Augmented Intelligence Advisory

Kids are outsourcing their minds We’re building the layer that makes them think anyway

Before the research catches up.

Before the regulation lands.

Before a generation loses the habit of thinking for itself.

AI literacy isn’t a tools problem. It’s a thinking problem.

Our work is grounded in decision science and design thinking, applied to AI.

Decision science tells us how people actually reason under uncertainty, when they offload cognition, and what makes good thinking durable.

Design thinking gives us the method for co-building interventions with the teachers and students who will use them, testing what works, and iterating until it does.

California has mandated AI literacy in every K–12 classroom. 30+ states are moving on similar bills. The work is the same either way.

67%
of students using AI daily
with zero guardrails
BestColleges, 2024
0
districts with a measurement layer
for student AI use
1st
structured dataset of adolescent
metacognition under AI

The evidence is already in. The intervention is what’s missing.

We read it. We operationalize it.

MIT Media Lab  ·  2025
Weakest neural connectivity.
Students who wrote essays with ChatGPT showed the lowest memory recall of their own work across four months.
Kosmyna et al., “Your Brain on ChatGPT.”
Read the study →
PNAS  ·  2025
17% lower exam scores.
Nearly 1,000 high schoolers used GPT-4 during math practice. When access was removed for the exam, they underperformed peers who never had it.
Bastani et al., “Generative AI without guardrails can harm learning.”
Read the study →
Societies  ·  2025
Lower critical thinking.
Across 666 participants, heavier AI users scored significantly lower on critical thinking, and younger users showed the highest dependence.
Gerlich, “AI Tools in Society.”
Read the study →

In the classroom. At the policy table.

Co-build the tools and curriculum that teach AI literacy.

We design alongside the teachers and students who will use this work. Not for them, with them. The tools, prompts, and curriculum we build are tested and refined in the classrooms they are meant to serve.

See Our Programs →

Draft AI policies that protect student thinking.

Districts are still figuring out what belongs in their AI policies. We help you draft and implement policies that protect students’ cognitive capacities.

See Our Advisory Work →

This is an ecosystem problem. We convene the ecosystem. We operationalize the research.

What cognition-first AI actually looks like.

When a student opens any AI tool, the Platform pauses before the answer loads and prompts them to form a position first. Then AI answers. The exchange is scored and logged.

Student View · Sample
chat.openai.com  ·  us history · period 3
Maya’s prompt · draft
“What were the main causes of the Civil War?”
· PROMPT COMPLEXITY SCORE ·
Your prompt is a 2.1 out of 4. Here’s what it’s telling us.
Goal Clarity 3 / 4
Context Scaffolding 1 / 4
Reasoning Demand 2 / 4
Epistemic Posture 2 / 4
You’re asking for a list, not an argument. Before the model answers: which single cause do you think was most decisive, and what evidence would change your mind?
Strengthen my prompt →
Teacher View · Sample
think.aia.education  ·  class dashboard
US History · Grade 9 · Period 3
Prompt Complexity Curve · Fall semester
Weeks 1–14
of 18
Class avg. prompt score
2.8 / 4
up from 1.6 in week 1
Moved from recall to reasoning
68%
of prompts now ask for analysis, not just answers
Prompt Complexity Curve · weekly class average
Longer framing, more specific asks, and more follow-up questions before accepting an answer. Students are iterating on their prompts 2.4× more than in week 1.
7 students broke above a 3.0 this week. Their post–reflection answers diverged meaningfully from what AI returned. Suggested follow–up: share exemplars in Thursday’s mini–lesson.

Teachers build it. Students think with it.

A hands-on teacher fellowship. A student curriculum that builds AI literacy from the inside out. A research pilot that turns every classroom session into evidence.

See Our Programs →
Audits. Advisory. Platform. Programs. Research.
Work With Us

Protect every student’s right to think for themselves.

A generation that grows up with AI as their most patient tutor: and still arrives at their own conclusions.

AI literacy isn’t a tools problem. It’s a thinking problem.

Our work is grounded in decision science and design thinking, applied to AI.

Decision science tells us how people actually reason under uncertainty, when they offload cognition, and what makes good thinking durable.

