Description
Summary We are an AI automation agency. We build workflow agents, chatbots and voice agents for business clients. We handle sales and client relationships, you handle the build. We are adding one or two engineers to our delivery bench for ongoing paid project work. The Work:- Most builds are backend workflow agents. Typical scope: - Ingest from a ticketing system, inbox, CRM or form - Classify by type and priority using an LLM - Route or assign based on rules, including time zone and business hours logic - Escalate important cases to a human instead of auto-handling - Log to a CRM or database and notify a team channel Tools we work with: n8n as the main orchestration layer, plus Make and Zapier. LLM APIs including OpenAI, Anthropic and Gemini. Integrations across Jira, Slack, Microsoft Teams, Google Workspace, Asana, HubSpot, GHL and Zoho. We also want people who can work outside low-code tools with custom code or direct API work when a client needs it. Chatbot or voice agent experience is a plus, not a requirement. Show us something you have already built. Send a walkthrough video or live link of a workflow agent you have delivered, ideally one that classifies, routes or triages incoming work. Ticket triage, support desk automation, lead routing or inbox handling are all relevant. Tell us what it does, what it runs on, and what it changed for the client. If you have built something close to support ticket triage, meaning classifying tickets by type and priority, handling routine ones automatically and escalating urgent ones to a human, say so clearly. That is the closest match to the work coming up. What matters most:- - Reliability over cleverness. Builds that hold up in production with real client data - Sensible guardrails. The agent escalates when it is unsure rather than guessing - Self-hosted deployments. Several clients need the system running on their own infrastructure - Comfort with technical clients. Some have their own engineering teams and will question your architecture and data handling - Clear turnaround and communication. We commit to client timelines and need to trust your dates - Ability to record a clean screen walkthrough. We show these to clients Questions: - 1. What is the closest thing you have built to ticket triage or routing? Link or describe it 2. How do you keep classification accurate, and how do you stop an agent acting on something it should have escalated? 3. Have you deployed self-hosted n8n on a client's own infrastructure? Describe the setup 4. When you call an LLM to classify content containing customer data, where does that data go, and how do you handle clients with strict data residency or privacy requirements? 5. When would you build with custom code rather than a low-code tool? Give an example 6. What is your current capacity, how many projects do you run at once, and what is your typical turnaround for this type of build? Proposals without a work sample and answers to these questions will not be reviewed. Strong fits go on our delivery bench and get first refusal on incoming projects. We have live work now, so we are moving quickly.