Colleague.ai

60+ Questions to Ask K-12 AI Vendors

jennifer — Thu, 14 May 2026 16:55:02 +0000

With AI platforms — as with any SaaS — the devil is in the details. What’s possible, what’s not, what’s claimed versus what’s actually delivered, and what kind of impact a tool can have on teaching and learning depend almost entirely on how the platform was built, what assumptions it carries, and how customizable it is to your district.

The list below is the one we hand to superintendents, curriculum directors, and technology leaders evaluating AI platforms — and it now incorporates the requirements districts are putting into their own AI tool approval checklists, AI addenda to student data privacy agreements, and state-specific privacy statutes. Use it in your next vendor meeting. If a vendor cannot answer most of these — clearly, specifically, without hand-waving — you have your answer.

Based on our research, evidence-backed approaches, academic papers, and implementations, we built a list of 60+ questions to ask K-12 AI vendors.

If a vendor cannot answer most of these — clearly, specifically, without hand-waving — consider that a yellow, orange, or red flag.

Pedagogical Architecture & Assumptions

What pedagogical theories and learning frameworks are encoded into how your AI responds — Bloom's Taxonomy, Webb's DOK, UDL, Vygotsky's ZPD, cognitive load theory, others?

Why it matters: Every AI tool encodes a theory of learning whether the vendor names it or not. If they can't articulate it, it's still operating in your classrooms.
How does your platform support the full teaching lifecycle — plan, instruct, assess, analyze, iterate — versus optimizing only one step?

Why it matters: Tools that only support planning, or only support feedback, force you to layer more tools to fill the gaps.
How does your AI avoid reinforcing weak instructional practices when teachers use it as a shortcut?

Why it matters: An AI that makes a bad lesson faster is not a benefit. It's a risk multiplier.
How does your platform distinguish between recall, application, analysis, and synthesis — and how does it generate work at the appropriate DOK level for the standard being taught?

Why it matters: Without DOK awareness, the AI cannot tell the difference between knowing and understanding, and neither will your assessment data.
How does the platform handle productive struggle — when does it provide an answer versus a hint, a scaffold, or a question back?

Why it matters: AI that always answers is AI that bypasses learning. The architecture has to defend struggle, not collapse it.
How does the platform identify and remediate student misconceptions, rather than just marking work right or wrong?

Why it matters: Effective instruction targets the why behind a wrong answer. AI that only grades cannot teach.
How does the platform support gradual release of responsibility — modeling, guided practice, independent practice — within a single assignment or across a unit?

Why it matters: One-shot generation is not instructional design. Real teaching is sequenced, and your AI should be too.
How does the platform align to evidence-based instructional practices — explicit instruction, retrieval practice, spaced practice, formative assessment, feedback?

Why it matters: If the vendor cannot name the practices their tool supports, they are guessing at pedagogy.
How does your platform support culturally responsive and identity-affirming pedagogy in content generation, examples, and student-facing prompts?

Why it matters: Generic AI defaults to a single cultural reference frame. That is a quiet equity problem you cannot afford to ignore.

AI Model, Customization & District Context

What specific underlying AI models do you use — foundation model name, version, and provider — and have they been fine-tuned for K-12 instruction?

Why it matters: A generic foundation model wrapped in a chat interface is not the same as a model tuned on instructional data, standards, and pedagogy.
What is your fine-tuning methodology — supervised, RLHF, instruction-tuning, custom — and what data was used for training and evaluation?

Why it matters: How a model was trained determines what it does well, what it does badly, and what biases it carries. Vague answers here are red flags.
How often are models updated, what is your version-control process, and how do districts get notified when model behavior changes?

Why it matters: AI is not static. Districts deserve change management, not surprise behavioral shifts in their classrooms.
What district-specific documents can the AI ground in — curriculum maps, scope and sequence, instructional frameworks, AUPs, technology use policies, school improvement plans, IEP guidance, MTSS protocols?

Why it matters: If the AI does not know your district's documents, it is answering from generic web data and calling it instructional support.
How are district documents indexed and retrieved — what embedding model, what chunking strategy, what retrieval method, and how is grounding verified per response?

Why it matters: This is where most "RAG-based" AI tools fall apart in practice. Quality of retrieval determines quality of answer.
How does the AI cite its sources when it makes a claim or recommendation, and how can administrators audit those citations?

Why it matters: Citations turn the AI from a black box into a defensible answer. Without them, you cannot stand behind anything it produces.
What is the platform's measured hallucination rate on district-specific content, and how have you tested for it?

Why it matters: A vendor that cannot tell you their hallucination rate has not measured it. You will measure it for them, in front of parents.
Can the platform be customized at the district, school, grade, course, and teacher level — and how do those customizations propagate when a district document changes?

Why it matters: AI without granular customization is one-size-fits-all instruction in a chat box.
How does the AI behave when district documents conflict, are silent, or are out of date — does it refuse, default, escalate, or guess?

Why it matters: How a system fails is more informative than how it succeeds. Ask explicitly.
What specific AI methods does your platform use — Intelligent Tutoring Systems, adaptive learning, NLP, ML-based recommender systems, chatbots / LLMs, computer vision, speech recognition, translation, time series analysis, reinforcement learning, dimensionality reduction — and for what educational purposes is each used?

Why it matters: District AI data privacy addenda increasingly require vendors to itemize the AI methods used by purpose. A vendor that cannot list this in plain language is not ready to sign your AI schedule of data.

Data Privacy, Security & Governance

Where does student and teacher data live (is it stored in U.S.-compliant jurisdictions?), who has access, how long is it retained — and is any of it used to train your models, third-party models, or to develop synthetic or inferred data?

Why it matters: Most district AI addenda now explicitly prohibit student data being used for training, synthetic data generation, or inferred data — even when de-identified or aggregated. A vendor that cannot answer in plain English is not contractable.
Does the contract confirm that student data remains the property of the district — and that sub-licensing of student data is prohibited without explicit written permission from the LEA or eligible parents?

Why it matters: Data ownership is the single most contested clause in district AI agreements. The default should be that the data is yours, not the vendor's.
What student data does the platform collect — name, DOB, student ID, demographics, academic records, special education info, health information, biometric data, behavioral data, location, input data — and how is data minimization enforced so only necessary PII is collected?

