
Prediction and forecasting
Models that show what is coming while there is still time to act on it.
Geometric spatial modeling
The modeling behind how our apps map progress and choose what comes next.
Multi-agent workflows
Reasoning software that sequences, drafts and checks the work, with every step traceable.
Our proprietary models, applied.
Three proprietary models sit beneath every product we build. Each one has a specific job inside it.
HDAIR Model
The product analyzes, iterates and scores its own work before a person ever sees it.
- Three candidate answers are drafted in parallel, each from a different angle.
- Each one is scored on five dimensions of reasoning.
- Each is refined on its own feedback, then all three are ranked on one index.
- The highest one that clears the pass mark goes forward. The others are kept as evidence.
HAICAM Model
A human gate sits on every consequential action: the product recommends, a person decides.
- Routine work runs on its own and is logged.
- Drafts wait until a person has reviewed them.
- Consequential decisions need a person on the way in and on the way out.
- A person can stop and reverse any run.
- Every release, sign-off and stop is written to the evidence ledger.
H-RSI Model
The product studies how it is used and raises its own baseline with every cycle.
- Each turn of the spiral is one cycle of real use.
- What the product learns in a cycle is proposed, and a person confirms it.
- The next cycle starts from the higher baseline.
How the Agentic Universe works.
The Agentic Universe is the multi-agent workflow system beneath every product we build. Picture it as rings around the person the work is for. The closer a ring sits to the center, the more authority it holds.
Every point is an agent with one job. Together they move as one system around the person at the center.
- 1A task arrivesThe person. It is anchored to one person’s record before any reasoning begins.
- 2A person authorizes itHuman authority clears the task before anything is composed.
- 3The orchestrator picks a teamOnly the agents this outcome needs are linked.
- 4The models set the methodHDAIR, HAICAM and H-RSI shape how the work is done.
- 5The Knowledge Bases supply what is knownWhat is already on record feeds the task.
- 6The specialist agents do the workEach one does its part in the domain field and passes it on.
- 7Verifiers check the resultAnything that does not hold is sent back.
- 8A person signs off and it is recordedThe decision and its evidence are written to the record.
How agents coordinate to reach the outcome you want.
Every task starts from the result a person wants.
Only the agents this outcome needs are linked.
Each one does its part and passes it on.
Anything that does not hold is sent back.
A person signs off, and the result is recorded.
How the Knowledge Bases feed every turn.
The H-RSI Model runs each task as a spiral, not a loop. On every turn the task draws on the Knowledge Bases, writes back what it learned, and starts the next turn from a higher baseline.
Perceive
What has just arrived, and what is already on record.
Analyze
How it measures up, and what follows from it.
Iterate
The draft is reworked and rebuilt until it holds.
Resolve
The result is checked, then signed off by a person.
Carry forward
The outcome is kept, so the next turn begins where this one ended.
Reasoning travels outward, from the models to the specialist agents. Evidence travels back inward, to the record. No route skips a ring.
1. Authorized
Nothing is composed, reasoned or executed until a person with authority clears it.
2. Disciplined
The team works one cycle: perceive, analyze, iterate, resolve. It repeats until the answer holds.
3. Validated
Verifier agents test the result. Anything unresolved goes back for another pass.
4. Recorded
The decision and its evidence are written to the record, so the next task starts from what is already known.
One intelligence beneath every product.
A multi-agent workflow system of 1,070 specialist agents in 139 squads across 25 domain constellations, orchestrated by the Agentic Learning Algo, with a human in command of every consequential decision.
The architecture stack
The same system as layers: super-agents and Models at the top, drilling down through the Agentic Learning Algo, the Platform Core, constellations, squads and agents.
AIMS Super-Agents & Models
led by H-Theory SDI · HDAIR Model · H-RSI · HAICAMAgentic Learning Algo
orchestration: runs the Models togetherPlatform Core
routing · memory · voice · safety · privacyDomain Constellations
curriculum · language · special education · law · care · moreSquads
each a companion or functionAgents
specialist roles running the workflowThe orchestration hub
How it is wired: the Models feed the Platform Core, which routes work to every domain constellation on its spokes, with data pulsing outward.
One mind, not many tools
Each proprietary Model threads through every constellation and converges on every agent, so the whole system reasons as one.
