AI Automation Specialist Job Description Template (2026)

Full JD (copy-ready)

Paste this into your ATS or careers page. Edit the company name and any bracketed placeholders.

# AI Automation Specialist — Job Description

## Role summary
An AI Automation Specialist wires LLMs into the business: mapping manual ops processes, building workflows in Zapier, Make, or n8n, plugging in OpenAI / Claude / Gemini for triage, extraction, enrichment, and drafting, and shipping internal dashboards in Retool, Airtable, or Notion so operators can see what the automations did. This is an operator-engineer role — mostly no-code and low-code with light Python or TypeScript glue — not a software engineering role and not an ML role.

## Responsibilities
- Shadow manual processes in sales ops, support, finance, or marketing to find the top automation wins by hours saved per week.
- Build Zapier, Make, or n8n workflows connecting HubSpot, Salesforce, Gmail, Slack, Airtable, Notion, and the rest of the SaaS stack.
- Wire OpenAI (GPT-4o, GPT-4o-mini), Anthropic (Claude Sonnet 4.5, Haiku), and Google Gemini into workflow steps for classification, extraction, drafting, and summarization.
- Build custom GPTs, Claude Projects, and OpenAI Assistants API setups for recurring internal tasks (proposal drafting, data lookup, onboarding Q&A).
- Ship document-processing pipelines: invoice and receipt parsing, contract extraction, resume screening — using LLM vision or OCR plus structured prompts.
- Design email triage and auto-response flows with classification, priority scoring, CRM lookup, and escalation to humans for complex threads.
- Run lead enrichment workflows: Apollo / Clearbit / ZoomInfo lookups, LLM-based ICP scoring, and push enriched records back to the CRM.
- Build Retool, Airtable, or Notion dashboards so the team can see automation runs, approve AI decisions, and correct mistakes.
- Design human-in-the-loop review queues for any workflow where an AI decision affects a customer or a dollar.
- Track LLM token spend per workflow against a monthly budget, route cheap queries to small models, and report weekly spend to the sponsor.
- Set up error handling, retries, and Slack alerting on every production workflow; debug and backfill when runs fail.
- Document every workflow in a shared runbook with trigger, steps, owner, error paths, and rollback procedure.

## Must-have skills
- 2+ years shipping production automations on at least one of Zapier, Make, or n8n — with workflows running against real business data.
- Has wired OpenAI, Anthropic, or Gemini APIs into production workflows with structured outputs and a cost tracker.
- Working knowledge of HubSpot, Salesforce, Airtable, Notion, and Slack APIs and webhook patterns.
- Comfortable reading and writing basic Python or JavaScript for the 10% of automation work that no-code cannot handle.
- Prompt engineering fluency: few-shot examples, structured output schemas, JSON mode, system prompts that hold up at scale.
- Understands rate limits, pagination, idempotency, and authentication (OAuth, API keys, service accounts).
- Strong written English — can write SOPs, runbooks, and status updates that ops stakeholders actually read.

## Nice-to-have skills
- Retool or Internal.io for building lightweight internal apps.
- OpenAI Assistants API or Claude Projects for persistent workflow agents.
- Apify or Browse AI for structured web scraping in automations.
- Pipedream or Trigger.dev for code-first automation when no-code platforms run out.
- Basic SQL for reading from Postgres or a warehouse inside workflows.
- Voice automation with Vapi, Retell, or Twilio + Deepgram for call summarization and outbound.

## Tools and technology
- n8n
- Zapier
- Make (Integromat)
- OpenAI / Claude / Gemini APIs
- Retool
- Airtable
- Notion API
- HubSpot / Salesforce
- Pipedream
- Apify

## Reporting structure
Reports to the Head of Operations, RevOps Lead, or Head of Growth depending on org. Works daily with sales, support, and marketing ops owners to understand processes, and partners with IT/engineering when automations need to touch production systems or handle sensitive data.

## Success metrics (KPIs)
- Hours saved per week across shipped automations — tracked with before/after measurement, reported monthly.
- Workflow reliability: greater than 99% successful runs on production automations with alerts on any regression.
- LLM spend within monthly budget per workflow; cost-per-run trending flat or down.
- HITL accuracy: AI suggestions accepted vs corrected rate — stable or improving as prompts iterate.
- Cycle time: median under 5 business days from intake ticket to shipped automation for scoped requests.
- Backlog health: shipped vs requested ratio above 70% quarter-over-quarter.