How It Works
Diagnose with peer-matched data. Prescribe evidence-based strategies. Equip teachers with a configured assistant. Six professional learning sessions across one full instructional cycle. Measure every quarter. The district owns it all.
Massachusetts districts face a familiar bind: MCAS scores still 8–12 points below pre-pandemic levels, an AI literacy requirement on the horizon with no implementation support, and professional development budgets stretched across competing priorities. The research on conventional PD is not encouraging — TNTP (2015) found only about three in ten teachers improved across the programs they studied, and nothing predicted who would. GAO (2026) confirmed the evidence is still mixed. The instructional strategies themselves are sound. The problem is getting them into daily practice.
The planning assistant changes that equation. A teacher who asks for a discussion protocol gets one built around the standards her diagnostic identified as weak — without having to specify that. The strategies show up every time she uses the tool, because they're built in. Coaching and community sessions build on what is already happening in her classroom, not on what a workshop covered months ago.
Phase 1
Benchmark your district against 5 comparable peers. Built from 9 years of MCAS data across 325 Massachusetts districts.
Matched on six dimensions: enrollment, economics, English learner population, high-needs population, per-pupil spending, and student-teacher ratio. Not statewide averages — operationally comparable peers. A district that looks fine by state standards may be 15-30 points below similar districts.
Not just "math is low." Which grade? Which strand? Where exactly are students losing ground — and where are they outperforming peers?
Every metric broken down by race, income, EL status, disability, and intersections. The question is always: for whom?
Strengths identified as deliberately as gaps. Where is the district above peers? What's working and how do we build on it?
This district appeared adequate by conventional measures — graduation rates above state average, stable teaching force, engaged families.
| Finding | Detail |
|---|---|
| At or above state average | 18 of 19 tested grades and subjects |
| Below peer average | 16 of the same 19 |
| Gaps exceeding 15 points | 8 grade-subject areas, up to 28 points below peers |
| Bright spots | 3 areas at or above peers — practices to learn from |
The diagnostic is free. Submit a request or email mluekens@directed.pro. I may follow up to confirm the peer group, and the snapshot is delivered within a few days as a branded report professional enough for a cabinet or school committee. No commitment. If the findings are useful, the conversation continues. If not, the district keeps the portrait. Built entirely from public DESE data — no student records collected or stored. See a sample →
Phase 2
The diagnostic produces a strategy prescription. The prescription drives everything downstream — how the planning assistant is configured, what coaching focuses on, and what gets measured.
Not a generic list of "best practices." The highest-priority strategies for your district, selected because your data shows these are the gaps where these strategies produce the largest effects. Two districts with different diagnostics get different prescriptions, different assistant configurations, different coaching sequences, different measurement targets. One principle applies everywhere: scaffolds give students access to grade-level work, not easier work.
Built on your adopted curriculum, not around it. The assistant starts from the lessons in your district's adopted programs and supports them with scaffolds, checks for understanding and discussion questions. It doesn't write replacement units. Professional development organized around curriculum materials has some of the strongest evidence in the field (Lynch et al., 2019).
A dedicated professional partner for every participating teacher — configured in advance with curriculum, standards, diagnostic priorities, and structured workflows. It supports teachers' preparation and paperwork; it does not work with students or take on any part of a paraprofessional's role.
MA Framework standards, MCAS architecture, workflow library (differentiation, formative analysis, lesson planning, parent comms, IEP notes), pedagogical knowledge base. Built once.
Scope and sequence, unit maps, standards crosswalks, common misconceptions. The assistant supports your adopted lessons rather than replacing them. "I'm starting Unit 5 Monday" — the assistant already knows what Unit 5 contains.
Diagnostic priorities, strategy prescription, school context. The assistant's defaults reflect where students need the most help — built in, not prompted.
Classes, students, preferences, curriculum position. Completed on Build Day, grows through use. Remembers across sessions.
The intelligence is in the prebuild, not in the prompt. A teacher says "I need practice for tomorrow's lesson" and is designed to get differentiated materials aligned to the right standards in a couple of minutes — because the curriculum and diagnostic priorities are already built in. The entry point is always burden reduction. Draft the parent email in 90 seconds. Sort yesterday's exit tickets by misconception. Then build from there.
The assistant is not a separate product. It is built inside the AI platform the district already uses or chooses (Claude, ChatGPT, Gemini, or an education platform like MagicSchool). The district owns the configuration and knowledge files, and they can be moved to another platform if the district changes tools.
In Practice
The effect shows up because the teacher directing the tool is skilled, not because the tool is sophisticated.