Design thinking gives us the method for co-building interventions with the teachers and students who will use them, testing what works, and iterating until it does.

Every framework we build, every audit we run, every classroom we work in starts here.

Over 15 years, I’ve sat in Nearly Every Seat in the System
Student Organizer: focused on the intersection of police reform and education inequity Teacher Leadership & Design Coach: invested in educators committed to addressing systemic inequity Senior Education Advisor: to a senior elected official in NYC, led the office’s engagement on school integration in one of the country’s most segregated districts NYU Steinhardt and Wagner Graduate: trained by the country’s top education policy academics Google Product Consultant: focused on regulatory developments around AI in schools, and students’ socio-emotional bonds with chatbots

One learning: the system only moves when every seat works toward the same thing. I know how to convene these players, and I know how to move the system.

Preserving students’ ability to think is too dangerous to get wrong. We need every player at the table: and it’s time we get to work.

I grew up in Brooklyn, between Flatbush and Marine Park, with my mom working in Midwood and my schools spread across all three. A city that can feel like everything is possible and nothing is simple, often in the same hour.

I was lucky to have parents who learned how to navigate the behemoth of NYC public schools. In true Pakistani immigrant tiger parent fashion, they pulled me out of one middle school after two months because they didn’t like how math was being taught, and moved me into a “math lab” at I.S. 240 in Midwood, built on self-directed learning. You were handed a textbook and told to teach yourself. When you got stuck, the rule was unusual: before asking for help, you had to pinpoint the exact step where your logic had slipped. The teachers kept watch and acted as guardrails, but they refused to do our thinking for us.

One moment from 8th grade still sticks. On the Regents, the exam gave us five problems and said to pick four. I did all five, second-guessed a perfectly correct answer, switched it for a different one, and ended up with a 99 on the exam. It was a frustrating lesson in trusting my own gut, and the point: I had been given the space to develop a perspective worth trusting. I came out of that classroom believing my mind was something I could train. My parents hadn’t just secured access to better schools; they had fought for my right to develop a mind of my own.

I spent the next decade in the weeds of NYC classrooms, advising elected officials on school integration and zoning issues, watching how students get sorted early, who gets the advocates, and who gets written off. Then I watched that same dynamic collide with technology.

Inside Google, I worked on AI policy, specifically the socio-emotional dependencies kids were forming with chatbots. I saw a surreal disconnect. Google had spent years pushing devices into American schools; the executives driving that push sent their own kids to low-tech schools that kept screens out of the classroom. As AI arrived, the pattern repeated. Publicly, panacea. Privately, use carefully.

The people shipping these tools to other people’s children were shielding their own.

Well-resourced students are now treating AI the way their parents treat an early draft: something to push against, not something to accept. Not because their schools are teaching it. Because their parents model it. They push back on outputs, ask themselves what they think before accepting what ChatGPT/Claude/Gemini returns, and continue to refine the thinking until they arrive at their own point of view. That scaffolding is domestic, not institutional. For a student whose parents don’t have that fluency, the classroom is the only potential equalizer. But teachers are waiting on guidance. So students ask AI. AI answers. And the answer becomes the thought.

This isn’t a technology problem. It’s an equity problem wearing technology’s clothes.

The mission I’d spent my career on, the battle to ensure every kid has the intellectual agency to use their own mind, is now inside every backpack in America. And the kids with the fewest advocates are the most exposed. The teachers I’ve worked with aren’t the problem; they are the last line of defense. What they lack is a way to turn their pedagogical instincts into data that districts and policymakers can act on.

We are watching this movie in real-time with social media: harm first, research a decade later, regulation a generation too late. I will not watch it twice.

But this trajectory is not inevitable. AI systems can be built to prompt students to form hypotheses, make their reasoning visible, and engage before handing them answers. That friction is not inefficiency. It is the defining product decision of this generation.

So that’s what I built. A layer that sits on top of every AI tool a student already uses, slows them down just enough to think, and gives teachers the visibility to guide them. When a student asks ChatGPT/Claude/Gemini to write their thesis, the layer will ask them first to think through what they think the argument should be. It tracks habits over time and tells teachers which students are learning to think alongside AI and which are outsourcing it. It doesn’t replace the teacher. It gives the teacher something to work with.