The cognitive core
Inside the nucleus: the AIMS super-agents that manage the whole system, and the proprietary Models, led by H-Theory SDI, that every other agent runs on. Squads shown as chips.
AIMS · Super-Agents
14The top tier of the system: super-agents that each own a whole domain and command the squads beneath them.
H-Theory SDI
16Proprietary theory of specially designed instruction, grounding the special education work in principle.
HDAIR Model
20Proprietary adaptive decision-and-intervention cycle: the reasoning loop at the heart of every companion.
H-RSI · Recursive Spiral Intelligence
20Proprietary recursive spiral Model: the system returns to ideas at ever-deeper levels and improves on each pass.
HAICAM
18Proprietary human-in-command Model: keeps human judgment, alignment and accountability central to every companion.
How the system expands
Growth radiates outward from the AIMS core in concentric stages: new agents spin up, new constellations bud and new domains layer on. Expansion is radial, not sprawling, and every new capability inherits the core's provenance and consent protocols.
Growth vectors
- Internal capacity grows through new agents.
- New use cases bud into constellations.
- Entire new domains layer on.
The living system
The whole architecture in motion: the Agentic Learning Algo at the core, the proprietary Models orbiting it with data flowing inward, and every constellation circling beyond. The system as one organism.
Scale, ranked
All 33 constellations ranked by agent count, bars growing into place: where the weight of the system actually sits.
Proportion by layer
The agents split into three layers, each constellation a tile sized to its share: a small orchestrating core above a vast field of domain work.
What our products share.
Agentic IEP, SammySchoolhouse, SammyTutor, Language Learner Companion and Next Commit serve very different markets. They are built the same way.
Workflow-driven
Each product is workflow-driven. It sequences, drafts and checks the work, and guides the user through it step by step.
A common engine
Every product runs on the Agentic Learning Algo, so each new one starts from proven ground.
Built by practitioners
Our specialists design for the real workday, so the product fits how its users already work.
Available turnkey.
Finished products available for lease or purchase, customized to your company and carried by your sales organization.
Agentic IEP
Proven in the field and ready to go to market under your name.
SammySchoolhouse
A companion for kids and grown-ups. She guides, supports and stays with the learner.
SammyTutor
Voice-first tutoring with Sammy, built around each learner's goals and interests.
Language Learner Companion
The full spectrum of emergent bilingual and English learner needs, in full compliance with federal and agency requirements.
Next Commit
A recruiting platform for high school, college and professional athletes.
Finished products, on every screen your customers use.
Delivered complete across laptop, tablet and phone, with the content already inside.
New intelligence-powered apps, built to lead their markets.
Agentic IEP, SammySchoolhouse, SammyTutor and Next Commit are the newest releases from Agentic Learning: AI-native, workflow-driven platforms engineered to markedly exceed every current competitor.
Agentic IEP, SammySchoolhouse and Next Commit are ready for full commercialization.
Exactly which agents do the work, and the law they follow.
Every Agentic IEP request is routed to a dedicated special education constellation, handled by specialist agents and checked against the curriculum and the full body of law before a person signs off.
One request's journey
A single IEP request travelling the pipeline: down through the Models and the orchestrator to the exact agents that do the work, ending at human sign-off.
"Draft an IEP from this evaluation."
H-Theory SDI grounds the SDI · HAICAM sets a human gate
Routes to the right constellation
Special Education · IEP Companion
Goals and objectives
PLAAFP Writer › Goal Writer › Objective Sequencer
A compliant draft, awaiting human sign-off
Curriculum and compliance
The layer that anchors the system to Texas classrooms: leading open curriculum and standards alignment, and the full body of federal and Texas law that governs children's education.
Curriculum & Open Content Alignment
24Accesses and aligns to leading free, public and open curriculum, the high-quality materials Texas districts prefer, and maps every lesson to the standards that matter.
Legislation & Title Funding · Federal & Texas
32Keeps every plan and program compliant with the federal and Texas laws that govern children's education: Title-funded programs, disability law, and the privacy and safety statutes.
From anchor to goal: the S·M·A·R·T engine
Tap each letter and watch the goal light up, piece by piece. Every goal Agentic IEP writes is built this way, from the present levels outward.
Given a randomly selected grade-level passage, Mateo will read aloud at 112 words correct per minute with at least 97% accuracy by the annual review date, measured twice monthly against the aimline.
