Projects | Prasad Vaidya

Agentic Email-to-Task Workflow with Reporting Dashboard

Problem: Important emails often contain tasks, commitments, and updates that never reach the project system cleanly.
Why it matters: A constrained workflow can reduce missed commitments while keeping humans in control of final task creation.

LLM classificationn8nNode.jsTypescriptReact

Local LLM / OpenClaw Experiments

Research and experimentation

Problem: Local AI workflows are promising but constrained by model quality, hardware, tooling, and orchestration reliability.
Why it matters: Local models can enable private, low-cost, and offline workflows when the product design matches their constraints.

OpenClawLM StudioLocal GGUF modelsUbuntuTool calling

Music Learning Experiments

Problem: Music practice can become unfocused when theory, technique, listening, and feedback are not connected.
Why it matters: Small learning tools can make deliberate practice more structured without removing the human feel of music.

LLMGPTAudio inputPitch detectionReact Native

Personal Knowledge Base

Problem: Knowledge workers lose context across meetings, notes, screenshots, documents, and scattered task systems.
Why it matters: The product aims to reduce mental load by making past context searchable and useful at the moment of work.

LLM ClassificationOCRVoice Transcription

Further Reading

Read the notes behind the systems.

The writing archive explains the architecture judgment, AI workflow thinking, and product constraints behind these experiments.