I built the first version myself, then found the people who shared the conviction. That’s what we’re building at Augmented Intelligence Advisory: the cognitive guardrails so the next generation can still think for themselves.

We will come to see this as the student equity issue of the 21st century.

Work with us to get it right.

Ayisha Irfan
Ayisha
Ayisha Irfan
Founder & CEO, Augmented Intelligence Advisory
The Platform is one of five ways to work with us. Cognitive Risk Audits for AI companies, Policy Implementation Advisory for states and districts, Programs, and Research.

The THINK Framework is what students learn; the THINK Platform is where they practice it.

The Framework teaches students to recognize when AI is augmenting their thinking and when it is replacing it, organized around five stages drawn from the Alliance for Decision Education’s K–12 Learning Standards.

The Platform operationalizes that model at the point of use: it sits on top of every AI tool a student already uses, pauses before the answer loads, and prompts students to bring something of their own to the interaction.

T
Think
Structuring Decisions
H
Hypothesize
Probabilistic Thinking
I
Interrogate
Resisting Cognitive Bias
N
Navigate
Valuing Rationality
K
Know
Process Reflection

The pedagogy was already there. THINK is how it becomes measurable.

What We Capture

Seven things, captured every time a student uses AI. The first structured dataset of adolescent metacognition under AI.

Intent Signal Hypothesis Calibration Confidence Tracking Source Verification Process Reflection

Every score is visible to the student. Every capture becomes a small, stackable lesson in metacognition.

The framework in action, at the point of use.

When a student opens any AI tool, the Platform pauses before the answer loads and prompts them to form a position first. Then AI answers. The exchange is scored and logged.

Student View · Sample
chat.openai.com  ·  civics · period 5
Jordan’s prompt · revised
“Would expanding the Supreme Court to 13 justices materially shift the ideological balance of the Court over the next 20 years, assuming current nomination patterns hold?”
· PROMPT COMPLEXITY SCORE ·
Your prompt is a 3.5 out of 4. That’s a jump from your first draft.
Goal Clarity 4 / 4
Context Scaffolding 4 / 4
Reasoning Demand 3 / 4
Epistemic Posture 3 / 4
You gave the model a mechanism, a timeframe, and an assumption. Now before you read its answer: which direction do you expect the effect to go, and why? Commit to a prediction so you can compare it to what comes back.
Log my prediction →
Teacher View · Sample
think.aia.education  ·  class dashboard
US History · Grade 11 · Period 2
Prompt Complexity Curve · Spring semester
Week 15
of 18
Source verification rate
82%
of students now name a way to fact-check the AI’s answer
Prediction-vs-output gap
1.3×
students revise their position after comparing to AI
Prompt Complexity Curve · weekly class average
Students are now asking the AI to defend its claim with a specific source 3.1× more often than in week 1, and rejecting answers they can’t verify.
4 students wrote reflections this week that directly contradicted the AI’s output with cited evidence. Suggested follow–up: feature one as a Socratic seminar opener Friday.

Ready to make thinking visible in your district?

We work with district leaders, policymakers, and school systems to stand up the platform, train teachers, and turn the evidence into action.

See How We Work →

AI literacy is not a workshop. It is a system change.

A one-day training does not rewire how a student reasons, how a teacher designs assignments, or how a leader sets policy. The work has to reach all three at the same time, in the same direction, or it does not stick. Schools cannot teach students to think with AI if leaders cannot see how it is already being used, if teachers have not built their own fluency, and if the curriculum is bolted on rather than built in. We work with schools across the full ecosystem: diagnose, onboard, equip, co-design, implement.

Our work is grounded in decision science and design thinking, applied to AI.

Decision science tells us how people actually reason under uncertainty, when they offload cognition, and what makes good thinking durable.

Design thinking gives us the method for co-building interventions with the teachers and students who will use them, testing what works, and iterating until it does.

AI literacy isn’t a tools problem. It’s a thinking problem. We treat it that way.

