How Coach Works
An AI-powered platform for K-12 CS coaches — designed to enhance your expertise, not replace it.
Built Around the Coaching Cycle
CS coaching follows a repeating cycle — Co-Plan, Co-Teach, Reflect — rooted in the Cornell Tech CS Coaching Toolkit. Each phase builds on the last, deepening the coaching relationship over time.
Co-Plan
Set goals, align expectations, map the CT framework together.
Co-Teach
Observe the classroom, document what you see, capture action items.
Reflect
Review observations, celebrate growth, set the next agenda.
Repeat this cycle as the coaching relationship deepens over time.
Coach supports you at every stage — from planning with AI assistance, to logging observations in the classroom, to reflecting with cited resources from the toolkit.
How the AI Works
Transparent, cited, and designed to sharpen your judgment — not shortcut it.
The Knowledge Base
The AI is grounded in the complete Cornell Tech CS Coaching Toolkit — 259 sections covering coaching principles, lenses, classroom practices, the CT Integration Framework, and real coaching scenarios. When you ask a question, the AI searches this knowledge base to find the most relevant sections.
No generic internet knowledge. No hallucinated frameworks. Every answer is anchored to the same toolkit your coaching practice is built on.
Cited Responses
Every AI recommendation cites its sources. When the AI references "Coaching for Equity" or the "Trust Lens", those citations link directly to the reference library — so you can always verify and go deeper.
Example Conversation
Resistance often signals a gap between the teacher's current comfort level and what they're being asked to do. I'd start with the Lenses: Trust — before addressing instruction, invest in understanding their perspective.
Consider asking: "What concerns do you have about adding CS to your math block?" This aligns with Principles: Agency — helping the teacher feel ownership rather than compliance.
How does this land for you? What do you already know about this teacher's background with technology?
Plan Generation
When you ask the AI to help build a coaching plan, it generates a structured draft with goals and milestones — mapped to the CT Integration Framework. Every plan is presented as a starting point for you to customize, not a final prescription.
The AI is explicit about its assumptions: it will tell you what context it used, and what it would need to know to improve the suggestions. You stay in the driver's seat.
Designed with Anchor & Steer
A design framework for AI that strengthens coaches, not replaces them.
Before building any AI feature in this platform, every decision had to pass three questions:
- What specific coaching skill does this feature enhance? If you can't name it, the feature doesn't have a clear reason to exist.
- What's the current workaround, and why is it inadequate? If coaches are already solving this well enough, a fancier tool won't get adopted.
- Does using this make the coach more skilled over time, or more dependent? Every feature should build judgment, not erode it.
Once value is anchored, five STEER dimensions shape every interaction between you and the AI:
Transparent Reasoning
You see the reasoning, not just the answer. The AI cites specific sections of the Cornell Tech toolkit and tells you how confident it is. If something is a best guess rather than a strong match, it says so.
Example
"I'm drawing on the Equity Lens here because you mentioned 70% of students are from historically excluded communities. But my confidence is moderate. What equity work has the school already done? That context would change my recommendation."
Your Guardrails
You set the boundaries. Before giving advice, the AI asks about your constraints, the teacher's experience level, and what's already been tried. It won't skip that step and jump straight to recommendations.
Example
"Before I suggest an approach, what's this teacher's experience with Scratch? Are there tools or strategies they've already told you they won't use? What are your non-negotiables here?"
Smart Handoffs
Some moments require a human, not an algorithm. When a situation involves student safety, personnel issues, or anything beyond the toolkit's scope, the AI says so explicitly and helps you prepare for that conversation instead.
Example
"A teacher expressing burnout goes beyond what coaching frameworks can address. Consider connecting with your school's support team. I can help you think through how to frame that conversation if it would be useful."
Evolving Trust
Every plan starts as a draft. The AI presents structured outputs with "you know this teacher best" and asks what you'd adjust. Over time, as you validate its suggestions, it can take on more. But you can always pull back.
Example
"Here's a starting structure for this coaching plan. What feels right? What doesn't fit your context or this teacher's current capacity? You're the one who was in the room."
Two-Way Learning
After every recommendation, the AI invites reflection. Your pushback and adjustments are signal, not friction. When you change a suggestion, the AI learns from it. Coaching flows both ways.
Example
"How does this land for you? What resonates and what doesn't fit your context? After you try this approach, I'd value hearing how it went. Your feedback makes the next suggestion better."
Read the full framework at jamelna.com/anchor-and-steer.
Built for Every Role
Coach is used differently depending on who you are. Here is how each role experiences the platform.
Your Coaching Command Center
Your dashboard shows all your active coaching relationships at a glance — how many plans are in progress, which milestones are coming up, and recent activity across your teachers. You see the full picture without digging.
When preparing for a co-planning session, open the AI Coach and describe your situation. The AI draws from 150+ coaching cards to help you think through your approach — always citing its sources so you can verify and go deeper before the meeting.
During observations, log what you see directly in the platform. Add action items for follow-up, tag observations to specific coaching plans, and share notes with the teacher. Everything feeds into the progress view automatically.
Your Progress, Your Way
As a teacher, you see your coaching plans, your classes, and observations your coach has shared with you. The CT Progress Grid lets you track each student's growth across 19 computational thinking competencies — updated as your coach logs observations.
When your coach logs a shared observation, you'll see it along with any action items assigned to you. The reference library is always available if you want to explore coaching frameworks independently — on your own schedule.
No PII leaves your classroom. Student names never appear in any AI prompt. The AI only sees aggregated, anonymized competency data.
District-Wide Visibility
As a district administrator, you see the big picture — all coaches, teachers, and schools in your district. The compliance dashboard is the centerpiece: every action in the system is logged with a complete audit trail and attribution.
Data export and deletion requests are built in for FERPA and GDPR compliance. You can generate reports, manage users, and ensure the platform meets your district's privacy requirements — without waiting on the vendor.
Role-based access control means every user sees exactly what they should — nothing more. Coaches see their teachers, teachers see their own data, and admins see the district.
Built With Care
CS Coach is designed with accessibility (WCAG AA contrast, semantic markup, keyboard navigation), privacy (FERPA and GDPR-aligned data handling with no student PII surfaced to the AI), and the Anchor & Steer framework at its core. The goal is a platform that makes coaches more skilled over time — not more dependent on the tool.
Read the full Anchor & Steer framework at jamelna.com/anchor-and-steer.