Specific: the learner, the behavior and the conditions are named outright. No vague verbs survive review.
Objectives beneath this goal are sequenced as incrementally progressive, clinically coherent benchmarks: stepping stones, not paperwork.
Every function, every product.
A complete reference to what each product does, in one place.
| Area | Function | What it does |
|---|---|---|
| Workspaces | Dashboard | Caseload status, deadlines and alerts at a glance |
| Workspaces | District Command Center | District-wide quality, compliance and meeting readiness |
| Workspaces | Assessment Center | Evaluation data intake, extraction and source tracing |
| Compliance | Compliance Trace | Every timeline for every protected student, including EB/TELPAS records |
| Compliance | Procedural Compliance | Notices, consents, safeguards and procedural steps |
| Compliance | Prior Written Notice | Every required element, in English and Spanish |
| Writing | IEP Draft Creator | 13-section draft plus compliance audit in less than four minutes |
| Writing | Agentic 504 | Complete 504 plan draft in less than one minute |
| Writing | Baseline Data Creators | Measurable baselines from evaluation and classroom data |
| Instruction | Supports Library | Accommodations, modifications and designated supports |
| Instruction | Resource Kits | Aligned specially designed instruction for each goal |
| Monitoring | Progress Monitoring | Goal graphs, aimlines and nine-week progress reports |
| Monitoring | Daily Tracking | Service minutes and accommodation implementation |
| Monitoring | Quick Tap | One-tap data capture during instruction |
| Planning | Planner | Meeting, service and teacher schedules |
| Planning | Caseload | Every student, every goal, every due date |
| Family | Family Comm | Clear, considered family communication |
| AI | AI Assistant Studio | 14 built-in tools plus Create Custom Tool |
| Standards | 50-state alignment | National and state-specific standards and regulations |
| Security | Zero-retention architecture | No boilerplate; all text original to each student |
| Area | Function | What it does |
|---|---|---|
| My Companion | Speak with Sammy | Voice conversation with Sammy on every page |
| Family | Family Dashboard | Schedules, reminders and progress in one place |
| Family | Learning Wall | Celebrates work and growth at home |
| Family | IEP Parent Coach | Helps families understand and support IEP goals |
| Learn | Path, Tutor, Explore | Personalized path through every standard and unit |
| Learn | Learning Lab & Challenge Arena | Hands-on activities, challenges and games |
| Learn | Standards & Thinking Areas | Standards coverage and thinking-skills development |
| Assessments | Quick Checks & Quiz Me | Fast formative checks with instant feedback |
| Assessments | Mind Measure & Thinking Skills | Cognitive domains measured and tracked |
| Wellbeing | Feelings & Check-In | Mood check-ins Sammy listens to |
| Wellbeing | Calm Space, Sensory & Focus | Self-regulation tools and focus sessions |
| Wellbeing | Diary, Goals & Reminders | Personal goals, journaling and routines |
| School | School Hub & Schedule | Assignments, announcements and timetable |
| Intelligence | Mastery Map & Action Board | Mastery by skill and next best actions |
| Progress | Insights & Progress Monitoring | IEP goal graphs updated from student work |
| Progress | Learning Profile | Strengths, needs, interests and preferences |
| Records | School File & Devices | Secure records and connected devices |
| System | Memory & Kernel of Truth | Permissioned long-term memory of the learner |
| Compliance | Texas program requirements | Built for homeschool, microschool, private and online school |
| Area | Function | What it does |
|---|---|---|
| Tutoring | Speak now with Sammy | Fast, voice-first tutoring conversation |
| Tutoring | Every standard, every unit | Interactive instruction across subjects |
| Tutoring | Hints, not answers | Adaptive hints when a learner hesitates |
| Tutoring | Visual models | Diagrams and manipulatives inside the conversation |
| Reading | Read-aloud listening | Words correct per minute and accuracy, live |
| Reading | Syllable and decoding practice | Structured-literacy word work |
| Adaptivity | Interest-built lessons | Lessons themed on what the child loves |
| Adaptivity | Leveled passages | Text matched to instructional level |
| Adaptivity | Next-day planning | Missed words carried into tomorrow's warm-up |
| Motivation | XP, streaks and badges | Rewards that keep learners coming back |
| Progress | IEP goal updates | Work flows into goal graphs automatically |
