Clay

The description states that Clay is an AI-native LMS designed to make the current education system more accessible for neurodivergent students. It connects to existing LMS platforms like Canvas, Blackboard, and Brightspace, and organizes courses, deadlines, assignments, and updates inside a more personalized interface.

It includes features such as:

• Clay Plan: Visual planning with milestones, time estimates, calendar blocks, progress tracking.
• One Next Step: Recommends one useful action based on deadlines, dependencies, workload, available time, and pace.
• Assignment Mold: Turns dense assignment instructions into clearer structure while preserving original requirements.
• Sensory modes: Includes multiple modes like Complete Calm, Focus Tunnel, Motion Safe, etc., which change interface behavior.
• Academic Memory: Remembers notes, decisions, preferences, working context, and exact place where student stopped.
• Memory Map: Shows connections between courses, assignments, milestones, notes, conversations, and preferences.
• Grounded companion: Uses actual course, assignment, plan, calendar, and memory context through chat.
• Voice-First Mode: Allows students to interact via voice with minimal interface.
• Personalizations: Lets students change fonts, themes, layouts, navigation, sounds, and interface behavior.
• Calendar: Combines deadlines, milestones, study blocks, and connected calendar events from multiple sources.

Inference Clay appears to be a hybrid system that uses both deterministic logic for academic facts (deadlines, permissions, progress) and AI for language reasoning and open-ended conversations. It is built using Next.js, TypeScript, Vercel, Neon Postgres, Drizzle, and various AI tools like Groq, Whisper, ElevenLabs, etc.

The author states that Clay is based on the idea that “every student processes, plans, and experiences education differently, so the learning system should be able to adapt ('mold') to them.” This suggests a positioning around personalization and adaptability rather than just accessibility.

It started focused specifically on neurodivergent students but evolved to include broader benefits such as calmer interfaces, clearer planning, voice-first access, and easier resumption of work — implying a potential expansion beyond its initial target.

Inference The positioning has shifted from being a niche tool for neurodivergents to something that could appeal to any student looking for better organization or interface control. However, this evolution is not backed by evidence of market validation or user testing outside the creator’s circle.

The primary stated customer is neurodivergent students who struggle with deadlines, tracking multiple tasks, time perception, and overwhelming interfaces.

Secondary audiences include:

• Any student seeking a calmer interface.
• Students wanting clearer planning tools.
• Users interested in voice-first access or easier resumption of work.

Inference While the author identifies a specific group (neurodivergents), there is no evidence of segmentation beyond this. No clear ICP definition exists beyond general student needs, and no data supports whether these users are already using or would adopt such a system.

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

Not evidenced

The project was built using:

• Framework: Next.js
• Language: TypeScript
• Hosting: Vercel
• Database: Neon Postgres + Drizzle ORM
• AI stack: Groq (for LLMs), Whisper (transcription), ElevenLabs / Amazon Polly (speech output), LiveKit (voice sessions)
• Tools used for development: ChatGPT 5.6, Codex

The system uses a hybrid approach:

• Deterministic logic handles academic facts like deadlines, permissions, progress.
• GPT models handle open-ended conversation and reasoning.
• Typed tools control what the model can access.
• Common actions bypass the LLM to save tokens.

Voice functionality includes:

• Turn detection via Silero VAD
• Real-time sessions with LiveKit
• Transcription using Groq Whisper
• Speech output using Amazon Polly or ElevenLabs

Inference The architecture shows a thoughtful balance between AI and deterministic logic, which may reduce risk in critical academic areas. The use of multiple AI services suggests an early-stage integration strategy rather than a fully integrated platform.

There is no evidence of revenue, customers, or adoption beyond the creator’s own testing and feedback from a professor and friends.

The author mentions:

• Showing Clay to a professor and several friends.
• Planning to use it during their upcoming semester.
• Running beta tests before wider release.
• Intending to test with neurodivergent students and educators.

However, no actual usage data, retention metrics, or product-market fit validation is provided.

Not evidenced

The description does not reference existing competitors in the LMS space. It implies that current systems are inadequate for neurodivergent users but does not compare Clay to other solutions like Canvas, Blackboard, Brightspace, or specialized accessibility tools.

Not evidenced

1. Single-person team: The entire project was built by one person (Shivam Arora), raising concerns about scalability, long-term maintenance, and lack of diverse perspectives.
2. No traction or user feedback: Despite claims about solving real problems, there is no evidence of real-world usage or validated demand.
3. Unproven AI integration model: While the hybrid approach seems intentional, it’s unclear how well this balances performance, cost, and reliability in practice.
4. Lack of product-market fit validation: The author states they want to test with students but hasn’t done so yet — meaning no real-world validation exists.
5. Unclear monetization path: No business model or pricing strategy is described.

1. What specific neurodivergent challenges did you observe in your research, and how do these translate into product features?
2. How are you planning to validate the effectiveness of Clay with actual users beyond informal feedback?
3. Can you describe the process for integrating with different LMS platforms (Canvas, Blackboard, etc.)? What level of API access is required?
4. How does Clay handle data privacy and security, especially when storing personal academic information?
5. Are there any plans to onboard educators or institutions early in development?
6. What are the technical limitations or bottlenecks you've encountered during prototyping?
7. How do you plan to scale beyond a single developer’s capacity?