OpenAI 2026 hackathon

Sumi — An AI-Native Learning Companion for 1StopQuantum

Sumi makes abstract concepts easier by explaining, acting inside the learning app like 1StopQuantum (Our Proof Case here) and performing experiments learners can observe, understand, and repeat.

Solo project by Tarun Chawdhury · 0 likes · 0 comments

Archive position — measured, not model output

0 likes on Devpost

2,264 of the 7,856 archived projects have more likes, and 5,592 share exactly 0 — so this project's #7,042 place in the like-ranked listing is a tie-break inside that group, not a ranking.

Projects (log scale)

1
10
100
1k
10k
05,592
11,758
2285
3–4132
5–975
10+14

Likes on Devpost. ▲ marks this project's group.

Show the figures
LikesProjectsShare of archive
05,59271.2%
11,75822.4%
22853.6%
3–41321.7%
5–9751.0%
10+140.2%
Devpost like counts for all 7,856 archived projects, captured when this archive was built.

Executive Summary

What the company appears to be

Sumi is described as an AI-native learning companion designed to enhance interactive education by understanding a learner’s current screen, demonstrating next steps, and enabling safe experimentation within learning applications. It is built around a reusable platform architecture that can be adopted by other educational tools beyond its first proof case, 1StopQuantum.

What changed

The author states that Sumi evolved from an initial idea sparked by a voice conversation about how current learning materials fail to adapt or demonstrate next steps when learners are confused. This led to the development of a system where an AI companion (Sumi) participates directly in the learning process, rather than just answering questions.

Single most important open question

Is there evidence that Sumi’s platform architecture is sufficiently mature and scalable for adoption by other educational applications beyond 1StopQuantum? The description implies reuse but lacks any demonstration of such reuse or integration with external systems.

Back to contents

What The Product Actually Is

The description states that Sumi is an AI learning companion designed to participate in the learning process itself, not merely answer questions. It operates within a host application like 1StopQuantum and provides contextual guidance, visual demonstrations, prediction prompts, and immediate feedback.

Sumi includes:

  • A browser SDK for integration into host apps
  • A CLI for scaffolding and validating screen integrations
  • A Control Plane managing tenancy, screen registries, approved actions, telemetry, and evaluation

It is described as a reusable platform rather than a feature confined to 1StopQuantum. The system interprets learner intent and routes requests either into response-only flows or approved-action flows based on registered capabilities.

Evidence

  • Sumi is built as a reusable SDK, CLI, and Control Plane.
  • It integrates with 1StopQuantum as the first proof-case application.
  • The platform separates voice/runtime concerns from governance and observability.
  • It uses typed, screen-scoped actions instead of arbitrary browser control.
  • Deterministic application output serves as the source of truth.

Inference

  • Sumi is intended to be modular and adaptable across different learning environments.
  • Its architecture supports bounded interaction models that prevent unintended UI manipulation.

Back to contents

Positioning & Claim Evolution

The author claims that Sumi pushes AI for education beyond simple answer generation. It aims to provide contextual guidance, visual demonstration, prediction prompts, and immediate feedback inside the learning activity.

Key Claims

  • Sumi is designed to help learners understand abstract concepts through guided experimentation.
  • It introduces a new model of interaction where the AI understands the current screen and acts accordingly.
  • The system supports both explanation and action within the learning interface.
  • Sumi is positioned as a reusable platform, not just a feature inside one app.

Evolution

  • Started with a voice conversation exploring how AI could go beyond Q&A.
  • Evolved into a structured approach to educational interaction using real-time UI state.
  • Transitioned from being a single-use quantum education tool to a general-purpose learning companion platform.

Evidence

  • The project began as an exploration of what AI could do in education, not just generate text.
  • It evolved into a reusable SDK and Control Plane architecture.
  • The author explicitly states that 1StopQuantum is the first proof-case application, not Sumi itself.

Inference

  • The positioning reflects a shift from product-as-tool to platform-as-infrastructure.
  • The evolution suggests an intent to build a scalable educational AI framework.

