Over the past year, RocketPD has spoken with hundreds of superintendents, principals, instructional leaders, and teachers. Across districts of all sizes, one theme keeps emerging:
AI will only be transformative in K–12 if it’s aligned to the real context of schooling — the people, workflows, relationships, and constraints that define everyday practice.
The national conversation tends to focus on tools — which model is best, how to write effective prompts, which features will save time. But in education, the real leverage point isn’t prompt engineering.
It’s context engineering.
What Is Context Engineering?
Most AI models — including popular tools like ChatGPT — are incredibly powerful. But by default, they operate with generalized context.
They know research.
They know patterns.
They know frameworks.
But they don’t know your district.
They don’t know:
- Your specific teacher evaluation rubric.
- Your collective bargaining agreement.
- Your trust dynamics between principals and teachers.
- Your feedback norms.
- Your state compliance requirements.
- Your limited observation time.
- Your instructional priorities for the year.
That’s the gap.
And in K–12, that gap is everything.
As AI researcher Philipp Schmid argues, many AI “failures” aren’t about the model at all — they’re context failures. When AI doesn’t understand the environment it’s operating in, it produces output that is generic at best, counterproductive at worst.
In other words:
Better context → Better inputs → Far more useful outputs.
A Visual Look at the Difference
The slide illustrates what we’re seeing in real districts:
- Generalized LLMs know research, data, frameworks.
- Context engineering layers in the realities of schooling — strategic goals, meeting notes, historical trends, local language, and relationship-based nuance.
The result:
Highly targeted, district-specific solutions.
This is especially critical in processes that are deeply relational and often administratively overwhelming.
Which brings us to one of education’s most urgent challenges.
Where Context Engineering Matters: Teacher Evaluation
Teacher evaluation is one of the most complex workflows in a school system.
It is:
- Time-intensive.
- Trust-dependent.
- Documentation-heavy.
- Highly variable.
- Often frustrating for both principals and teachers.
And leaders are right to ask:
How can AI help without undermining relationships?
The answer lies in contextualization — not generic tools.
This is why RocketPD is partnering with Swiftscore, in consultation with Catalyst @ Penn GSE, to launch the National Reimagining Teacher Evaluation Research Cohort.
This is not a typical professional learning series.
It’s a multi-state, research-driven initiative to explore how contextualized AI can support human-centered educator growth.
What Makes This Initiative Different
Districts in the cohort will participate in real-world testing of AI solutions that are grounded in their systems — not somebody else’s.
Our shared design principles:
1. Context In First
Districts ingest real rubrics, observation histories, past feedback, and local workflows so the AI understands their environment.
2. Trust at the Center
AI supports clearer, more consistent feedback — but leadership judgment and relationships always stay primary.
3. Human + AI Workflows
Principals and teachers remain at the center of the process. AI accelerates the work; it doesn’t replace it.
4. Continuous Adaptation
The system learns from district input, new trends, and ongoing feedback — evolving over time instead of staying static.
Early partner districts are reporting:
- .Significant administrative time savings
- Improved feedback consistency.
- More meaningful instructional conversations.
- Stronger transparency.
- Increased trust.
But only when the AI is aligned to their real context.
Why This Matters for 2025 and Beyond
As AI becomes more common in classrooms, district offices, and instructional leadership teams, the question won’t be:
“Which tool is best?”
The question will be:
“Which system understands us well enough to actually help?”
Districts that invest in contextualized AI today will get ahead of the curve — not by automating people out, but by freeing leaders to spend more time on coaching, relationships, and instructional improvement.
How Your District Can Get Involved
We are building this national cohort with leaders who want to help define what comes next.
► 1. Download the Guide
A practical, research-backed overview of AI’s role in strengthening teacher evaluation.
[Get it here ->]
► 2. Register for our Webinar
A 60-minute conversation featuring district leaders, AI experts, and RocketPD/Swiftscore partners.
[Register now->]
► 3. Apply to Join the National Research Cohort
We’re accepting the first 30 districts for the 2025-2026 launch phase, with events kicking off in the fall and spring.
[Learn more & claim your seat]
► 4. Have a conversation with our team
If you’d like to understand what contextualized AI could look like in your district, we’re always open to a conversation.
Email us directly at info@rocketpd.com; Subject line: “Teacher Evaluation Cohort”
Let’s Build the Future of Educational AI — Together
AI in K–12 will succeed or fail on context.
The most impactful systems will be those that can inhabit the real world of schooling, not force everyone into generic workflows.
At RocketPD, we believe the next generation of educational AI must be built with districts, for districts — and in the context that makes each district unique.
If your district wants to help shape that future, we would love for you to join us.
