AI can generate faster.
It still needs a method.
I help school, district, charter-network, and university leaders move from scattered AI use to a coherent instructional approach — one that expands teacher capacity, protects professional judgment, preserves rigor, and improves multilingual access.
The challenge is not only what teachers know. It is what they can execute — consistently.
We train teachers to reach multilingual learners. Then we ask them to demonstrate that knowledge in every lesson, across multiple proficiency levels, every period, every day.
The expectations are right. The professional learning matters. But understanding how to make rigorous content accessible — and having the hours to design that access consistently — are two different things.
It was never only a knowledge gap.
It was an execution gap.
What becomes possible when design comes first.
The impossible ask
- One lesson, five proficiency levels, and no extra hours to design for each.
- Differentiation happens sometimes — for some students, on the good days.
- AI used raw scales a generic lesson faster, for more classrooms at once.
- No shared standard for what "good" access actually looks like.
Every learner, every day
- Customized, language-accessible instruction as ordinary daily practice.
- The teacher's judgment directs the tool — the thinking stays human.
- Grade-level rigor held; the production burden moves to the machine.
- One shared method a whole team — and a whole system — can hold.
One task. Designed before it's generated.
A real move from the framework — turning a grade-level science task into language-accessible instruction without lowering the rigor.
Labeled diagram + sentence frames: "Light energy is converted into ___."
Cloze paragraph with a cognate word bank and a cause–effect connector list.
Write the process as a paragraph, then explain it aloud to a partner.
Did the student explain the process and use cause–effect language to do it — in content and in language?
Same rigor for every student. Built in minutes, not a lost planning period.
Four moves that keep human judgment in control.
The repeatable method behind the sample above — for directing, evaluating, and refining AI-supported instructional design.
Name the demand.
Identify the content, language, and participation demands students must navigate.
Meet the range.
Plan access around the range of language development and entry points in the room.
Protect the rigor.
Create supports that expand access without removing the thinking or lowering the rigor.
Read the evidence.
Examine what students understood, how they used language, and what instruction should do next.
Before asking educators to use AI, leaders need to define what good use looks like.
A working session for the leadership team of a school, district, charter network, or university — built to move you past tool adoption to the decisions responsible AI implementation actually requires. Valuable on its own, and the first step of a longer engagement.
- Establish a shared instructional stance.
- Identify quality and risk criteria.
- Define commitments to multilingual access and rigor.
- Select an appropriate pilot population or use case.
- Determine what evidence leaders need to examine.
- Identify a practical 30–90 day next step.
A leadership decision — not a generic AI demonstration.
A training day is not an implementation strategy.
Leadership alignment is the beginning. Depending on need, the work extends into role-specific professional learning, applied lesson design, coach support, and evidence review.
The goal is not more AI activity. It is stronger, more feasible instructional practice.
I've led this work where the plan has to survive Monday morning.
I was an English learner myself. I know what it is to sit in a classroom reaching for language you don't yet have — and I've spent 25 years making sure other students don't have to reach alone.
I've also watched extraordinary teachers quietly exhaust themselves trying to give every learner that access, by hand, every single day. That gap — between what we ask and what a human can sustain — is the reason this work exists.
- Higher edLed a state-mandated reform of a 2,000-student college ESL program — 50+ faculty — and moved it fully online in weeks.
- K–12Dean of Multilingual Education at a KIPP high school, then district-level multilingual program facilitation.
- CredentialsMA TESOL; ESL professor and department chair; 25+ years across classroom, program, and system.
- NowCertified AI Consultant — pairing that pedagogy with responsible, disciplined AI implementation.
"I am not an AI tech person teaching humans. I am a human-development person using AI as a tool."
CODE is becoming a larger body of work.
The forthcoming CODE book, the leadership workshop, educator tools, and implementation resources are being developed as one coherent system for making rigorous, language-accessible instructional design more feasible — for multilingual classrooms first, and for responsible AI-supported learning more broadly.
Define the instructional approach before scaling the tool.
For schools, districts, charter networks, universities, and education organizations seeking a coherent approach to AI-supported, language-accessible instruction.