Back to contents

Target Customer & ICP

The description identifies three main user groups:

  1. Learners: Receive contextual guidance, visual demonstrations, prediction prompts, and immediate feedback.
  2. Educators: Get structured ways to define learning objectives, review explanations, and manage approved actions.
  3. Organizations: Benefit from reusable screen registries, action policies, observability, evaluation, and governance across multiple learning applications.

Evidence

  • Sumi is described as providing a way for educators to define learning objectives and review explanations.
  • The Control Plane offers governance features like tenancy, screen registries, and telemetry.
  • The SDK allows adoption by other learning apps without importing quantum-specific logic.

Inference

  • The target ICP appears to be educational institutions or platforms looking to enhance their interactive learning experiences.
  • There is an emphasis on scalability and reusability for broader adoption beyond a single domain (quantum computing).

Back to contents

Business Model & Pricing Evidence

Not evidenced.

There is no mention of pricing, monetization strategy, revenue streams, or business model in the description.

Back to contents

Technical & Delivery Signals

Sumi's technical architecture includes:

  • A browser SDK and CLI for integration
  • A Control Plane managing organizations, applications, environments, screens, approved actions, prompt layers, telemetry, and evaluation
  • Voice interaction with speech-to-text (Whisper), text-to-speech (Kokoro), and interruption handling
  • End-to-end testing using Playwright
  • Integration with Qiskit and Cirq for quantum circuit generation and simulation
  • Model-agnostic runtime architecture supporting various LLMs, STT, and TTS providers

Evidence

  • Uses Codex with GPT-5.6 for development.
  • Implements typed, screen-scoped actions to prevent arbitrary UI control.
  • Employs deterministic verification of results.
  • Separates voice/runtime concerns from governance and observability.
  • Supports PWA behavior and offline capabilities.

Inference

  • The architecture supports modularity and adaptability across different domains.
  • The use of Playwright indicates a focus on functional correctness and UI fidelity.

Back to contents

Traction & Maturity Signals

Not evidenced.

There is no mention of revenue, customers, user adoption, or any traction metrics in the description.

Back to contents

Competitive Context

Not evidenced.

The description does not reference existing competitors or market positioning beyond its own claims.

Back to contents

Key Risks & Red Flags

  1. Lack of Traction: No evidence of users, customers, or revenue.
  2. Unproven Reusability: While described as reusable, there is no demonstration of adoption by other platforms.
  3. Single Developer Team: The team size is listed as one person (Tarun Chawdhury), raising questions about scalability and long-term maintenance.
  4. Dependency on Specific Tools: Heavy reliance on Codex with GPT-5.6 for development may limit flexibility or portability if those tools change.
  5. Limited Scope of Use Case: Currently focused only on quantum computing, which limits perceived applicability to broader markets.

Back to contents

Diligence Questions To Ask The Founders

  1. Has Sumi been successfully integrated into any other learning applications besides 1StopQuantum?
  2. What are the key challenges in scaling the platform beyond its current architecture?
  3. How does Sumi ensure consistency and accuracy of explanations across different domains (e.g., quantum vs. general science)?
  4. Are there plans to support more than one language model or speech-to-text provider?
  5. What is the roadmap for monetization or commercial viability?
  6. How will the platform handle updates or changes in host application interfaces?

Back to contents

Investment/Partnership Verdict

Not evidenced.

The description does not provide sufficient information to assess whether Sumi has strong potential for investment or partnership. While it presents an interesting architectural approach and a clear vision, there is no evidence of traction, revenue, customer base, or demonstrated scalability beyond the initial proof case. The single-developer team raises concerns about execution capacity, and the lack of external validation makes it difficult to evaluate commercial viability.

Confidence Level Low

Reasoning

Self-reported only; no third-party data, no revenue, no customers, no product-market fit evidence.

Back to contents

Source

Submitted to the OpenAI 2026 hackathon on Devpost. Project home on DevPost.

The analysis above was generated by a language model from the project's own one-line description. It is not independent research and contains no verified traction, revenue or customer data.