A teacher whose students are struggling with proportional reasoning has her assistant produce three real-world scenarios at different levels of complexity, each requiring proportional reasoning and each containing a deliberate error for students to find. She builds a structured discussion around them. The logistics of differentiation are already handled, so her attention goes to the discussion.
A teacher uses her assistant to analyze student drafts against her rubric, producing specific, actionable feedback for each student: "Your third paragraph claims community support but gives no evidence. What data would strengthen it?" She reviews and edits every comment before it reaches a student, and she sees patterns across the class that shape her next mini-lesson.
A teacher planning a force-and-motion investigation has her assistant produce inquiry prompts for each lab group: questions, not answers. "What variable are you testing? What are you holding constant? What result would disprove your hypothesis?" She uses them on cards and in conversation, and spends her time with groups on experimental design.
Build Day
One day. Teachers walk in, meet an assistant that already knows their curriculum and diagnostic priorities, customize it for their own classes and students, and walk out with materials for Monday. Everything they build is for their real classes. Build Day is organized by school level so content stays focused. Larger districts can split Build Day into more than one session.
Every minute is designed so teachers leave with something they can use the next morning. If an activity doesn't serve that purpose, it doesn't belong in the day.
Morning: Live demo across subjects — no slides, real output, real time. Then five tasks using the teacher's own content for their own classes. The first thing they do is the admin task they most want off their plate. Relief first, then capability.
Afternoon: Personalization — classes, students, preferences. Department-specific deep dives. The diagnostic data, presented after hands-on experience so it explains rather than threatens. Working group formation. 30-day challenge. Baseline survey.
What teachers leave with: A working assistant that knows their curriculum and students. Materials they will actually use next week. Five or more practiced workflows they can repeat independently. Confidence that the assistant works across all their preps. And the beginning of the director skill that develops across the engagement.
The Progression
Teachers move through five observable levels of capability over the course of the engagement. Nearly everyone starts as a user of any new tool. The point is not where teachers begin but how far the work helps them go. These levels are never used in evaluation.
Simple requests, accepting what comes back. "Give me a quiz on Chapter 3." Because the assistant is already configured, even a simple request is designed to return material aligned to the right unit and standards. Build Day starts here. Build Day onward
Teachers provide context: the class, the unit, the range of students, the level of thinking required. "I need a warm-up for Period 3 on slope-intercept form. They have a wide range; some still struggle with plotting points." Weeks 2–6
Teachers judge output against professional standards and push back: "This version lowers the rigor instead of scaffolding access." "The exit ticket is at DOK 1, but I taught to DOK 3." Often prompted by seeing colleagues' approaches in coaching and community sessions. Sessions 2–3
Teachers connect formative data to the next instructional decision on their own: "Last week's exit ticket showed 40% of Period 3 hadn't mastered equivalent fractions. Build scaffolded practice on that before I move to Unit 4." They build their own recurring workflows. Sessions 3–5
Teachers use the assistant as a thinking partner: "I want a three-week inquiry sequence on evolution that moves from observation to hypothesis to evidence to argument, with each day building on the last. Let's plan it together." These teachers become the internal leaders who sustain the work. Sessions 5–6+
Professional Learning
Six professional learning sessions across the engagement where teachers deepen their practice, share what's working, and build toward district independence. The sequence is driven by the strategy prescription.
Coaching runs in grade-level or department teams of four to six teachers working on the same prescribed practice: a team planning conversation, brief observations focused on that practice, and a team debrief. The district's instructional coaches and working-group members lead the cycles after the first round. DirectEd trains them, models a first cycle with each coach, and observes and gives feedback on their coaching across the engagement. All observation data is kept entirely separate from educator evaluation.
About an hour where teachers share what's working, troubleshoot what isn't, and see one another's approaches. Early sessions support teachers at Levels 1 and 2; later sessions develop Levels 3 and 4. The working group co-facilitates by session 2, leads by session 4, and runs the work independently by the end of the engagement.
Phases 3 + 4: Measure + Adjust
Baseline on Build Day. 30-day adoption check. 90-day practice change indicators. Short time-use logs at baseline, midpoint, and end of engagement. A quarterly audit of a sample of assistant output for accuracy, alignment, rigor and curriculum fit. Quarterly reviews connecting investment to practice change to student outcomes, disaggregated by subgroup. Impact report comparing against diagnostic baseline and peer districts. The measurement plan includes a comparison group — a staggered start across schools or grades, or a comparative interrupted time series against the peer set.