Why it matters: Districts now require AI addenda to enumerate exactly which categories of student data are collected. Vague answers force the district to assume the worst — and to deny approval.
What FERPA, COPPA, PPRA, SOC 2, and state-specific (Washington RCW 28A.604 / RCW 42.56.590, California SOPIPA / AB 1584) compliance and certifications do you maintain — and where can we see the documentation?

Why it matters: Federal compliance is the floor. State student privacy statutes are where most vendors quietly fail.
Is data encrypted at rest and in transit, and what is the full list of subprocessors that touch student data?

Why it matters: Data flow is the story. A short list of subprocessors is good news; a long, vague list is a procurement risk.
What admin controls, audit logs, and guardrail configurations exist — and are they enforced at the model layer, not as a teacher-side reminder?

Why it matters: Policy enforced in a banner that teachers can ignore is not enforcement. The AI should behave correctly by default.
Do you have a written incident response and breach notification plan — and what is your notification SLA to districts after a confirmed incident?

Why it matters: Districts are obligated to notify families on a clock. A vendor without a written breach plan extends that clock and puts the district at legal risk.
Does the platform make any automated educational, instructional, or employment recommendations to students — and if so, how is that disclosed to teachers, students, and families, and do you certify that no response is determined in whole or part by paid third-party consideration?

Why it matters: Washington's Student User Privacy in Education Rights statute prohibits AI recommendations being influenced by paid third parties. The vendor must certify this in writing.
Do you commit, in the contract, to no targeted advertising to students and to no monetization of student data in any form?

Why it matters: This is a baseline contractual requirement in most state student privacy laws. A serious vendor commits to it without hesitation.
What is your process for notifying districts of material changes — new AI features, new subprocessors, new data collection, or changes to terms of service — and how much advance notice do districts get before changes take effect?

Why it matters: AI tools evolve silently. Most district AI approval checklists now require re-evaluation when ToS, AI features, or subprocessors change. Vendor advance notice is the only way that works.
What happens to district data at the end of the contract — return or destruction within 30 days, backup purging timeline, and what verification do you provide?

Why it matters: Year-one privacy promises are easy. End-of-contract data handling is where most vendors are quiet, and where many state privacy laws now place explicit obligations.

Student Learning, Growth & Cognitive Development

How does the platform support student growth and learning, not just task completion?

Why it matters: Productivity-first AI optimizes for finishing the worksheet. Learning-first AI optimizes for what happens in the student's head.
What specific safeguards prevent students from outsourcing their thinking to the AI?

Why it matters: Cognitive offloading is the single biggest risk of AI in education. The vendor must have a deliberate answer, not a deflection.
Does the platform scaffold student work by Depth of Knowledge (DOK) level, and does it track student progression across DOK levels over time?

Why it matters: Without DOK awareness, the AI cannot tell the difference between recall and analysis — and neither can your data.
How does the AI promote metacognition — does it prompt students to explain their reasoning, plan their work, or reflect on their understanding?

Why it matters: Metacognitive prompts are one of the highest-leverage instructional moves AI can make. Their absence is telling.
How does the platform identify a student's zone of proximal development and adjust difficulty in real time?

Why it matters: Static AI delivers the same response to every learner. Effective AI adjusts to the student in front of it.
How does the platform track standards mastery and longitudinal student growth — across units, semesters, and years?

Why it matters: A snapshot is a vanity metric. Longitudinal data is the basis for instructional decisions.
How does the AI distinguish between a student who is stuck and a student who is learning slowly — and what does it do in each case?

Why it matters: Same surface behavior, very different instructional needs. AI that cannot tell them apart will harm both.
How does the platform support student self-regulation, goal-setting, and reflection — versus delivering content at them?

Why it matters: Self-regulated learners outperform tutored learners over time. AI that does not build that muscle is borrowing future learning.
How does the platform support long-term retention — spaced practice, retrieval practice, interleaving — rather than only performance on the assignment in front of the student?

Why it matters: Students who pass the quiz and forget by Monday have not learned. The platform's architecture should reflect the science.

Research Foundation & Evidence Base

Where did your platform come from — who built it, and with what research foundation?

Why it matters: AI tools built first and sold into education second carry a different risk profile than tools built with educators and researchers from day one.
What independent, peer-reviewed evidence exists that your platform improves teaching practice or student outcomes?

Why it matters: Marketing claims are not evidence. Districts are entitled to ask for studies, not testimonials.
Was your platform built with school districts and education researchers in the room — and is it backed by federal research funding (NSF, IES, DOE)?

Why it matters: Federal funding is one of the few external signals that a tool was built with rigor, not just speed.
What research methodologies do you use to validate your platform's instructional claims — and have you published RCTs, quasi-experimental studies, or longitudinal effectiveness data?

Why it matters: A vendor claiming "research-backed" without rigor behind it is using the word as marketing, not as substance.
Who are the named education researchers, learning scientists, and former district leaders on your team, and what is their visible role in product design?

Why it matters: A learning-science team with names you can look up is a different signal than a vague reference to an "advisory board."

Preparing AI-Ready Graduates & Academic Integrity

How does your platform prepare students for an AI-enabled workforce — not by replacing their thinking, but by building it?

Why it matters: Your graduates will enter a labor market in which AI fluency is a baseline. The platform should be visibly contributing to that fluency.
How does your platform make it harder, not easier, for students to cheat — and how do you preserve student authorship and accountability?

Why it matters: An AI tool that turns assignments into generated artifacts undermines the entire purpose of the assignment. The architecture has to defend integrity by default.
Does the platform build AI literacy — teaching students how AI works, when to use it, when not to, and how to evaluate its outputs critically?

Why it matters: AI literacy is the new digital literacy. The tool itself should be part of the curriculum.
How is student work product attributable — what evidence does the platform retain that a student's thinking, not the AI's, produced the final artifact?

Why it matters: A teacher cannot grade what they cannot see. The platform should make student thinking visible, not hidden behind a chat interface.

Equity, Fairness & Uneven Outcomes

How does the AI avoid creating unfair advantages or widening achievement gaps across student subgroups?

