Anish Sabharwal

Inqora includes Blinky (a desktop AI copilot like cluely.com and heyclicky.com), AgentOS (a lightweight K8s-like agent orchestrator for deploying agentic systems, built in Golang), experiments with RLHF, RLVR, SFT, and CPT on open-source SLMs, an Image Memory System using VLMs (Gemini models), and a Panel of AI Personas simulating simplistic elections. It also includes a RAG system for retrieval over curated financial and legal data stored in ChromaDB, with web and mobile app interfaces. Worked on creating voice flows for customer-service voice agents running on Twilio phone numbers, as well as a life-coach agent. Built a Document OCR system using PaddleOCR and Tesseract.js.

Launched Paxio - “AI that works with you”. Ask Paxio to order on Zepto, Ask it to Doomscroll the entire web, Connects Gmail, Notion, Calendar, Inbuilt memory for context + Autonomous Workflows. Traction: 10k+ impressions and 50 users.

Desktop Native Interface containing Personal Agent (integrating gmail, calendar, notion, forms and many more), Financial Agent to manage end to end budget tracking & expense analysis, Legal Agent to interact ith the Law in an interactive way powered by custom RAG piplines and Experiments with Shopping Agent to order autonomously on Quick Commerce. [Check out Technical Blog below]

Includes a Canvas Chat Interface (Interact with your favourite LLMs in Canvas native interface, Multi mode where a single query runs parallely on multiple LLMs, Vote Mode wherein LLM vote upon best response of each other), Workflow Builder (Build simple workflows with many Nodes like Input, Output, Processing, Condition, Notes in a interactive drag and drop interface), Explainer Video Maker (Generate simple explainer videos using AI of any topic with gameplay backgrounds)