| Progress | Mastery Map | Skill-by-skill mastery view |
| Family | One-tap family updates | Progress and recordings sent home |
| Safety | HAICAM human gate | A person stays in command of decisions |
| Area | Function | What it does |
|---|---|---|
| Profile | Athlete profile | Stats, measurables and highlight film in one place |
| Profile | Academics | GPA and eligibility tracked alongside athletics |
| Recruiting | Recruiting timeline | Every contact, visit and offer on one timeline |
| Recruiting | Coach outreach | Outreach tracking and follow-up reminders |
| Recruiting | Readiness score | Recruiting readiness with next steps |
| Analytics | Film and profile views | Who is watching and how often |
| Family | Family dashboard | Athletes and families follow progress together |
| Levels | High school, college, pro | One platform across every level |
| Area | Function | What it does |
|---|---|---|
| Platform | Student information system | Enrollment, demographics, attendance, scheduling, grades, transcripts |
| Platform | Learning management system | Courses, content, assignments, grading, gradebook |
| Platform | Benchmark & interim assessment | Adaptive benchmarks with RIT-compatible scale scores |
| Platform | Universal screening | Fall, winter and spring screeners with risk flags |
| Platform | Curriculum-based measurement | Probes, aimlines and decision rules |
| Platform | Progress monitoring | Real-time goal graphs for every student |
| Platform | Adaptive practice | Personalized practice in reading and math |
| Platform | AI tutoring | Voice-first tutoring through Sammy |
| Platform | Special education management | Agentic IEP: evaluations, ARDs, IEPs, timelines |
| Platform | Section 504 management | 504 plans, reviews and accommodation logs |
| Platform | IEP goals & SDI library | Goals, objectives and specially designed instruction |
| Platform | English learner program | Language Learner Companion: identification, LPAC, monitoring |
| Platform | Dyslexia services | Screening, identification and services tracking |
| Platform | Gifted & talented | Identification and service tracking |
| Platform | MTSS / RTI | Tiers, interventions and decision rules |
| Platform | Behavior & discipline | Incidents, interventions and reporting |
| Platform | Social-emotional learning | Check-ins, self-regulation and wellbeing |
| Platform | Family communication | Messaging, digests and home-language updates |
| Platform | Compliance & state reporting | Audit-grade timelines, Texas reporting |
| Platform | Medicaid (SHARS) billing | Service logging, claims and denial prevention |
| Platform | Analytics & dashboards | District, campus and teacher intelligence |
| Platform | Professional development | On-demand training and CPE tracking |
| Platform | Data integration | LTI 1.3, OneRoster and Ed-Fi connections |
| Platform | Records & documents | Secure student files and evidence trails |
| Platform | Security & privacy | Role-based access, consent and full audit trail |
| Area | Function | What it does |
|---|---|---|
| Identification | LPAC Builder | Newly enrolled identification and placement |
| Identification | Home language survey | Survey intake and screening triggers |
| Notification | Parental notification | Notices in the family's home language, on time |
| Assessment | PreLas Companion | Initial proficiency screening and decisions |
| Assessment | LAS Links Companion | Annual proficiency results and progress monitoring |
| Decisions | Middle-of-year assessment decisions | Committee decisions documented and tracked |
| Exit | Reclassification and program exit | End-of-year review and process |
| Exit | Post-exit monitoring | Monitoring years tracked for every exited student |
| Programs | Program Builder | Fall, winter and spring program cycles |
| Instruction | Instructional strategies | Strategies matched to each proficiency level |
| Compliance | Status meters | Every student by name with days on the clock |
| Compliance | Agency Knowledge Bases | Each edition driven by its own rules, clocks and funding |
| Leadership | Leader dashboards | Campus and district views of every timeline |
| Records | Student portfolio | Every form, decision and data source in one record |
Proven in the field: the Houston ISD pilot.
Findings from the spring 2026 pilot of workflow-driven IEP authoring inside Houston ISD's S1S platform, March to May 2026, as Phase 1 of the HISD Special Education Automation Project. Houston ISD went live district-wide on April 16, 2026, and purchased the system for 276 schools serving about 29,000 students with IEPs.