Why it matters: Generic AI tools risk widening every gap they touch. The vendor must show that they have tested for this, not just hoped it isn't happening.
What testing have you done to ensure the AI performs comparably for English learners, students with IEPs, and students from historically underserved communities?

Why it matters: Ask for specifics. Bias and quality drift across populations is a measurable phenomenon, and a serious vendor will have measured it.
How does the platform support MTSS, differentiation, and intervention — not just one-size-fits-all responses?

Why it matters: If the AI cannot differentiate, it cannot serve the students who most need it to.
How does the platform support students with IEPs and 504 plans — does it read and operate within their accommodations automatically?

Why it matters: Accommodations as a manual teacher task are accommodations that get forgotten. The platform should know.
Is your platform regularly audited (at a defined cadence) for biases and fairness in AI decision-making — and can you produce documentation of those audits and the results?

Why it matters: State AI addenda increasingly require vendors to certify regular bias audits with documented results. A vendor without an audit cadence has not measured what it claims.
What documented strategies do you implement to identify and mitigate discriminatory effects in AI decision-making — and can you produce examples of bias issues you have actually surfaced and remediated?

Why it matters: A vendor that has never found a bias issue has never looked. Ask for messy real examples, not a polished statement.

Multilingual Support & Accessibility

What languages does your platform support — for student-facing AI tutoring, teacher-facing tools, and content generation — and at what reading levels?

Why it matters: Language access is an equity issue. "Supports translation" is not the same as "provides instructional support in a student's home language."
Is speech-to-text, text-to-speech, and other accessibility support built into the platform — not bolted on as an add-on or third-party integration?

Why it matters: Accessibility built in is durable. Accessibility bolted on breaks under load.
Can you provide a current VPAT (Voluntary Product Accessibility Template) — and what level of WCAG 2.1 AA and Section 508 conformance does it document?

Why it matters: VPAT is the industry standard accessibility disclosure. A vendor without a current VPAT (within the last 12 months) has not done the work, or has done it and won't show it.
Has your platform been tested with the assistive technologies your students already use — VoiceOver, JAWS, NVDA, screen magnifiers, switch access — and what gaps were identified and remediated?

Why it matters: WCAG compliance is the floor. The real question is whether the tool actually works for the students who depend on assistive technology.

Administrator & District Leadership Tools

What dedicated workflows exist for principals, curriculum directors, and superintendents — distinct from the teacher-facing tools?

Why it matters: A teacher tool sold to leaders is not a leadership tool. Districts need their own surfaces, not a different login to the same interface.
Does the platform support school improvement plan development aligned to state requirements and evidence-based research?

Why it matters: School improvement is one of the highest-stakes documents a district produces. If the AI can't help here, it isn't supporting district leadership.
Can the platform produce school-level and district-level AI impact reports — including which teachers are using it, how, and connected to student outcomes?

Why it matters: If you cannot see AI usage and connect it to student growth, you cannot defend the AI investment to your board.
Does the platform include enrollment forecasting, budget allocation, and resource planning tools that connect AI usage and outcomes to district priorities?

Why it matters: Strategic planning is where leadership lives. AI that stops at the classroom door is not a leadership platform.
Can the platform support communications to parents, board members, and the public — board reports, parent-facing summaries, community updates — grounded in district data?

Why it matters: Communication is leadership work. AI that can defend itself to a board in the leader's voice is a different category of tool.

Teacher Coaching, Feedback & Evaluation

Does the platform provide structured feedback to teachers on their instructional practice — and connect their AI usage patterns to student growth?

Why it matters: AI usage by itself is a vanity metric. AI usage tied to outcomes is professional development.
Can the platform support teacher evaluation or professional growth cycles aligned to our district's instructional framework — Danielson, Marzano, or a local framework?

Why it matters: If the AI is silent on the instructional rubric the district has already adopted, it is operating outside the district's existing systems.
Can principals and instructional coaches see classroom-level AI usage patterns and use them to plan coaching cycles?

Why it matters: Coaching at scale requires data. The platform should be feeding instructional leadership, not running parallel to it.

Implementation, Training & Long-Term Partnership

What does implementation actually look like — is this a classroom-by-classroom install, an individual-teacher signup, or a true institutional rollout across roles?

Why it matters: Tools that only deploy classroom-by-classroom never become district infrastructure. They stay anecdotal forever.
Does the platform support the full set of K-12 use cases — personalized learning, instructional support, automated grading and feedback, learning gap identification, special education, engagement, administrative support, parental engagement — or are you fundamentally a single-use tool that will force us to layer other AI tools to fill the gaps?

Why it matters: Most district AI addenda now require vendors to itemize every purpose of AI use across the platform. A single-purpose vendor will quickly become five overlapping vendors and a procurement and privacy nightmare.
What does training look like — fully self-serve, lightly-supported, or backed by a dedicated implementation team that knows education?

Why it matters: Self-serve training works for some districts. For most, the difference between adoption and shelfware is the depth of the vendor's training team.
What integrations are supported with our LMS, SIS, SSO, and identity systems — and how are those integrations maintained over time?

Why it matters: A demo that does not integrate with your stack is a demo, not a deployment.
What does ongoing partnership look like — year two, year three? How do you support districts in deepening instructional impact, not just product support?

Why it matters: A vendor whose support function ends at Tier 1 ticketing will not be a strategic partner. Ask what later years look like.
What is your process for incorporating district feedback into product development — and what concrete features in your roadmap came from district leaders?

Why it matters: A vendor who builds with districts will name the districts and the features. A vendor who builds for them will not.

Colleague AI Newsletter, April 21st

Victor Tian — Tue, 21 Apr 2026 10:00:00 +0000

Feature Highlight ✨

Welcome to your organizational and personal library!

At Colleague AI, we support content rigor and customizable needs.

Now with the new “Public Library” and “School Library,” you can find the best of two worlds. Derived from the previous “Find Lessons” feature, the Public Library lets you find high-quality OER lessons from Illustrated Mathematics, OpenSciEd, and more. You can then bring this content to Brainstorm ideas, use it as a template in My Documents to adapt, or generate interactive activities and games based on it.

