Embedded Files
Activity Overview
What the Research Says equipped you with two approaches to think about removing learning barriers:
Executive function calibration asks what is happening in the brain of a specific learner, and adapts support to that profile.
Universal Design for Learning asks what is built into the lesson itself, offering more than one way in so no learner needs a diagnosis to be reached.
CAST’s 2024 UDL guidelines fold executive function in as one of its three design categories. Every venture below picks up one, both, or neither approach.
Conversational AI tutors like Khanmigo and ChatGPT’s Study Mode sit outside this forecast. They respond to the task rather than to the underlying construct—such as working memory, attention, transfer, or self-regulation—that may be driving the struggle. This forecast focuses on tools that identify and adapt to that construct directly.
The Three Ventures
Each venture below is a bet on a different approach to AI-enabled adaptive tutoring. These three categories split along two questions:
Does the design act before a struggle appears, or react after it?
For the designs that act first, do they profile one learner’s specific need, or build for every learner at once?
Each profile is a strategic venture category, not a single company. Real, featured ventures anchor each as evidence.
Real ventures rarely sort cleanly into one box. You are assessing a category's design philosophy and scoring potential, rather than one founder’s execution.
NeuroScaffold AI
Its guiding philosophy is to know the learner, delivered through diagnostic-profile adaptation.
This video is an interview with Tobey’s Tutor founder Arlyn Gajilan, showing what this looks like in practice. Watch from 16:00 to 22:00 to hear how the platform builds a student’s profile from an 11-part quiz and IEP upload, then rebuilds its own interface around it.
Women Talkin’ ‘Bout AI. (2026, January 21). Vibe coding and building AI for kids: Inside Tobey’s Tutor with Arlyn Gajilan [Video]. YouTube.
How It Works
Maya’s problem, back in Activity 1, was never a shortage of ideas. It was managing them: starting a task, holding a multi-step instruction in mind, building a plan and sticking to it.
NeuroScaffold calibrates to executive function, specifically working memory, inhibitory control, and cognitive flexibility, the same three constructs you explored in What the Research Says, and uses that profile to remove the specific barriers a learner is facing. A student who struggles to start a task gets it broken into smaller steps. A student who loses track of multi-step instructions gets them delivered one piece at a time.
Who’s Building It
Tobey’s Tutor, whose founder you just heard from, names executive function challenges directly in its own design, built for kids with dyslexia and ADHD, profiled from report cards, evaluations, and IEPs, teaching through homework instead of completing it. CereBRO commits to the same population, naming ADHD, dyslexia, autism, and twice-exceptional learners (gifted and disabled at once) directly, but never names executive function as part of how it works.
Universal Tutor
Its guiding philosophy is to build for everyone from the start, delivered through proactive inclusive design.
This video shows a live demonstration of SchoolAI by a teacher setting up a video lesson so an AI chatbot checks understanding as students watch, then groups students by what they actually grasped. A different feature than the translation and text-to-speech example below, same underlying claim: one lesson, more than one entry point for learners.
Edutopia. (2025, December 10). AI tool demo: Turning passive video watching into active learning with SchoolAI [Video]. YouTube.
How It Works
Universal Tutor works by offering more than one way into every lesson, so no student has to be identified or diagnosed to be reached. A student who reads slowly can listen instead. A student who struggles to write can explain out loud instead. The lesson does not change; the way in does.
Who’s Building It
This is UDL applied to AI. IncluLearn AI, a 2026 peer-reviewed, user-centered adaptive tutoring system for digital-literacy development among women with disabilities in Jordan, uses UDL as a governing philosophy, alongside participatory design and explainable-AI principles, to provide personalized, accessible instruction (Maberah, 2026). SchoolAI, already in classrooms at scale internationally, ties UDL to specific features instead. Its own marketing describes “translation tools for English language learners, text-to-speech for students with reading disabilities, and adaptive content for advanced learners, aligning with UDL principles".
AccommoBot
Its guiding philosophy is to ada[pt, but never identify why a learner is stuck, delivered through performance-reactive adaptation.
This video shows a live demonstration of the adaptive tutor, ALEKS, the same reactive scaffolding this category is based on.
AleksCorporation. (2011, May 19). See ALEKS in action [Video]. YouTube.
How It Works
AccommoBot is not built for a specific population. Like Universal Tutor, it reaches any learner without diagnosing anyone. Unlike Universal Tutor, there is no design theory behind that reach, just a reaction to right and wrong answers.
Who’s Building It
ALEKS and ASSISTments both offer multi-step scaffolding: worked solutions broken into steps, and layered hints when a student gets stuck. But it is the same steps for anyone who asks, and it does not identify why a specific student is stuck, or which scaffolding steps that student needs. ASSISTments and ALEKS show this pattern in math: adaptive support, but with no diagnosis behind it. ReadTheory does the same in reading. Alpha School applies this same reactive, performance-based model at far larger scale than ALEKS or ASSISTments.
Which Venture Best Fits?
Which Venture Best Fits?
Each venture approaches learner support differently. This activity asks you to compare their strengths, limitations, and market potential before moving into the deeper investment analysis.
Instructions: Drag the name of the venture that best completes each statement into the blank. You may use the same venture more than once.
Record Your First Investment Read
Record Your First Investment Read
Based on the three venture categories presented so far, identify which one you would investigate first and what feature or concern most influenced your initial judgement.
Page updated
Google Sites
Report abuse