Executive summary
Each Agentic IEP draft was scored against a legacy IEP written from the same Full Individual Evaluation, on a single rubric anchored in IDEA, Endrew F., Texas special education rules and the Standards for Educational and Psychological Testing.
| Indicator | Agentic IEP | Legacy IEPs |
|---|---|---|
| Overall quality rating (of 100), pilot average | 94.0 | 62.0 |
| Adjusted quality rating, final four weeks | 97.8 | 62.0 |
| Score references with standard score and percentile | 98% | 54% |
| Score references with a confidence interval | 89% | 11% |
| Evaluation-documented needs with a matching goal | 93% | 41% |
| Template language in IEP text (lower is better) | Less than 1% | 61% |
| Fabricated psychometric data found | None | n/a |
| Time for a complete 621-field draft | Less than 4 minutes | n/a |
Where the pilot came from
Houston ISD's technology leadership proposed a proof of concept and launched the HISD Special Education Automation Project, with IEP Companion as Phase 1. Each district requirement maps to a pilot result.
| What the district asked for (October 2025) | What the pilot delivered (spring 2026) |
|---|---|
| Smart logic that generates compliant, customized IEPs | Adjusted quality rating of 97.8 against 62.0 for legacy IEPs; less than 1% template language |
| Real-time validation against federal, state and district rules | Every draft checked against IDEA 2004, 34 C.F.R. Part 300 and 19 TAC Chapter 89, with 23 automatic checks added during the pilot |
| Seamless data pull from district systems | Ran as the IEP writing engine inside S1S, cross-checked field by field against the S1S record |
| Output aligned with state reporting requirements | A complete draft in the district's 621-field IEP format in less than four minutes |
| A fully guided, step-by-step workflow | Guided drafting, with every draft reviewed, changed and adopted by the ARD committee |
| User interface built on the S1S platform | Delivered inside S1S, the district's IEP platform |
| Implementation as early as spring 2026 | Educator testing opened March 18, 2026; district-wide go-live April 16, 2026 |
Study design
A matched comparison inside a live district deployment.
How drafts were scored
The four rubric dimensions
| Dimension | What it measures | Anchored in |
|---|---|---|
| Educational accuracy | Goals, services and accommodations target the needs the evaluation documents | Endrew F.; IDEA §1414(d) |
| Data quality | Scores cited with standard score, percentile, confidence interval and date; a traceable chain from evaluation to goal | AERA/APA/NCME Standards (2014); IDEA §1414(b) |
| Individualization | The IEP describes this student, not a template | 34 C.F.R. §300.320 |
| Legal and psychometric standards | Meets IDEA 2004 and 19 TAC Chapter 89; instruments used and interpreted as publishers intend | IDEA 2004; 19 TAC §89.1050 |
Findings
Quality rose to an adjusted 97.8
Drafts averaged 94.0 of 100 across the pilot against 62.0 for legacy IEPs, and reached an adjusted 97.8 in the final four weeks. Template language made up less than 1% of draft text, against 61% for legacy IEPs.
Same student, same evaluation
In matched pairs written from the same evaluation, Agentic IEP scored 96 against 64 for the district IEP, and its lead grew from 16 to 32 points after the March system changes. It also outscored a 93-point hand-written reference IEP that had a fuller evaluation behind it. Where the district IEP was stronger, for example a full nursing health care plan, the report says so.
Every educator issue answered within days
The first written round of educator feedback named 11 distinct issues, 4 of them critical. Each had a matching system change within a week: 43 changes in all.
No fabricated data
No fabricated psychometric values were found in extraction, analysis or reporting. Standard score and percentile appeared in 98% of score references and confidence intervals in 89%, against 54% and 11% for legacy IEPs.
Effect sizes
Converting results into standard effect-size metrics makes the size of each difference comparable. Every value is calculated directly from the reported figures. Benchmarks for Cohen's h: 0.2 small, 0.5 medium, 0.8 large.
| Indicator | Agentic IEP | Legacy | Cohen's h | Odds ratio | Omission ratio |
|---|---|---|---|---|---|
| Standard score and percentile reported | 98% | 54% | 1.21 | 41.7 | 23.0× |
| Confidence interval reported | 89% | 11% | 1.79 | 65.5 | 8.1× |
| Needs with a matching goal | 93% | 41% | 1.22 | 19.1 | 8.4× |
| Template language (lower is better) | <1% | 61% | 1.59 or more | n/a | n/a |
Odds ratio = [p₁ / (1 − p₁)] ÷ [p₂ / (1 − p₂)]. Omission ratio = how many times as often legacy IEPs left the element out. All four indicators exceed the threshold for a large effect. Goal coverage, the link from evaluation to program, reaches h = 1.22: legacy IEPs left an evaluation-documented need without a goal 8.4 times as often.