In addition, if your account is under an enterprise license, you can upload materials to the School Library to make them available to your colleagues. This includes materials generated on Colleague AI or documents from your cloud or local drive. Your uploaded materials will be sent to your OrgIT and OrgLeader for approval before publication. Afterward, all your colleagues in your organization can use your content as a template.

Try It Now →

Your Milestone 🏅

The AICE Certified Educator Badge

🏅 The AICE Certified Educator Badge, offered by the University of Washington College of Education in collaboration with Colleague AI, recognizes educators who have shown they can use AI as an effective teaching assistant — not just for planning, but for designing impactful lessons, supporting all learners, and modeling safe, responsible use for students.

🚀 Why It Matters

Earning this badge means you’ve built a modern teaching skill set that helps you:

Save time without lowering quality.
Reach every learner with customized scaffolding.
Prepare students for a world where AI is part of everyday learning and work.

This badge shows your school, your colleagues, and your professional network that you are ready to lead in AI-supported teaching.

Learn More →

Our Community 📣

Colleague AI at ASU+GSV

Grateful to have been part of the conversation at the ASU+GSV Summit.

We were proud to bring a practitioner- and educator-centered perspective to the stage, grounded in how AI can meaningfully support teaching and learning in real classrooms.

Colleague AI at ASU+GSV Highlight

Our Co-Founder and CEO, Min Sun, joined Charles Elliott, Chris Clark, Jennifer Lee, and Kavitta Ghai to discuss the future of ambient AI in education.

Dr. Min Sun tackled one of the most pressing questions in EdTech: How do we design systems that empower rather than surveil? As AI becomes the “air” of campus life, how do we scale personalization without eroding agency?

For Colleague AI, the answer is simple: The teacher is the anchor. Data is not collected to monitor — it’s collected to inform.

We give educators a clearer view of where every student stands, and enable teachers to do what they do best: connect with students in meaningful, human ways.

“The human part of this is where I put most of my energy and thinking. I want students to enjoy school, be prepared for the future, and have a fulfilling life.”

It was also great connecting with so many of you at GSV Demo Day and at Booth #68. We appreciated the thoughtful conversations with educators, leaders, and partners exploring what responsible, classroom-ready AI can look like.

If we didn’t get a chance to connect in San Diego, we’d still love to continue the conversation — feel free to reach out.

Our Research 🔬

Heading to the 2026 NARST Annual International Conference in Seattle, WA?

Join our research partner’s session on How Science Teachers Use AI: A Descriptive Portrait from a National Survey.

Session Time: 8:15–9:45, April 21, 2026

Location: Jefferson B (L4)

Presenter: Dr. Joshua Rosenberg, University of Tennessee, Knoxville, USA

Artificial intelligence is reaching science classrooms, but how teachers use it remains unclear. We report results from a national survey of 566 U.S. public-school science teachers (RAND American Educator Panel, 2025). We defined generative AI as tools that create content or automate tasks using large data patterns.

About half of teachers reported any use of GenAI; roughly one in four use it weekly or daily. Use concentrates on teacher-facing work — planning lessons (40%) and creating assignments or assessments (32%) — with smaller shares for in-class support, differentiation, and routine paperwork. Only 5% said they teach students to use GenAI as a learning tool.

About a quarter said GenAI has changed how they interact with students, most often through more personalization, better support for varied needs, and more ways to explain difficult ideas. Teachers also reported limited student use overall.

Taken together, GenAI currently functions more as a planning assistant than a classroom learning tool. We close by outlining implications for professional learning, classroom-ready resources, and policy.

We value your opinion!

Your feedback helps us make Colleague AI better. Please take 2 minutes to complete our survey.

Take the 2-Minute Survey →

Colleague AI Newsletter, April 13th

Victor Tian — Mon, 13 Apr 2026 10:00:38 +0000

Our Commitment to Data Privacy and Trust 🔒

At Colleague AI, respecting user data privacy is our highest priority. Our platform does not access any cloud data, such as Google Drive or OneDrive, without explicit user permission. Every individual user, whether part of a district partnership or using Colleague AI independently, must actively authorize access before any integration is enabled. Users also retain full control to review and revoke permissions at any time through their Google account settings.

We are committed to transparency in how our systems work and to building trust with educators, schools, and districts. If questions ever arise, we encourage reaching out to our team directly so we can provide clear, accurate information: info@colleague.ai

We will continue to prioritize responsible AI design that supports educators while safeguarding their data.

Feature Highlight ✨

Access the Latest in Agentic AI with Colleague AI

Wondering how to access the latest in Agentic AI? Look no further. Colleague AI has integrated these advanced models directly into our AI Grading and Generate Interactive features. We've moved beyond simple prompting. You can now deploy autonomous AI Agents to handle your most complex and time-consuming workflows.

With our Generate Interactive (& report) and Agentic Batch AI Grading* features, you can:

✨ Analyze complex datasets using simple, conversational commands

✨ Process high-volume files with unmatched precision and speed

✨ Generate dynamic assets—from interactive worksheets and assessments to comprehensive dashboards—that do the heavy lifting for you

Experience the power of an AI that doesn't just respond, but acts. Use natural language to manage multi-step tasks and unlock deeper insights, all without needing any technical expertise.

*Agentic Grading (beta) is only available for enterprise use currently.

Try It Now →

Our Community 📣

Colleague AI Is at the ASU+GSV Summit

Our Co-Founder and CEO, Dr. Min Sun, will join Charles Elliott from Google, Chris Clark from Salesforce, Jennifer Lee from GSV Ventures, and Kavitta Ghai from Nectir on stage to discuss the future of ambient AI in education.

📅 Monday, April 13

🕒 3:00 PM

We're excited to contribute a practitioner- and educator-centered perspective on how AI can meaningfully support teaching and learning in real classrooms.

You can also find us at GSV Demo Day on April 14, and visit us at Booth #68 to see Colleague AI in action.

If you'll be in San Diego, we'd love to connect. DM us to meet the team and continue the conversation.

Our Podcast 🎙️

Testing Times: Relieving the Pressure of Testing Season with AI

April 15, 2026

3:30–4:30 PM PT

Published via YouTube Live @ColleagueAI

In this episode of the podcast, we examine how artificial intelligence is changing the way teachers approach test preparation before state testing season.

