Great question! While there is some overlap in the AI assistant space, here’s what sets Nityasha apart from Glean, Gumloop, ChatGPT apps, and Claude MCP:
Persistent Memory Across Everything. Nityasha remembers your preferences, context, and conversations across all sessions and devices. You can pick up a conversation weeks later right where you left off. Most ChatGPT apps reset context, but Nityasha builds a long-term understanding of you.
Actual Action-Taking, Not Just Chat. Unlike pure chatbots, Nityasha directly connects with your tools to take real actions. Glean focuses on enterprise search across apps, whereas Nityasha actually performs tasks like sending emails, scheduling meetings, and managing your workflow.
Built-In Roadmap & Sprint Planning. We offer a native Kanban-based project management tool with 30-day growth sprints, priority tracking, and four-column workflows (Planned, In Progress, Updated, Completed). This is integrated and not a separate tool that requires context-switching.
Personalized for Individuals, Not Enterprises. Glean targets enterprise knowledge management, Gumloop focuses on workflow automation, and Claude MCP is about protocol-based integrations. Nityasha is designed as your personal productivity companion, featuring AI CRM for contacts, habit tracking, and goal breakdown.
Human-Centric Learning. Our Study Mode uses multiple types of input (voice, image, text) and does not just provide answers—it guides you to learn. It’s designed for teaching, not just for answering questions.
The main difference: a unified experience versus fragmented tools. Instead of switching between 15 or more apps or trying to connect multiple AI agents, Nityasha provides one conversational interface that remembers everything and manages email, calendar, tasks, research, and planning in one flow.
Fair point! I could have done a better job explaining the technical depth upfront rather than leading with my age.
What we built: persistent memory using vector embeddings (Pinecone), semantic search across conversations, Socratic teaching system, and unified workflow integration. The value isn't in the base model - it's in the architecture layer we added on top.
I should've let the technical work speak first. Appreciate the feedback.
Thank you for the kind words! You're absolutely right - I didn't do this alone. My dad has been incredibly supportive throughout this journey, helping me learn to code and working alongside me on this project. The technical architecture and many of the hard problems were definitely team efforts.
I'm grateful to have that support system. It's easy to just consume content and games, so having someone push me to create instead has made all the difference. Your parents sound like they did the same for you - that's awesome.
Thanks for taking the time to share your experience!
Fair point on the "wrapper" label. Let me clarify what we're building on top of base LLMs:
Yes, we use OpenAI/Anthropic APIs - we're not training models from scratch (like you said, neither does Perplexity, Jasper, or most AI tools).
What we add (technical details):
1. Persistent Memory Architecture
- Vector embeddings of user context stored in Pinecone
- Semantic search across past conversations (not just in-session)
- Retrieval pipeline: query → embed → cosine similarity → top-k memories → inject in prompt
- Challenge: Managing token costs while maintaining context
2. Socratic Teaching System (Study Mode)
- Question analysis: detect knowledge gaps
- Progressive hint generation (not just Q&A)
- Tracks learning progression
- Example: Instead of "here's binary search code", asks "what property of sorted arrays makes this possible?"
3. Unified Workflow Integration
- Email parsing + calendar sync + task extraction
- Single interface reduces context switching
- Memory persists across all tools
Architecture overhead:
- User sends query
- Retrieve relevant memories (vector search)
- Build augmented context window
- Send to LLM with enriched prompt
- Generate + store new embeddings
- ~200ms additional latency for memory operations
You're right that the base intelligence is GPT-4/Claude. But saying "wrapper" feels like saying Notion is "just a wrapper around PostgreSQL" or Stripe is "just a wrapper around payment processors."
The value is in the layer we built, not the underlying model.
That said - we should've been clearer about this upfront. Our first comment didn't explain the technical depth. That's on us.
Re: the fake comments - you're absolutely right to call that out. Those were friends/early users we asked to support the launch. That was a mistake and goes against HN's culture of authentic discussion.
I'm 13 and this is our first HN launch. We didn't understand how much the community values genuine engagement over orchestrated support. Won't happen again.
Apologies to the HN community for trying to game the system. Should've let the product speak for itself.
Appreciate you taking time to give honest feedback instead of just downvoting. This is exactly why we launched here - to learn from people who know better.
What would make this feel genuinely useful vs "just another wrapper" to you?
Quick clarification though - Nityasha isn't actually an SLM (Small Language Model) itself. We use existing large models (like GPT-4, Claude) but add a persistent memory layer on top + specialized features like Study Mode.
The "showing results" part is what we're most proud of - we tried to focus on actually being useful for daily tasks rather than just being impressive in demos.
What kind of tasks are you using it for? Would love to hear more about your workflow!
Thank you so much for using Nityasha, Abir! Really appreciate you being an early user.
We spent a lot of time on the UI trying to make it feel natural and not overwhelming. My dad kept saying "if a 13-year-old can't understand it instantly, we failed"
Out of curiosity - what features do you use most? And what would you like to see added?
Also, when you say "better than some new gen slms" - are there specific areas where you find it more helpful? Always trying to understand what's working so we can double down on it.
$19.90/month for a browser seems steep, especially when most AI features are available elsewhere for free or built into existing tools. Feels like they’re trying to bundle convenience with exclusivity, but I wonder how many people will actually pay for it long-term.
Wow, this is a clever hack! Turning a printer into a scanner is such a simple yet elegant solution — reminds me how creative people can get with hardware limitations. Makes me wonder what other “hidden” functionalities old devices could have if we just experiment a bit.
Persistent Memory Across Everything. Nityasha remembers your preferences, context, and conversations across all sessions and devices. You can pick up a conversation weeks later right where you left off. Most ChatGPT apps reset context, but Nityasha builds a long-term understanding of you.
Actual Action-Taking, Not Just Chat. Unlike pure chatbots, Nityasha directly connects with your tools to take real actions. Glean focuses on enterprise search across apps, whereas Nityasha actually performs tasks like sending emails, scheduling meetings, and managing your workflow.
Built-In Roadmap & Sprint Planning. We offer a native Kanban-based project management tool with 30-day growth sprints, priority tracking, and four-column workflows (Planned, In Progress, Updated, Completed). This is integrated and not a separate tool that requires context-switching.
Personalized for Individuals, Not Enterprises. Glean targets enterprise knowledge management, Gumloop focuses on workflow automation, and Claude MCP is about protocol-based integrations. Nityasha is designed as your personal productivity companion, featuring AI CRM for contacts, habit tracking, and goal breakdown.
Human-Centric Learning. Our Study Mode uses multiple types of input (voice, image, text) and does not just provide answers—it guides you to learn. It’s designed for teaching, not just for answering questions.
The main difference: a unified experience versus fragmented tools. Instead of switching between 15 or more apps or trying to connect multiple AI agents, Nityasha provides one conversational interface that remembers everything and manages email, calendar, tasks, research, and planning in one flow.