Share of the gap to a perfect score closed
Gap closed = (Agentic IEP − legacy) ÷ (maximum − legacy). By the final four weeks, the distance between the legacy IEP and a perfect score had shrunk to less than one-seventeenth of its original size.
Improvement within the pilot
| Measure | Earlier | Later | Change |
|---|---|---|---|
| Quality rating (pilot-wide composite → final four weeks) | 94.0 | 97.8 | +3.8 points |
| Distance to a perfect score | 6.0 | 2.2 | −63% |
| Lead over the district IEP, matched pairs (before → after the March fixes) | +16 | +32 | +100% |
| Lead over legacy IEPs (composite → final four weeks) | +32.0 | +35.8 | +12% |
Responsiveness and throughput
Educators shaped the results: each round of feedback became a tracked change.
| Measure | Value |
|---|---|
| Educator issues resolved within a week | 11 of 11 (100%) |
| Critical issues resolved within a week | 4 of 4 (100%) |
| System changes per educator issue | 3.9 (43 changes ÷ 11 issues) |
| Fields completed per minute of drafting | 155 or more |
| Fabricated psychometric values found | 0 |
The pilot at a glance
Agentic IEP ran as the IEP writing engine inside S1S, the district's IEP platform, during a 36-school pilot that opened March 18, 2026.
| District | Houston ISD |
| Where it ran | S1S, the district IEP platform, with Agentic IEP writing the IEP draft |
| Pilot scope | 36 schools; special education case managers and specialists |
| Pilot window | Code freeze March 17; user acceptance testing from March 18; findings through May 2026 |
| District scale after the pilot | 276 schools, about 29,000 students with IEPs |
| Output | A complete draft in the district's 621-field IEP format, in less than four minutes |
| Rules every draft is checked against | IDEA 2004, 34 C.F.R. Part 300, 19 TAC Chapter 89, Endrew F. |
How the pilot ran
A tight test-and-fix loop: district reviewers judged real drafts, and every problem they named became a tracked change, retested before the next round.
District testing first
The district cross-checked output against its own IEP and the S1S record, field by field. Changes were frozen at 8 a.m. on March 17 so the district could run final tests before educators joined.
Educators review real drafts
User acceptance testing opened. Special education reviewers, including case managers and a speech-language pathologist, ran drafts and marked up each section by hand.
Every comment sorted and ranked
The first round of handwritten feedback was transcribed the same morning, checked line by line against the originals, and ranked critical, high, medium or enhancement.
One ticket per change
Each fix was logged as its own ticket, so the district could see who owned it and when it closed.
Fix, retest, show the result
The updated system produced a full exemplar IEP in the test environment, shared with the district alongside a running accountability table of every item and its status.
Specialists set the guardrails
A short survey went to the district's special education specialists so their own practice could shape the rules the system writes by.
Measure quality against human IEPs
Drafts were scored side by side with IEPs written by district staff from the same evaluation, using one rubric for both.
Side by side: the same student, two authors
A matched pair written from the same evaluation. The district's hand-written IEP scored 64; the Agentic IEP draft scored 96, higher even than a 93-point reference IEP written from a fuller evaluation.
Goals give only "Percent Accuracy."
Every goal names its method, schedule and mastery rule.
Most measures start at 1%, though its own present levels say the student decodes CVC words with good accuracy.
Baselines drawn from the evaluation and consistent with present levels.
Parallel sub-skills, each starting from the same 1% baseline.
Objectives step support down toward independence.
A math goal targets quantities to 50 for a student its own present levels place at 1 to 20.
Targets set from the student's measured level.
Two checkbox phrases with no evidence behind them.
Cites the teacher's social ratings, the parent's report and the speech-language recommendations, then states which factor actually drives the placement decision.