Join Hayley Spohn from Federal Way Public Schools (WA) as she discusses how educators can leverage tools like Colleague AI to quickly identify common misconceptions, generate targeted review activities, and support multilingual learners during high-stakes testing periods. Drawing from classroom practice, the conversation also explores how teachers can integrate social-emotional supports and explicit test-taking strategies to help students approach assessments with confidence and clarity.

Grounded in real classroom experience, this episode highlights how AI can reduce teacher workload while helping students engage more deeply with the skills and thinking processes that matter most.

Watch Now →

Our Research 🔬

The Colleague AI R&D Team Presented at AERA and NCME

The Colleague AI R&D team presented at several sessions at the American Educational Research Association Conference and at the National Council on Measurement in Education conference.

AERA

How teachers use generative AI tools

Alex Liu presented work exploring how teachers use generative AI tools. This session highlighted the fundamental research underpinning the development of Colleague AI's features.

See Session Details

AERA

Responsible AI use in K-12 learning

We were at a structured poster session with Colleague AI's academic partner, the University of Washington AmplifyGAIN Center. At this session, four federally funded AI R&D centers shared their work researching responsible AI use in K-12 learning.

See Session Details

NCME

LLM graders and human benchmark score alignment

Victor Tian presented research evaluating the alignment of LLM graders' scores with human benchmark scores.

See Session Details

Reimagining AI in education. Explore how our research and our platform meet at Colleague AI to provide full-cycle learning support.

Watch the Research Video →

Colleague AI Newsletter, April 6th

Victor Tian — Mon, 06 Apr 2026 10:00:48 +0000

Feature Highlights 🚀

Colleague AI Classroom just got an upgrade. 🚀

Teachers can now upload and organize their own files in the new Classroom Learning Library.

Give your students a 24/7 digital hub for every resource you share in class, all in one place and easy to access whenever they need it.

Check out the new upload tool now!

Try It Now →

Our Research 🔬

Meet Colleague AI R&D Team at AERA & NCME 2026 in Los Angeles, CA

Interested in learning more about our research? We will be presenting at several sessions at the American Education Research Association Conference and at the National Council on Measurement in Education conference.

AERA Session 1

How Teachers Use Generative AI Tools

Thursday, April 9 at 2:15 PM
The Westin Bonaventure
Floor: Lobby Level, Santa Barbara C

Join Dr. Alex Liu as she presents her work exploring how teachers use generative AI tools. Learn more about the fundamental research underpinning the development of Colleague AI's features.

See Session Details

AERA Session 2

Responsible AI Use in K-12 Learning

Friday, April 10 at 3:45 PM
Los Angeles Convention Center
Floor: Level Two, Room 515A

Join us for a structured poster session with Colleague AI's academic partner, the University of Washington AmplifyGAIN Center. At this session, four federally funded AI R&D centers will share their work researching responsible AI use in K-12 learning.

See Session Details

NCME Session

LLM Graders and Human Benchmark Score Alignment

Friday, April 10 at 3:30 PM

Victor Tian will present research evaluating score alignment between LLM graders and human benchmark scores. Come learn about quadratic weighted kappa and the statistical analysis underlying Colleague AI's features.

Session Details

Survey 📋

We'd love your feedback

Your feedback helps us make Colleague AI better. Please take 2 minutes to complete our survey about the new design.

Our Workflow 🎬

Discover how AI can help teachers build meaningful Tier 2 supports without pulling students away from grade-level learning

In this video, Ryan, a Multilingual Learner Specialist in Bellingham, Washington, shares practical ways to use Colleague AI to create standards-based interventions, scaffold prerequisite skills, generate accessible worksheets, build interactive learning supports, and even design a custom multilingual chatbot for students.

What stands out most is the flexibility. Teachers can bridge unfinished learning, support multilingual learners, customize materials with dignity in mind, and gain insight from student performance data and chat summaries.

This is a powerful example of how AI can support inclusive, research-based instruction and help every student participate in the general learning community.

Watch to see how teachers can:

✨ Create targeted prerequisite practice from existing lessons

✨ Scaffold materials for multilingual learners

✨ Generate interactive supports aligned to lesson flow

✨ Build custom chatbots for differentiated student help

✨ Use student interaction data to inform next steps

If you are interested in AI for teaching and learning, this is a great example of what is possible in real classrooms.

Watch the Video →

Colleague AI Newsletter, March 31st

Victor Tian — Tue, 31 Mar 2026 10:00:54 +0000

Community Challenge 🏆

Show Off Your Colleague AI Creations!

Our Call for Use Cases ends by the end of this Friday, Apr 3, 2026.

Winners of the CAI Spring 2026 Innovation Challenge will be announced by Apr 10, 2026.

Are you using Colleague AI (CAI) to transform your classroom? Whether you've built an immersive digital storybook, a high-stakes review game, or a complex simulation, we want to see it!

Colleague AI Generate Interactive is designed to be the dream tool builder for your instruction. From generating interactive activities and simulations to building worksheets and formative assessments that provide instant data reports, the possibilities for evaluating and supporting learning are endless.

Now, it’s your turn to take the stage. Show the community how you're using CAI to spark “aha!” moments.

🏆 What's in it for you?

Win a 3-month Plus Subscription to Colleague AI.
Score a $50 gift card to treat yourself or your classroom.
Be a guest on our monthly podcast and demonstrate your AI-supported instruction to educators.

📝 How to Enter

It's simple. Just send us:

The Share Link: A link to your generated CAI interactive activity.
The Context: A brief explanation of how you integrated this tool into your classroom instruction, such as your learning objectives, student grade level, or how you used the generated reports to pivot your teaching.

Ready to inspire?

Submit Your Use Case Here →

Our Use Case 🎨

Stop fearing AI in your ELA classroom. Start teaching students how to use it ethically.

Colleague AI can help educators move beyond fear-based conversations about AI and toward practical, responsible classroom use. When students learn how to use AI ethically, they build stronger critical thinking, reflection, and communication skills while staying grounded in authentic learning.

This use case highlights how teachers can create space for thoughtful AI use in ELA instruction, support student voice, and model what responsible technology integration looks like in real classroom practice.