Targets are agreed with the district in advance. A portion of the fee is tied to implementation targets. At the midpoint review, the district may end the engagement for any reason with no further fees. The evaluation design is pre-registered before work begins, so results are reported against a plan fixed in advance. When a school committee member asks "what did we get for that investment?" — the answer is documented, specific, and transparent.
Shared Responsibility
Every engagement is designed so the district carries the work and DirectEd builds its capacity to do it.
| Work | DirectEd | District |
|---|---|---|
| Diagnostic and prescription | Builds the diagnostic; drafts the prescription with district leaders | Provides data and context; approves the prescription |
| Planning assistants | Builds the first version with the working group; trains them to maintain it | Owns and maintains the configuration |
| Build Day | Designs and leads it, with working-group members co-facilitating | Provides the PD day |
| Coaching cycles | Trains coaches; models a first cycle; observes and gives feedback | Coaches and working group lead team cycles after the first round |
| Community of practice | Supplies the sequence; co-facilitates early sessions | Working group leads by session 4 |
| Measurement | Supplies protocols and templates; trains reviewers; analyzes early quarterly data | Collects observations, work samples, time-use logs and output reviews |
| Quarterly reviews | Facilitates the first reviews | Leads reviews by the second half of the engagement |
Privacy & Responsible Use
The assistant is built from curriculum, standards, and district-level diagnostic results. When teachers use it, students are identified by first name, initials, or number — never by full name, student ID, or other identifying record. Work that involves individual student records — such as finalizing IEP progress reports or grades — stays in the district's own systems.
Only education or enterprise accounts covered by a signed Massachusetts Student Data Privacy Agreement are used, consistent with FERPA, the Massachusetts student records regulations (603 CMR 23.00), and DESE's AI guidance. Vendors under those agreements may not use student data to train their models.
Teachers review everything the assistant produces before it reaches a student. Build Day includes practice in catching common failure modes: invented facts, misaligned standards, arithmetic errors, and tasks that quietly lose rigor. The engagement provides a template staff acceptable-use policy aligned with DESE's AI guidance. Whether and how students use AI directly remains a separate district decision.
The planning assistants and everything that configures them belong to the district — the configuration files, the curriculum knowledge base, the workflow library. Teachers keep their developed director skills. The working group leads coaching and community sessions independently. Nothing is licensed, nothing expires, nothing requires a renewal. Districts that continue into a second year receive an updated diagnostic and revised prescription. Districts that don't continue still keep everything they built.
Engagements are structured to remain under $100,000, below the Massachusetts Chapter 30B simplified procurement threshold. Where federal Title II-A funds are used, federal procurement rules (2 CFR 200) apply instead.
Questions Districts Ask
No. Nothing in this framework reduces teaching or paraprofessional positions. The time the assistant saves is supposed to go back into teaching — and the measurement system checks whether it actually does. Every engagement's terms, including the commitment to keep all observation data out of evaluation, get worked out with the local association before any work starts.
One Build Day, held on a PD day the district already has. Six professional learning sessions across the engagement — team coaching cycles of about an hour to an hour and a half, during common planning time wherever possible, plus community sessions of about an hour inside existing meeting time. And three one-week time-use logs over the course of the engagement. What teachers get back is time, starting in the first week.
No. The assistant is built on the curriculum the district has already adopted. It supports lessons with scaffolds, checks for understanding and discussion questions — it doesn't write replacement units. Part of each quarterly review looks at whether supplementary material is drifting away from the curriculum. If your district has invested in high-quality instructional materials, think of this as a way to get more out of that investment.
The assistant is built from curriculum, standards and district-level results. Students are identified by first name, initials, or a number — never by full name, student ID, or any other identifying record. Only accounts covered by a signed Massachusetts Student Data Privacy Agreement are used, and vendors under those agreements may not use student data to train their models.
Yes, and the skepticism is earned. Three things are different here. First, AI isn't the active ingredient — the instructional practices are. AI makes them practical. Second, measurement is built in from the start, with a comparison group and a publicly registered evaluation plan. Third, I'm clear about what's proven and what isn't. The teaching practices are well established. Using AI to make them sustainable is the part being tested. Overpromising is how education technology lost teachers' trust, and I don't intend to do it again.
The working group runs the coaching, the community sessions and the cycle. The assistants and everything that configures them belong to the district. Districts that continue get an updated diagnostic and revised prescription. Districts that stop keep everything they built.
Free. No commitment. I'll benchmark your district against peers and show you what state reports don't.
Request a Free Diagnosticor email mluekens@directed.pro