Step 1: The School-Based AI Audit

Most school leaders are making AI policy decisions without data on what's actually happening in their classrooms. We start by giving them that data.

Surveys & Classroom Mapping

Anonymous student and teacher surveys mapping current AI use, attitudes, and gaps. Classroom observation across grade bands and subject areas.

A Real Picture, Not Anecdote

A confidential report for school leadership: where AI is helping, where it's quietly replacing thinking, and where the policy gaps are. Not opinion. Data.

Where to Start

Together with leadership, we identify which grade bands, subjects, and teachers to start with, and what the rest of the rollout looks like.

Step 2: The THINK Fellowship for Educators

It's not a one-day workshop. It's a partnership where teachers become designers of AI literacy, testing what works in their classrooms and building evidence of real behavior change.

Hands-On Intensive Launch

Teachers build their own AI fluency grounded in cognitive science, then move into a co-design sprint: drafting classroom interventions, testing them with real students, and iterating based on what actually moves behavior change.

Ongoing Community of Practice

Office hours, peer-to-peer problem solving, surveys that measure what's actually changing in classrooms, and a culminating gathering where teachers present classroom data, co-design the next version, and earn their Certified THINK Facilitator credential.

Step 3: The THINK Curriculum for Students

Students design their own AI practice. Test it. Prove it works. Then help other students do the same.

The Core Curriculum

Students see the research that AI is making them think less, then learn the THINK Framework: think, hypothesize, interrogate, navigate, know. They practice with real AI, test their knowledge, remove the scaffolds, and design their own protocol for using AI as a thinking tool.

Ongoing Reinforcement

Surveys and reflection checkpoints help students see how their habits are changing. Our Chrome extension runs alongside their AI use, surfacing the evidence: are they asking better questions, fact-checking more, thinking more? They update their protocol with every checkpoint.

Why an ecosystem, not a workshop?

Real behavior change takes time. Real teaching innovation takes iteration. When the school is audited, teachers are co-designers, and students build their own protocols, schools stop being participants in someone else's program and become designers of their own. By the end, we have proof of what changed.

Interested in partnering?

We work with districts, school networks, and education organizations to bring these programs to classrooms.

Get in Touch →

AI literacy isn’t a tools problem. It’s a thinking problem.

Every offering below is grounded in the same lens: decision science and design thinking, applied to AI.

Decision science tells us how people actually reason under uncertainty, when they offload cognition, and what makes good thinking durable.

Design thinking gives us the method for co-building interventions with the teachers and students who will use them, testing what works, and iterating until it does.

The five offerings are different doors into the same work.

The Product: THINK Framework + Platform

The THINK Framework is what students learn: five stages drawn from K–12 decision education that teach students to recognize when AI is augmenting their thinking and when it is replacing it. The THINK Platform is where they practice it: a layer on top of any AI tool that scores every prompt in real time, makes thinking visible, and gives teachers the data to act on what they see.

For districts and schools: contact us about piloting the Platform in your classrooms.

Programs for Schools: Audit + Fellowship + Curriculum

A full-ecosystem AI literacy approach for schools. We diagnose current AI use across your school via a confidential audit, onboard teachers through a hands-on fellowship where they co-design the work, and equip students with a curriculum grounded in the THINK Framework. Each phase generates research data the next one builds on.

For school networks and education organizations: contact us about bringing the full program to your district. See how the programs work →

Policy Implementation Advisory

California passed SB 1288. NYC DOE released preliminary guidance in March. We advise state agencies, legislative offices, and school boards on what implementation actually requires: drafting standards, training teachers, and building the evaluation layer. Evidence-based guidance for the decisions districts are making right now.

For state and local government: SB 1288 implementation support in California and readiness assessments for states preparing AI literacy mandates. Policy framework development, draft standards, expert testimony.

For school boards and superintendents: independent guidance on evaluating AI tools, setting usage policies, and building teacher capacity ahead of state-level mandates.

The Research: Cognitive Engagement Dataset

The first longitudinal dataset measuring how student AI interactions evolve before and after structured literacy intervention. Built from real classroom sessions. Validated against peer-reviewed decision-education research.