Try It Now →

Our Research 🔬

ISEA Hackweek

Featured partner: AmplifyLearn AI Center

We're excited to highlight ISEA Hackweek as part of our growing research and innovation ecosystem. This initiative brings together educators, researchers, and builders to explore how AI can support meaningful teaching and learning experiences.

Programs like this help advance practical experimentation in education and create space for collaborative discovery around the future of AI in schools.

Learn More About ISEA Hackweek →

Our Podcast 🎙️

Testing Times: Relieving the Pressure of Testing Season with AI

Date: April 15, 2026

Time: 3:30–4:30 PM PT

Where: YouTube Live @ColleagueAI

In this episode of the podcast, we examine how artificial intelligence is changing the way teachers approach test preparation before state testing season.

Drawing from classroom practice, the conversation also explores how teachers can integrate social-emotional supports and explicit test-taking strategies to help students approach assessments with confidence and clarity.

Grounded in real classroom experience, this episode highlights how AI can reduce teacher workload while helping students engage more deeply with the skills and thinking processes that matter most.

Watch on YouTube Live →

We value your opinion!

Your feedback helps us make Colleague AI better. Please take 2 minutes to complete our survey about the new design.

Take the 2-Minute Survey →

Colleague AI Newsletter, March 3rd

Colleague AI — Tue, 03 Mar 2026 10:00:08 +0000

AI in Action: Recapping Our Time at NCCE 2026

Practical AI, Parent Comms, and Data. From the CAI Summit to the Exhibit Hall, here is what we learned with you in Seattle.

Our Summit 🚩

Practical AI, Parent Comms, and Data. From the CAI Summit to the Exhibit Hall, here is what we learned with you in Seattle.

It was a pleasure meeting so many of you last week in Seattle. Whether we spoke at the CAI Summit, caught up in the Exhibit Hall, or connected during our Presentation Sessions, we appreciate the time you took to engage with us. Summit Highlights: Moving from Theory to Practice The Colleague AI Summit provided a dedicated space for educators and district leaders to move beyond the hype and into implementation. Vision & Roadmap: Dr. Min Sun and the team opened with Colleague AI roadmap, followed by a live look at how our platform is streamlining lesson planning and data analysis. Leadership Perspectives: Justin Talmadge (Snoqualmie Valley SD) and Dr. Min Sun held a candid fireside chat on the district-level implications of AI and the current research driving development. Hands-On Learning: The afternoon concluded with differentiated workstations, allowing participants to explore AI at their own readiness level, from foundational concepts to advanced instructional orchestration. Spotlight Educators: Voices from the Field A major highlight was hearing directly from the practitioners using Colleague AI to navigate real-world classroom challenges. We want to extend a special thank you to our spotlight educators for sharing their "warts-and-all" implementation stories.

Slide deck created by Colleague AI Generate Slides

Our Exhibit 🛎️

At the Exhibit Hall: Connections & Conversations Outside of the Summit, our team spent time at the Colleague AI booth, where the focus was on real-time problem-solving. It was great to move from high-level strategy to the granular details of your daily workflows. We appreciated the chance to walk through live platform demos, answer your technical questions, and hear directly how you're tailoring AI tools to fit your specific educational goals. Thank you to everyone who stopped by to share their progress and challenge us with new ideas.

Our Session 📈

Data-Informed Leadership: Driving Decisions with Data On Thursday morning, we kicked off in Room 606 with Driving Decisions with Data: Your District at a Glance. Led by Tim Chalberg (High School Math Instructional Facilitator, Tacoma Public Schools) and Dr. Min Sun, this session moved beyond the technical specs of the Colleague AI District Dashboard to focus on its practical application. Participants explored how to translate raw dashboard data into actionable insights that actually improve student outcomes. By focusing on real-world district needs, Tim demonstrated how administrators and facilitators can use these tools to support educators and maintain a high standard of instructional rigor across their buildings.

AI Tools Every Teacher Needs - See What's Possible!

Watch Video

Our Use Case 💜

When parent emails feel like attacks, it’s easy to spiral Authored by Hayley Spohn You’re managing behavior, coregulation, academics, and everything in between… then an email lands that feels like an accusation. Your nervous system goes into defense mode, and suddenly you're up at 10 PM drafting and redrafting a response that proves you're doing your job. We get it. It's exhausting. Here’s the shift that helps: Most of the time, that email isn't really about you. Parents are doing what they’re wired to do. They’re protecting their child. They heard their child’s version, emotion came first, and advocacy followed. That’s care. They just don’t have the full picture yet.

This work is hard enough without carrying the weight of every charged interaction.If you're in the thick of a tough parent situation and need help reframing, drafting a response, or just processing the emotional labor of it all, use Colleague AI as a thought partner. Sometimes you just need someone to help you see it differently.

Colleague AI Newsletter, February 17th

Colleague AI — Tue, 17 Feb 2026 10:00:56 +0000

Newsletter

Batch AI Feedback and Grading in One Upload 📤

Our session at the 2026 NCSM AI Virtual Summit, and continued alignment with leading educational technology standards.

Feature Highlights ✨

New Batch AI Grading in Colleague AI

Grade multiple files with feedback in a single upload.

With Colleague AI’s AI Grading, upload your answer key and a full batch of student work, and the system evaluates everything at once—providing detailed, rubrics-aligned feedback for you to review and confirm.

What makes it powerful:

Upload multiple student samples (10 files max in my request) in a single step
Apply one consistent, standards-based rubric across all submissions
Receive scores and formative feedback for every student
Maintain alignment with your answer key and expectations

Batch upload. Detailed feedback. Review and finalize.

Our Community 💜

Colleague AI at 2026 NCSM AI Virtual Summit

11:00 AM - 4:30 PM (ET)

February 17, 2026

Zoom

CAI Speakers

Dr. Min Sun, Professor at University of Washington; CoFounder and CEO of Colleague AI

Alex Liu, University of Washington; Colleague AI

Guided by NCSM’s Educational Technology and AI Guidance , this summit looks beyond tools and trends to explore how artificial intelligence can enhance high-quality mathematics instruction, expand equitable learning opportunities, strengthen authentic assessment, and support informed leadership.

