AI Skill & Automation Specialist

Hotwire is the pre-eminent global tech communications and marketing consultancy. We are a group of expert communicators and marketers who are fanatical about tech and here to help businesses ignite their possibilities. Our local experience and global scale enable us to deliver bold yet pragmatic guidance to tech companies worldwide.

This role sits within Hotwire's AI Lab, but its remit spans the wider Enero Group: Hotwire, BMF, Orchard, and ROI·DNA. The skills you build will be used by colleagues across all four agencies. We have been named a Best Place to Work by PR Week, PR News, and Crain's New York year after year.

The AI Skill & Automation Specialist is the builder behind the Group Skill Library and its automation layer. They take a raw requirement from an agency delivery team (a client brief, a media analysis, a pitch deck assembly) and turn it into a well-structured, well-documented, production-ready Skill or automation, built in Claude, Microsoft Power Automate, or Copilot, that others can trigger reliably without knowing how it was made. This is a hands-on craft role, not a research or strategy role.

The work demands a logical mindset. You write for a machine that takes you literally, so you must specify precisely. You resist treating fluent output as correct output, so you evaluate rather than admire. And you make a real judgment call, early, about whether a skill or automation is worth building at all, and which platform is right for it. Not every workflow deserves one. The Specialist works alongside the Claude Specialist for Skills and Prompt Engineering, who sets library standards; the Specialist ships the skills and automations that populate the library.

• Build production-grade Claude Skills end-to-end, from an SRD (Skill Requirement Document) or agency brief through to a packaged, tested, documented skill in the Group Library

• Make the build-or-don't-build call early, and pick the right platform for it, a Claude Skill, a Power Automate flow, or a Copilot extension. Not every workflow deserves one; sometimes a prompt or a template does the job better

• Author clear, trigger-accurate SKILL.md files with precise descriptions, correct YAML frontmatter, and progressive disclosure across references/ and scripts/

• Break complex workflows into the right skill structure and chaining logic

• Design guardrails, context systems, and required references/schemas/connectors to avoid guessing/hallucinations

• Create supporting assets like schemas, examples, prompt templates, and light Python utilities

• Test for trigger accuracy, quality, edge cases, and failure modes; iterate and version properly

• Turn agency briefs into clear technical specs (without losing intent) and user-friendly documentation

• Stronghold on prompt engineering (Structured prompting, reasoning, and evaluation) and context engineering (design the reference files, input schemas, retrieval patterns, and progressive disclosure)

• Hands-on Claude experience - Claude.ai, Projects, API, or CoWork. Candidates must be prepared to showcase skills, prompts, or workflows they have built

• Hands-on experience building automations in Microsoft Power Automate (flows, connectors, approvals) and Microsoft Copilot (Copilot Studio agents or Copilot extensions), in addition to Claude. Candidates must be prepared to showcase flows or agents they have built

• Strong evaluation mindset: check what the skill produced against what it was supposed to produce

• Working knowledge of how LLMs fail (context degradation, spec drift, tool selection error, hallucinated structure) and how to design around these patterns

• Excellent written English and a collaborative style, with clear communication for both LLMs and agency teams

• Comfort with light Python scripting for helper scripts inside skills. Not a hard requirement, but useful for the moments where a skill needs a small utility to preprocess or validate

• Familiarity with code review practices. You do not need to ship production code, but being able to read a pull request and spot obvious issues is a plus

• Working comfort with Git, YAML, and Markdown. Standard tools of the craft; easy to pick up if not already familiar

• Prior experience building and shipping Claude Skills, including packaging and trigger tuning

• Familiarity with MCP, Claude Connectors, or other agentic tooling patterns

• Exposure to LLM evaluation frameworks such as eval suites, rubrics, or benchmark design

• Public portfolio: open-source skills, prompt libraries, or published prompt engineering work

· Degree in engineering, computer science, or a related discipline (preferred)

· Strong portfolio of skills, prompts, or AI workflows valued more than formal education

· 2–3 years in marketing, PR, advertising, or creative agency is a strong plus

· Agency understanding of briefs, deliverables, and client work is beneficial.

✦ Prompt Engineering (Advanced)

✦ Context Engineering

✦ SRD & Specification Writing

✦ Claude Ecosystem Fluency

✦ Systems Thinking

✦ SKILL.md Authoring & Packaging

✦ Output Evaluation Discipline

✦ Guardrail & Scope Design

✦ LLM Failure Mode Literacy

✦ Build-vs-Don't-Build Judgement

✦ Power Automate & Copilot Fluency

✦ Multi-Platform Automation Judgement

· Flexibility. Life doesn't always fit around a 9-to-5; we are serious about work/life integration, and this role is fully remote

· Global reach and opportunity. You will work with teams across Australia, Singapore, the UK, Europe, and the US, and your skills will be used by colleagues on every continent

· Strong benefits. Paid time off, holidays, sick leave, cultural celebration days, paid time off for community involvement

· Development opportunities. A strong learning and development programme with training available at every level

· Autonomy. The freedom and support to be limitless, and the opportunity to own your career and create the future you want

· Variety. No day is the same. You will navigate change daily across different agencies, clients, and problem types

· Our team. Connected, ambitious, and friendly. We don't step on toes to get ahead; we build each other up

We engage in Thoughtful Working, empowering our people to determine the best way to work for our clients, our team, and Hotwire every single day