Building a 1:1 tutor was one of the original dreams of AI.

It’s finally possible.

To maximize human potential, we must scale 1:1 teaching.

We now have the tools to do this.

We’re building the components of an AI teacher in three parts.

The key to unlocking AI teaching for young kids is Child Speech Perception

Children learn by talking, so the best interface for young kids to interact with technology is natural language. But existing speech recognition does not work for children.

We’re leveraging advances in self-supervised machine learning to build 10x better child speech perception.
See research


A great teacher knows the child in front of them and continuously maps their cognitive state.

An AI teacher builds a granular picture of a child’s understanding and situates them along a learning pathway. Think of it as continuous phenotyping for every stage of early learning.


The best teachers find a way to engage learners at just the right level for them to stretch and grow.

Large generative models unlock the ability to create personalized stories and teaching materials that evolve and interact to meet children at that just-right apex for learning. We're building Language Models that are safe and produce outputs aligned with pedagogical best practices and produce content driven by the interests of each unique child.

Building a Learning Companion

We’re building delightful UI that sparks a love for learning and centers the child, not the technology


The tools for building a real AI teacher have arrived

Our team has played a central role in researching & developing the building blocks of AI teaching.

wav2vec 2.0: A framework for self-supervised learning of speech representations.
A. Baevski, H. Zhou, A. Mohamed, and M. Auli.

Advances in neural information processing systems. 2020;33:12449-60.

Systems and Methods for Speech Validation. U.S. Patent Application
C. Voss, T. R. Sayer, and A. Nag.

U.S. Patent Application No. 17/644,767

Data-Driven Diagnostics and the Potential of Mobile Artificial Intelligence for Digital Therapeutic Phenotyping in Computational Psychiatry
P. Washington, N. Park, P. Srivastava, C. Voss, A. Kline, M. Varma, Q. Tariq, H.
Kalantarian, J. Schwartz, R. Patnaik, B. Chrisman, N. Stockham, K. Paskov, N. Haber, D. P. Wall

Biological Psychiatry: Cognitive Neuroscience and Neuroimaging. 2020 Aug;5(8);759-69

Improved Digital Therapy for Developmental Pediatrics Using Domain-Specific Artificial Intelligence: Machine Learning Study
P. Washington, H. Kalantarian, J. Kent, A. Husic, A. Kline, E. Leblanc, C.Hou, O. Cezmi Mutlu, K. Dunlap, Y. Penev, M.Varma, N. Tyler Stockham, B. Chrisman, K. Paskov, M. W. Sun, J. Jung, C. Voss, N. Haber, D. P. Wall

JMIR Pediatrics and Parenting. 2022 Apr 8;5(2):e26760.

Active world model learning with progress curiosity.
Kim, Kuno, M. Sano, J. De Freitas, N. Haber, and D. Yamins.

Proceedings of the 37th International Conference on Machine Learning, 2020;119:5306-15.

Learning to play with intrinsically-motivated self-aware agents.
N. Haber, D. Mrowca, L. Fei-Fei, D. Yamins.

Part of Advances in Neural Information Processing Systems 31 (NeurIPS 2018).

Our team brings together world-class AI and learning science expertise


Evidence-based approaches to learning & engagement that motivates.

Artificial Intelligence

A radically different way for young children to interact with technology.

Learning Philosophy

Learning is done by a child, not to a child. A teacher puts the child at the center of the learning.

Read with Ello is the first AI reading coach that works with real books to support readers with evidence-based instruction & motivation, but that’s just the beginning. Language is the most natural way for a young child to learn, and the heart of engagement between student and teacher. Ello’s technology will unlock this critical interactive learning model for all students. Access to 1:1 support that is specifically tailored to the child’s interests and developmental needs will accelerate learning and maximize the potential for every child, across all subject domains.
Research by Education Reimagined

Organization of learners

Current paradigm
Organized in age cohorts
Learner-centered paradigm
Learners learn individually and in diverse and shifting groups consistent with their developmental, social, and competency needs.


Current paradigm
Standardized linear curricula divided into subjects
Learner-centered paradigm
Relevant, contextualized curricula organized by competency
Current paradigm
Learner-centered paradigm

Organization of learners

Organized in age cohorts
Learners learn individually and in diverse and shifting groups consistent with their developmental, social, and competency needs.


Standardized linear curricula divided into subjects
Relevant, contextualized curricula organized by competency

Learner goals / Progress indicators

Required credit hours and seat time
Development of competency in agreed domains of knowledge, skills, and dispositions in timeframes appropriate to each learner

Evidence-based literacy instruction

At Ello, our approach to literacy development is deeply rooted in evidence-based learning science. Ello follows a phonics scope and sequence, providing direct instruction to children as they learn phonics skills, as well as the opportunity to practice these skills by reading decodable books out loud and getting feedback from Ello. Just as Scarborough’s model highlights the multifaceted nature of reading, incorporating both language comprehension and word recognition skills, our program recognizes the importance of addressing various components of literacy simultaneously.

We understand that successful reading acquisition requires a synthesis of phonological awareness, decoding skills, vocabulary development, fluency, and comprehension strategies. Through engaging activities and personalized learning experiences, Ello strengthens the foundations of a child’s literacy skills, ensuring that children not only learn to read, but can harness the power of literacy to achieve academically and beyond.

The Science of Reading

The Science of Reading is a comprehensive approach based on decades of research into how children learn to read. It emphasizes the importance of teaching reading through explicit, systematic instruction in phonics — the relationship between letters and sounds.
This method helps children decode written language, laying a solid foundation for fluency, vocabulary, and comprehension. By aligning with this evidence-based approach, Ello ensures your child is on the best path to becoming a confident and proficient reader.

Decodable Books - A Vital Tool in Learning to Read

Decodable books are specially crafted reading materials that use simple, repetitive text aligned with a child's phonics knowledge, enabling them to practice and apply their reading skills effectively.

The Role of Decodable Books in Reading Education

Decodable books align closely with a child's phonics skills, enabling practical application in an engaging way. These books enhance word recognition and decoding abilities, making reading easier and more enjoyable. As children successfully read these books, their confidence and enthusiasm for reading grow, paving the way to fluency and confidence.


Witness your child's reading transformation: from basic phonics to engaging narratives, fostering a sense of achievement.