All times listed in Eastern Time (ET)

Summit Schedule

11:30 AM – 12:00 PM: Opening Session (Welcome & Framing)
12:30 – 1:30 PM: Concurrent Session Block 1 (choose 1 of 2)
2:00 – 3:00 PM: Concurrent Session Block 2 (choose 1 of 2)
3:30 – 4:30 PM: Concurrent Session Block 3 (choose 1 of 2)

Registration Details

NCSM Members: FREE
Non-Members: $85 Includes a one-year NCSM membership, with access to future member-only events and resources

Our Credential 🏆

We’re proud to announce that Colleague AI has been awarded the prestigious ISTE Seal!

To earn the ISTE Seal, our product underwent rigorous evaluation by a distinguished panel of experts in education, instruction, and technology. After careful review, our solution was recognized for its excellence in three key areas:

Practical Usability: Our product was evaluated for its user-friendliness and practicality, ensuring that educators and students can easily navigate and utilize its features.
Alignment with the ISTE Standards: Colleague AI was found to align closely with the ISTE Standards, ensuring that it meets the highest benchmarks for educational technology.
Digital Pedagogical Implementation: Reviewers determined that Colleague AI supports effective digital pedagogy, helping educators integrate technology seamlessly into their teaching practices.

This achievement demonstrates that Colleague AI is designed for modern-day teaching and learning. The Seal is a testament to our commitment to providing top-quality educational tools that empower educators and enhance student learning. Thank you to the ISTE panel and to all our users for your continued support and feedback!

February Live Podcast 🎙️

In this second episode of the Colleague AI: AI Implementation at School and District series, we shift the focus to instructional coaches and their role in supporting AI integration.

Join Ryan Fries from Bellingham Public Schools (WA) and his colleagues as they share firsthand experiences, support strategies, and challenges encountered while guiding educators through AI adoption. The conversation highlights opportunities created through professional learning communities, coaching cycles, and dedicated AI learning hours, offering practical insights into how instructional coaches help translate emerging AI tools into sustainable instructional practice across schools.

In this podcast we’ll explore how instructional coaches can support the responsible adoption of AI in the classroom. Whether it’s teacher facing or student facing applications, we’ll discuss how coaches can leverage PLCs to help prepare teachers and students for the future

Colleague AI Newsletter, February 10th

Colleague AI — Tue, 10 Feb 2026 10:00:07 +0000

Newsletter

How Colleague AI Brainstorm Ideas Stands Apart from Other AI Chat Tools

Meet Colleague AI at AASA 2026 and in our February podcast 🎟

Feature Highlights ✨

This isn’t just another chatbot.

Here’s how Colleague AI Brainstorm Ideas stands apart from other generative AI chat tools:

1) Simpler input, classroom-ready output
With minimal prompting, Colleague AI delivers well-rounded, classroom-ready responses that are organized, practical, and ready to use in the classroom.

2) Built-in tools, integrated into one response
Instead of sending you elsewhere, it can automatically generate visualizations and search credible online sources, bringing everything together in a single, coherent response.

3) Pedagogically grounded and research-informed
Colleague AI is rooted in educational research and sound instructional practice. It promotes transparency by clearly connecting generated content to underlying learning theories and instructional reasoning, helping educators understand not just what to teach, but why.

4) Supports educator agency, not replaces it
Brainstorm Ideas offers thoughtful suggestions that fully address instructional needs while respecting educators’ professional expertise and decision-making. The goal is to support, not substitute, professional judgment.

Try taking this same prompt to other tools:

“ Create a worksheet for K-PS2-2 Motion and Stability: Forces and Interactions. Analyze data to determine if a design solution works as intended to change the speed or direction of an object with a push or a pull. ”

You’ll quickly see the difference in depth, structure, and instructional alignment.

Our Community 💜

Colleague AI at AASA 2026

8:00 AM - 9:00 AM (local time)

February 14, 2026

Nashville, TN

Speakers

Dr. Min Sun, Professor at University of Washington; CoFounder and CEO of Colleague AI

Dr. Kelly Aramaki, Superintendent, Bellevue School District, Bellevue, WA

Dr. Trevor Greene, Superintendent, Yakima School District, Yakima, WA

Do More with Less: How Two Districts Enhance Teaching and Learning with AI Amid Fiscal Constraints

In an era marked by budget cuts, educational leaders face urgent questions about how to maintain instructional quality while navigating resource constraints. The rise of AI has sparked both excitement and skepticism about its role in classroom. This session highlights how two WA state districts, Bellevue and Yakima, are responding to these challenges by implementing AI-powered technological innovations to enhance student learning and optimize teacher efficiency. The AI innovations empower teachers' lesson planning, facilitate in-class discussions, provide students with 24/7 tutoring, and streamline feedback and assessment processes. Superintendents Aramaki and Greene will share their implementation stories and their partnership with a national R&D center of GenAI in classroom instruction housed in University of Washington. The session offers actionable insights for district leaders exploring responsible and practical AI integration to improve system efficiency and personalize learning.

February Live Podcast 🎙️

In this second episode of the Colleague AI: AI Implementation at School and District series, we shift the focus to instructional coaches and their role in supporting AI integration.

Colleague AI Newsletter, February 3rd

Colleague AI — Tue, 03 Feb 2026 10:00:40 +0000

Newsletter

A Closer Look at New Generate Slides

Teaching through current events; Colleague AI at AMTE 2026 and IETA 2026; Community event

Feature Highlights

Update: Colleague AI "Generate Slides" 2.0 ✨

Building a slide deck usually involves two time-consuming stages: the struggle to get words on the page and the tedious hours spent formatting them. With the latest update to Colleague AI, we are bridging that gap.

We are excited to roll out the new Generate Slides capability. We have streamlined the creation workflow to be more intuitive, but the real magic is in the quality jump . The slides are cleaner, more professional, and ready for the boardroom with minimal intervention.

Streamlined Creation: A simplified flow takes you from prompt to presentation in seconds.
Elevated Aesthetics: Generated slides now feature improved layout logic and visual fidelity.
Seamless Adjustments: Utilize AI assistance to refine content per-slide without starting from scratch.

