Agentic AI Engineer — Build a CRM-Connected Chatbot with Composio

We're building an agentic AI chatbot that talks to our CRM through Composio (150+ CRM/tool integrations) to look up customer data, update records, and answer customer questions in real time. We're looking for an experienced engineer who has actually built and shipped agentic AI systems with Composio and CRM tools — not someone learning it for the first time on our project. This is an hourly, ongoing engagement. We want a reliable long-term partner who can help us architect the system correctly from day one, build it, and keep improving it. WHAT THE CHATBOT NEEDS TO DO - Read CRM data: fetch contacts, deals, tickets, order history, and account info to answer customer questions accurately - Create and update records: log conversations, update contact fields, create tickets/tasks, and write activity back to the CRM - Customer-facing support: deployed to our customers as a support/sales assistant (must be reliable, safe, and on-brand) - Work across channels: website chat widget, WhatsApp/SMS, and Slack/Teams THE STACK - Composio for CRM and tool integrations (required — you must have hands-on Composio experience) - An LLM/agent layer (OpenAI, Anthropic, or similar) with function/tool calling - An agent framework of your recommendation (e.g. LangGraph, LlamaIndex, CrewAI, or a lightweight custom loop) — tell us what you'd use and why WHAT WE NEED HELP WITH 1. Architecture & setup: design the right agentic architecture — how the agent plans, calls CRM tools via Composio, handles auth, and returns safe, accurate answers to customers 2. Build the integration: connect the chatbot to our CRM through Composio for both read and write actions 3. Guardrails & accuracy: make sure the bot doesn't hallucinate CRM data, respects permissions, and confirms before making changes 4. Multi-channel deployment: website widget, WhatsApp/SMS, and Slack/Teams 5. Ongoing iteration: this is long-term; we'll keep adding capabilities and channels REQUIRED EXPERIENCE (please only apply if you have this) - Proven, hands-on experience with Composio — please describe a specific project where you used it - Built and shipped agentic AI / LLM agents with tool/function calling (not just simple prompt chatbots) - CRM integration experience (HubSpot, Salesforce, Zoho, or similar via Composio) - Strong understanding of designing safe, reliable customer-facing AI with guardrails - Comfortable with API auth, webhooks, and connecting multiple channels - Clear communicator who can explain trade-offs in plain language NICE TO HAVE - Experience deploying chat to WhatsApp Business API, website widgets, and Slack/Teams - Prompt engineering and evaluation/testing of agent behavior - Experience with vector databases / RAG if we later add knowledge-base answers TO APPLY, PLEASE ANSWER THESE (applications without answers will be ignored) 1. Share a specific example where you built an agentic AI system using Composio connected to a CRM. What did it do, and what was the outcome? 2. What agent framework would you recommend for this project and why? 3. How do you prevent the bot from hallucinating or giving customers wrong CRM data? 4. How would you handle write actions (updating/creating records) safely in a customer-facing bot? 5. Your hourly rate, weekly availability, and time zone. We'd like to begin with a short paid call to review architecture before the full build. We're prioritizing depth of real Composio + agentic AI experience over the lowest bid. If you've genuinely built this before, we want to talk.