Gumbo Agent Wrangler Remote · Contractor Company website

Apply for Agent Wrangler

Building the future of work by making AI multiplayer. Strategy, design, and technology for humans and agents, cooking side by side.

Role Summary

We’re looking for an Agent Wrangler to own the messy middle between AI potential and actual operational value. This person will help design, run, evaluate, and improve AI-agent workflows across internal operations, product development, research, and client delivery.

This is part operator, part systems thinker, part prompt/tool/workflow designer. The right person is excited by making agents useful in the real world — not just demoing them, but building the human + process + tooling layer that makes them dependable.

What You’ll Do

- Design and run agent-based workflows for research, execution, documentation, and delivery

- Turn recurring team work into reusable AI-assisted processes, playbooks, and operating patterns

- Improve agent reliability through better context, memory, prompts, routing, tooling, and review loops

- Help define where agents should be autonomous, where they need human review, and where they should not be used

- Work across product, ops, engineering, and client-facing teams to identify high-leverage use cases

- Build the connective tissue around agents: documentation, evaluation criteria, fallback paths, handoffs, and QA

- Help maintain durable team context so agents can work from shared knowledge instead of one-off chats

- Pressure-test new tooling and workflows, then operationalize the ones that actually help

What We’re Looking For

- Strong systems and operational thinking

- Experience working with LLMs, agents, automation tools, or AI-assisted workflows in practice

- Comfort with ambiguity, experimentation, and rapid iteration

- Excellent written communication and documentation habits

- Ability to debug broken workflows across people, prompts, tools, and process

- Good judgment about reliability, risk, and when human oversight matters

- Strong operator instincts: you notice where work falls apart and fix the process, not just the symptom

Nice to Have

- Light technical ability: scripting, APIs, structured data, prompt design, or workflow automation

- Experience in product ops, technical program management, solutions engineering, or AI operations

- Experience designing evaluation loops or review systems for AI outputs

- Familiarity with engineering workflows, project management systems, or knowledge-base tooling

What Success Looks Like

- Agents save real time on real work, not just in demos

- Team knowledge compounds instead of disappearing into chats

- AI-assisted workflows become more reliable, reviewable, and repeatable

- Humans spend less time wrangling chaos and more time making decisions