Ready to dive in? Check out the new workflow in action in our step-by-step video and generate your first deck in minutes.

Use Case

What to do when current events make their way into your classroom?

Authored by Hayley Spohn, teacher from Federal Way Public Schools, Colleague AI Education Success Partner.

When students come to school carrying fear about things happening in the world, teachers are often left wondering what to say—and what not to say.

This check-in generator helps you respond to current events in a way that:

Centers student safety and dignity
Avoids putting anyone on the spot
Gives students choice in how they engage (write, draw, speak, move, or pass)
Uses simple, age-appropriate language
Helps you transition back to learning with clarity and care

Designed for moments when students are asking questions before you’ve had time to prepare.

Save it for the days when your classroom needs steadiness, structure, and care.

Image generated by Colleague AI Generate Image Advanced mode.

Try Now →

Our Community

Colleague AI will present at AMTE 2026 | Portland, OR

Amplifying Mathematics Instructional Practice with AI: Real-World Use Cases Across Education Contexts

Extended Session (9:30–11:15 AM)
Session 190 | Willamette, Main Lobby Level

Presenters

Alex Liu, University of Washington; Colleague AI

Dr. Lorraine M. Males, University of Nebraska–Lincoln

Dr. Min Li, University of Washington

Session Description
This panel showcases four real-world use cases of AI-powered instruction across mathematics teacher education contexts, ranging from lesson planning and critical thinking development in preservice programs to formative assessment use in field placements. Drawing from teacher preparation programs and K–12 district implementations, panelists will illustrate how AI augments pedagogical practices and supports reflective learning. Attendees will explore how educators use conversational agents and rubric-based tools, and examine how school districts leverage teacher–AI interactions to guide professional learning.

This session is appropriate for mathematics teacher educators, methods instructors, and education leaders, and aligns with AMTE’s Candidate Standard C.1.6 (supporting teachers in using technological tools).

Colleague AI will present in the NCCE room at IETA 2026 | Boise, ID

Support Evidence-Based Decision-Making with AI: From Data Analysis to Student Success

Date: February 5, 2026
Time: 1:15–2:45 PM
Presenter: Dr. Bridget Palmer-Chavez, Colleague AI Education Success Manager

From Lesson Design to Feedback: Teacher Tools in Colleague AI

Date: February 6, 2026
Time: 8:00–9:30 AM
Presenter: Dr. Bridget Palmer-Chavez, Colleague AI Education Success Manager

Our Live Podcast

In this second episode of the Colleague AI: AI Implementation at School and District series , we shift the focus to instructional coaches and their role in supporting AI integration.

Colleague AI Newsletter, January 20th

Colleague AI — Tue, 20 Jan 2026 10:00:09 +0000

Newsletter

Colleague AI's New Year Message 💌

Funding opportunities for ESCs; Join Colleague AI Summit at NCCE 2026, Feb 26th, Seattle, WA; Feb Podcast-AI implementation story with intructional coaches

Our Community 📣

AmplifyGAIN will award five contracts ranging from $8,000 - $20,000 , based on scope and anticipated impact. Selected ESCs will join a national cohort focused on advancing responsible, classroom-ready AI use and contributing to a growing peer learning network.
This opportunity is open to all eligible ESCs nationwide.

What Selected ESCs Receive
🏆 Contract funding to support professional development work
🏆 Ongoing technical assistance and collaboration with the AmplifyGAIN research team
🏆 Access to high-quality training materials and national dissemination opportunities
🏆 Free 3-month enterprise licenses to the Colleague AI learning platform for participants
🏆 Eligibility for participating educators to earn the AI Competency for Educators (AICE) Badge from the University of Washington

🔗 For more information

RSVP Info Session →

Our Event 💎

Join Colleague AI Summit at NCCE 2026

1 - 4 PM, February 26, 2026

Room 612 | Seattle Convention Center

This is a hands-on learning lab for educators and school and district leaders. Hosted at NCCE 2026, this three-hour summit—presented by Colleague AI—is designed for K-12 teachers, instructional coaches, and administrators who want practical, responsible ways to use AI in real classrooms. Participants leave with a clear vision for AI’s role in teaching, strategies they can use immediately, and pathways for continued growth, whether they are just starting or already leading AI innovation.

What to Expect

1:00–1:45 PM | Keynote + Live Platform Demo
Leading the Next Wave of AI in Education (2026 and Beyond)
Dr. Min Sun opens with where AI is heading in 2026, grounded in current research and district realities, followed by a live demonstration of how Colleague AI supports lesson planning, assessment, classroom implementation, and data analysis.

1:45–2:45 PM | Education Innovator Spotlight
Educator Voices: What AI Looks Like in Practice
Four Washington-based education innovators share concrete classroom and district use cases—what worked, what didn’t, and what they wish they had known earlier—drawing directly from their implementations.

2:45–3:45 PM | Educator Workstations (Hands-On)
Choose Your Path: AI at Your Readiness Level

New to AI: Foundations, core concepts, and safe entry points for classroom use
Intermediate: Designing lessons, assessments, and student supports while maintaining instructional intent and rigor
Advanced Practice: Workflow optimization, instructional orchestration, differentiation, feedback, and reflection

Each station is facilitated and built for active exploration.

3:45–4:00 PM | Your Voice
What Will You Try Next?
Reflect, share takeaways with peers, and provide feedback to shape future educator-centered AI tools and professional learning.

Why Attend : Walk away with clarity, confidence, and concrete next steps—AI in action, aligned to classroom practice.

RSVP now for a chance to win a free one-day (Thursday) registration for NCCE 2026.

* Registration to NCCE 2026 is separate. If you RSVP now, you will enter a chance to win a free one-day (Thursday) registration for NCCE 2026. If you win, we will notify you by January 31st, 2026.

RSVP for Colleague AI Summit* →

Our Podcast 💜

In this second episode of the Colleague AI: AI Implementation at School and District series , we shift the focus to instructional coaches and their role in supporting AI integration.

AI in Action: From Vision to Classroom Practice

RSVP for Colleague AI Summit* →

Our Podcast 💜

In this second episode of the Colleague AI: AI Implementation at School and District series , we shift the focus to instructional coaches and their role in supporting AI integration.