Product

Lirvana Labs is creating an AI‑powered writing companion for students in Grades 4 and up. It is intended to support learners, especially multilingual students and those in under‑resourced classrooms, through the full writing process: planning, drafting, reviewing, revising, and publishing. The tool aims to help students develop self‑regulation, focus their revisions, celebrate progress, and build confidence in their creative voice.

Age and Target Demographic

Students in Grades 4+, including:

  • Multilingual learners
  • Striving or reluctant writers
  • Students in under‑resourced or high‑need classrooms
  • Learners who benefit from personalized, scaffolded writing support

Team and Partners

Joan Ganz Cooney Center Sandbox, Lirvana Labs, The GIANT Room

 

Product Development Stage

Early‑stage product concept. The team’s goals for the engagement included:

  • Building foundational writing‑pedagogy alignment
  • Integrating personalization and safety features
  • Re‑scoping core interaction flows based on co‑design insights
  • Developing a clearer logic model and structured writing‑process supports

Background

This work took place within the Joan Ganz Cooney Center’s Sandbox for Literacy Innovations, which provides product teams with literacy consultations, Universal Design for Learning (UDL) support, and co-design sessions with children implemented by The GIANT Room. Before joining, Lirvana had an early concept for an AI writing agent and no direct involvement of students in the design process. Through the Sandbox, the team deepened their understanding of writing as a sequenced, self‑regulatory process and began building their product around established literacy frameworks and learner variability principles.

Design Question(s)

  • How might we support students in planning, goal‑setting, and drafting in their writing process?
  • How do students envision an AI writing agent helping them during moments of struggle or uncertainty?
  • At which points in the writing cycle would students choose to use AI, and what kind of relationship do they want with the agent?
  • What keeps students motivated during evaluation, editing, and revision?

Process

  • Across the Sandbox experience, Lirvana worked through three interconnected strands—literacy consultations, UDL and whole‑child design, and co‑design with children—that collectively reshaped their product and team practices.
  • Through literacy consultations, the team strengthened its instructional approach, shifting from outcome‑based writing support to a process‑aligned model grounded in the research-based Model–Practice–Reflect cycle. These conversations clarified how the AI should guide planning, drafting, revising, and reflection, and how feedback, tone, and task structure should adapt to different writing stages.
  • In the UDL and Learner Variability workshops, the Lirvana team emerged with plans for multilingual versions, neurodiversity‑aligned scaffolds, adaptive interaction modes, and richer personalization settings.
  • These earlier components informed the research questions for the co‑design sessions where children provided transformative insights after completing a short writing task to get into the writing mindset. Using techniques such as Layered Elaboration, roleplay, and rapid prototyping, the team learned how students want to interact with an AI writing partner, what helps them stay motivated, and what makes feedback feel supportive. These insights led to major product shifts, including customizable book‑character coaches, live scribing, task‑specific agent tone, relational features like memory of preferences, and tiered safety interventions.

Final Product

The product is still in development but evolved substantially through the Sandbox experience. It now has clearer instructional grounding, stronger whole‑child design, and is shaped directly by children’s voices.

The finding that learner involvement in the design of the feedback-providing AI increases engagement has led to Lirvana co-designing their agents with student users, ensuring not only that they collect inputs from the would-be users themselves to inform design decisions but also that they ensure a wide accessibility to AI by students of varied backgrounds.

In the current iteration through an offering called ConvAI, students of grades 4-12+ engage in spoken dialogue with agents on various topics with purposes ranging from end-of-year speaking state exams to debate practice and literary analysis. These exercises help students organize and develop ideas, strengthening their fluency, persuasive skills, and critical thinking, which are building blocks to growing as writers. Students have engaged in 9,910 ConvAI sessions since May 2025.

Project Photos and Video