Shahan A.

I build AI agents, voice AI agents and n8n automations for startups, and I keep them running in production. My team and I handle the whole product: the agent, the backend, the web app and the mobile app, from the first call to launch. 10+ years in software engineering, now focused on agentic AI and AI-native development. What we have built and run Bookerise: a booking marketplace with an AI assistant that books appointments into business calendars. Next.js, Node.js, Supabase, LangGraph, OpenAI embeddings, AWS. Voice booking agent: talks to customers, answers from a knowledge base, books into Google Calendar and updates the CRM. We started on OpenAI Realtime and moved to our own full-duplex voice model, GPT-Live-1, when costs got too high in real use. Self-healing OpenClaw setup: OpenClaw running on Hetzner and AWS, with Hermes Agent watching it and bringing the containers back up when it goes down. AI content pipeline in n8n: daily news research, AI drafting with a validation step, human review, then auto publishing. It runs on our own site and inside EasyRep. Mobile apps: EasyRep on Android (React Native, with AI features), Muhassin and Muhassin Law on iOS, Armor POS and Unlocked Academy. AI agents and multi-agent systems Autonomous agents that reason, call tools and finish multi-step tasks. LangGraph, LangChain, CrewAI, AutoGen, MCP (Model Context Protocol) servers, tool calling, ReAct and plan-and-execute patterns. OpenClaw and Hermes Agent OpenClaw setup, configuration, MCP tools and Docker deployment on AWS or Hetzner, with monitoring and auto recovery through Hermes Agent. Voice AI agents AI voice assistants and AI receptionists for booking and customer support. OpenAI Realtime API, Deepgram, ElevenLabs, Gemini Live, LiveKit and GPT-Live-1, with RAG, memory, CRM and calendar integration. Safe agents with human approval An approval layer built on Jev: every agent action is scored for risk and whether it can be undone. Safe actions run, bigger ones wait for a human yes, dangerous ones are blocked. AI automation and n8n workflows Agent-triggered n8n workflows for research, content, lead generation and CRM updates (HubSpot), with human review where it matters. LLMs, RAG and AI chatbots OpenAI GPT, Claude, Gemini, Groq, Mistral and local models with Ollama. RAG chatbots and semantic search on pgvector, Pinecone and ChromaDB. Self-hosted LLMs and small language models on GPU servers. Full stack web and mobile development SaaS products and MVPs with Next.js, React, Node.js, NestJS, Python and FastAPI. Mobile apps with React Native and Flutter. PostgreSQL, Supabase, Docker, AWS and GCP. Integrations with Stripe, Google, Microsoft Graph, Zoom and HubSpot. How we work Daily progress updates, clear scope and support after delivery. We use Claude Code, Codex and Cursor to move fast, and we prefer long-term work with startups and product teams. Tell me what you want to build or automate, and I will tell you honestly whether AI agents are the right call and how we would approach it. 100% Job Success | Top Rated Agency | CTO, ArmorTech