What the pilot surfaced, and how it was resolved
The pilot did its job: it found the weak spots early, and each one was traced to a cause and fixed rather than worked around.
| When | What surfaced | Cause | Resolution |
|---|---|---|---|
| Early March | A draft left the service table blank, skipped transition services for a 15-year-old and left related services empty despite speech eligibility | Age and eligibility triggers not yet wired to those sections | Transition flags at 14 and 18, service minutes required, related services tied to eligibility |
| March 10 | Assessment accommodations returned a sentence instead of the district's list | Output field shape did not match the district format | Corrected the same day |
| March 19 | Four critical and seven other educator issues | Goal format, extraction and language rules not yet matched to district practice | 14 updates, 23 new automatic checks, 6 new input fields |
| March 31 | One draft scored 72 despite the strongest present-levels section yet | Arithmetic errors in goals: mastery percentages, service minutes (300 stated where frequency and duration gave 150), one objective above its own goal | Calculation rules corrected; new release April 1 and retested |
| April 3 | IEP runs paused for about two hours | Model account credit ran out; automatic top-up failed | Second account administrator added the same day |
| May | Goal coverage thinner than the evaluation supports in complex multi-disability cases | Goals not yet written for every need when needs span many areas | Named the top development priority for the next version |
What the results mean for assessment
The IEP becomes the place where an evaluation does its most consequential work: each score arrives intact, interpreted and tied to a goal, instead of being summarized away.
Scores read the way the test was built
Standard score and percentile in 98% of score references and confidence intervals in 89%, against 54% and 11% in hand-written IEPs. Scores are cited with the instrument and the date it was given.
Interpretive cautions carried through
In the reference case, the system flagged that a documented fine-motor impairment likely depressed KTEA-3 Math Computation and Written Expression, and carried that caution into present levels.
Every subtest has a downstream use
93% of needs documented in the evaluation had a matching goal, against 41% in legacy IEPs. A score that drives a goal is a score a district values.
A full clinical battery, used in full
The reference evaluation drew on WPPSI-IV, KTEA-3, CELF-5, GFTA-3, PPVT-4 and EVT-2. A more complete battery gives the IEP more to work with.
One district, 29,000 IEPs a year
Each built on an evaluation record, with a reevaluation at least every three years.
Built by measurement people
Psychometric conventions are written into the rules every draft must pass.
Outcome and next steps
Method note. Quality ratings were scored by the development team. Effect sizes are descriptive and computed from reported proportions; standard errors and confidence intervals will be reported with the formal reliability study.
Proposed next steps
Close the goal-coverage gap
In complex multi-disability cases, the one area where drafts still trailed the evidence.
Score a larger sample
District IEPs scored at the end of the 2026–27 school year with the same rubric, to confirm the 97.8 rating across the full district.
Track student outcomes
Goal progress and ARD agreement rates for students whose IEPs were drafted with the system, not only document quality.
Measure time saved
Per IEP, for case managers, against a recorded baseline.
Run a formal reliability study
Blind scoring of matched pairs by two or more trained raters, reported with inter-rater agreement and effect sizes for each dimension.
Better than your competition, by a significant margin.
We will only build industry-leading, intelligence-powered, analytics-driven apps that markedly exceed all current competitors. We are motivated by revenue, so our approach is simple: we will be better than your competition by a significant margin. That is our guarantee.
Straightforward and simple.
Our terms are straightforward and simple. We discuss them directly, and privately, with each firm we work with.
A boutique of specialists.
Highly specialized, highly trained professionals. We take on a limited number of engagements, and we are as selective as the firms who work with us.
Neuropsychology
How learners think, attend and remember, built into the product.
Psychometrics
Measurement science applied to test development.
Predictive analytics
Forecasting systems that put the next decision in front of the user.
Deep learning & teaching systems
Systems designed around how teaching actually works.
Cognition & learning science
Research-grounded design in every workflow.
Practitioner experience
Hands-on experience doing the work our products are built for.
Advanced forecasting
Models that project growth, risk and outcomes before they happen, so decisions are made early.
Built for the rules your customers work under.
Your customers are trusted with sensitive records. The products we build for you are designed like it.
Privacy first
Data is always protected, minimized, and used only for the purpose it was collected.
Secure by design
Enterprise-grade security and compliance, with a full audit trail on every record.
Legally aligned
Built to the laws and regulatory requirements of your market.
Always evolving
Continuous improvement driven by research and outcomes.
Bring the next great product to your market.
Six intelligence-powered apps, one platform underneath, and a team that builds to exceed the best product in your industry.
When quality and security matter, work with Agentic Learning.
We work with a select group of firms. Tell us about your company and your market, and we will tell you whether we are the right fit.