Ramanjit SinghFounder · AT Innovators · Ludhiana, IN
AI that answers your calls, books your meetings, and runs your busywork.
Voice agents, document chatbots, and workflow automation for small businesses — scoped, built, and shipped by the person you talk to. No layers.
Stack
PythonFastAPINode.jsReactOpenAIClaudeLLaMAPostgresn8nWebRTCDocker
Featured work
My own products, built end to end
01 · Voice AI
VoiceLead AI
- Problem
- Sales calls missed after hours; leads go cold before anyone follows up.
- Built
- A real-time voice agent that answers, qualifies interest, and books meetings against live calendar availability.
WebRTCWebSocketsLLMCRM Sync
Case study →
02 · Intake automation
ClientPilot AI
- Problem
- Inquiries pile up in inboxes; booking takes days of back-and-forth.
- Built
- Intake that answers from the business’s own documents and turns a form fill into a scheduled appointment — webhooks, SMS, calendar.
RAGWebhooksSMSGoogle Calendar
Case study →
03 · Knowledge chat
EcoMind AI
- Problem
- Answers buried in thousands of pages of PDFs and internal content.
- Built
- An assistant that searches it all semantically and streams answers grounded in the organization’s own material.
LLaMAEmbeddingsSemantic Search
Case study →
04 · Marketing automation
FlowForge
- Problem
- Campaign steps live in sheets and inboxes; the same manual flows get re-run by hand every week.
- Built
- A node-canvas workflow engine — triggers, email, SMS, Canva, and Sheets wired together, with watchable step-by-step runs.
OrchestrationGoogle SheetsEmail + SMSCanva
Case study →
05 · RAG advisory
EduNavigator AI
- Problem
- Study-abroad answers need explanation and data together; generic chatbots invent deadlines.
- Built
- An advisor that pairs cited explanations with SQL-derived program tables from a maintained catalog — student app plus consultancy admin.
RAG + SQLCitationsStructured Data
Case study →
06 · OCR search
ProductSense AI
- Problem
- Product specs trapped in scanned datasheets that keyword search can’t read.
- Built
- An OCR-to-embeddings pipeline with grounded chat — answers quote the scan and link back to the extracted page, confidence included.
OCREmbeddingsVector Search
Case study →
Services
Fixed scope · you own the code
- RAG applications
- AI chatbots
- AI agent development
- LLM fine-tuning
- Workflow automation
- CRM integration
- Website development
How it works
Fixed scope · no surprises
01
Discovery
You describe the workflow. I ask the awkward questions.
02
Scope
Written scope, tools, and running costs — before any money moves.
03
Build
I build it myself, with demos as it takes shape.
04
Deploy
Shipped with error handling, retries, and logging.
05
Handover
Code, exports, docs — yours. I remove my own access.
Contact
Direct line · no handoffs
Based in Ludhiana, India · working worldwide · every message answered by me, not a team.