AI-Driven n8n Workflow Automation
Budget: ₹12,500 – ₹37,500 INR
I need production-ready n8n workflows that use Claude (Anthropic) or any other fit-for-purpose LLM to automate one core business goal: streamlining customer service requests that touch our finance operations. We also
The workflows must be able to:
• Ingest the incoming request—whether it arrives by email, form, or chat—identify it as finance-related, and extract the key facts.
• Send the request through an LLM prompt chain for classification, summarisation, and suggested next actions.
• Trigger follow-up steps inside n8n (routing to the right teammate, updating our ticketing system, posting a Slack/Microsoft Teams message, or calling an external API) with graceful error handling and clear logging.
Acceptance criteria
• Deployed on our self-hosted n8n instance with version-controlled JSON files.
• Clear, reproducible environment variables, node credentials, and test data so engineering can run locally and in staging.
• Inline comments plus a short README that explains each step, the LLM prompts, and how to retrain or swap models later.
You will work directly with Founders & Product Owner and liaise with finance ops and engineering as needed. Quick iteration, clean code, and thoughtful prompt design are more important than flashy UI. If you already speak TypeScript, Postgres, or have experience tuning Claude for enterprise workflows, that’s a bonus we can tap into.
The workflows must be able to:
• Ingest the incoming request—whether it arrives by email, form, or chat—identify it as finance-related, and extract the key facts.
• Send the request through an LLM prompt chain for classification, summarisation, and suggested next actions.
• Trigger follow-up steps inside n8n (routing to the right teammate, updating our ticketing system, posting a Slack/Microsoft Teams message, or calling an external API) with graceful error handling and clear logging.
Acceptance criteria
• Deployed on our self-hosted n8n instance with version-controlled JSON files.
• Clear, reproducible environment variables, node credentials, and test data so engineering can run locally and in staging.
• Inline comments plus a short README that explains each step, the LLM prompts, and how to retrain or swap models later.
You will work directly with Founders & Product Owner and liaise with finance ops and engineering as needed. Quick iteration, clean code, and thoughtful prompt design are more important than flashy UI. If you already speak TypeScript, Postgres, or have experience tuning Claude for enterprise workflows, that’s a bonus we can tap into.