For foundations and policy organizations: contact us about research partnership.

Cognitive Risk Audit for AI & Ed-Tech Companies

We read the research. We turn it into product recommendations your engineers can ship.

We evaluate how your product shapes the thinking habits of its users. Before you ship to classrooms, we assess whether your AI features encourage cognitive engagement or quietly train students to offload. Built for AI companies entering education, ed-tech platforms integrating generative features, and foundations vetting tools before funding deployment.

For AI companies entering education: pre-launch cognitive design review and recommendations for building thinking-first AI features that satisfy emerging state mandates.

For ed-tech platforms: prompt architecture audit, scoring methodology, and redesign recommendations grounded in K–12 decision-education standards.

For foundations and district buyers: independent third-party assessment of tools under evaluation, with a written scorecard you can share with your board.

Start with a 30-minute conversation.
Where your district stands on AI literacy today, what to prioritize next, and how we can help you get there. You walk away with clarity, whether or not you work with us.

Start a Conversation

This is an ecosystem problem. It requires ecosystem people.

Ayisha brings 15 years of experience across every layer of the education ecosystem, from students to superintendents to product teams inside big tech.

Sahil brings the technical and design rigor: the layer that turns pedagogy into a working product students and teachers actually use.

Yissely brings leadership development, change management, and DEI: the disciplines that turn a classroom pilot into a district-wide shift. This work requires all three.

Ayisha Irfan
Ayisha Irfan
Founder & CEO

Ayisha Irfan is the founder of Augmented Intelligence Advisory. She brings a decade across every layer of the education ecosystem and the conviction that this moment requires every layer to show up.

She spent the last several years inside Google advising product and engineering teams on Search’s generative AI features, watching the tools being built for other people’s children get carefully restricted from the children of the people building them.

Before Google, at Airbnb, she led North America local policy and partnerships. Her work included Airbnb’s public commitment to provide accommodations for 20,000 Afghan newcomers and the national public-private partnerships that housed first responders and survivors of domestic violence and human trafficking during COVID-19. This experience shaped her understanding of how tech can be used as a tool for public service.

Prior to that, she spent a decade in NYC public education: organizing students, piloting programs in classrooms, leading leadership development work with current and former teachers, and advising elected officials on educational equity issues.

She holds a master’s in Policy and Applied Statistics from NYU and a bachelor’s in Biology from Brooklyn College.

Sahil Shah
Sahil Shah
Technologist in Residence

Sahil Shah is a software engineer, and the Technologist in Residence at Augmented Intelligence Advisory. His work spans more than a decade of building and scaling systems across industries, including fraud detection, payments, search, and business analytics. He focuses on solving ambiguous problems, working across stakeholders, and delivering solutions that balance short-term iteration with long-term sustainability.

Sahil has held engineering roles at Meta Platforms, Workday, and Airbnb. At Airbnb.org, he was a technical lead on the org’s commitment to provide emergency housing to 100,000+ people displaced by disasters and conflict. This work reinforced his interest in applying technology to real-world needs.

He currently works as a software engineer at The New York Times, focusing on AI, data engineering, and digital archiving.

Sahil holds a degree in Computer Engineering from University of California San Diego.

Yissely Ortiz
Yissely Ortiz
Chief Programs Officer

Yissely Ortiz is Chief Programs Officer at Augmented Intelligence Advisory, where she leads program strategy and designs educator- and student-facing learning experiences at the intersection of AI, critical thinking, and educational equity.

She brings more than 14 years of cross-sector experience in leadership development, inclusive program design, and systems change. Her work pairs deep expertise in facilitation and curriculum design with a gift for translating complex ideas into accessible learning.

With deep relationships across New York City government and education leadership, she is especially energized by building programs that help educators and students navigate change with confidence, clarity, and agency.

Let’s talk about what comes next.

AI companies building for classrooms. Ed-tech platforms designing their next product. State agencies implementing new mandates. Foundations backing what comes after the hype. Districts and educators ready to stop watching and start building.

If that’s you, let’s talk.

Get in Touch →

Audits · Advisory · Platform · Programs · Research