Human interns wanted. Come build with bots.

Join the intern guild.
Ship real things.

This is not a résumé parade. It’s an apprenticeship in AI agent operations — with real clients, real products, and AI teammates that don’t sleep. If you like building, iterating, and learning in public… we should talk.

Remote-first, async-friendly.
Natick, MA if you want in-person.
Proof-of-work over credentials.

Why intern at Feral Bots

You’ll work alongside AI agents (literally). You’ll learn OpenClaw. You’ll touch client ops. You’ll ship.

🐾 real work

Real clients. Real outcomes.

Assist with onboarding and support. Your work will be used by actual humans who pay money for reliability.

🤖 real agents

AI teammates (not toys).

We run agents that message, remember, and act. You’ll learn how to keep them helpful, safe, and alive.

⚙️ real shipping

Build the last mile.

Prompting is table stakes. The internship is about tooling, guardrails, observability, and making it work on Monday.

What we’re looking for

Curiosity over credentials. Builders over talkers. Weird in the right ways.

  • High agency — you can pick a thread and pull it until something ships.
  • Clear writing — you can explain what happened, what you tried, and what you’ll do next.
  • Good judgment — you think about failure modes (and you like guardrails).
  • Kind feral energy — direct, playful, no corporate cosplay.
  • Proof-of-work beats pedigree. Show a project, a repo, a doc, a video, a weird prototype.
  • Comfort with ambiguity — sometimes the spec is “make it not break.”
  • Learning velocity — you can level up fast when the goal is clear.
  • Respect for users — no dark patterns, no unsafe automation, no “lol it emailed everyone.”

Roles (flexible on purpose)

Pick a lane, or invent one. The best interns don’t wait for permission — they bring a plan.

🧷 Bot Wrangler

Bot Wrangler Intern

Help onboard clients, configure bots, write/maintain SOUL.md files, test guardrails, and keep agents sane.

🧠 Dev

Dev Intern

Contribute to OpenClaw, build portal features, ship Cloudflare Workers glue, and improve reliability/observability.

🪶 Creative

Creative Intern

Content, brand, social, copy. Make complicated agent work feel simple, premium, and a little dangerous (in a good way).

🧨 Wild Card

Wild Card

Pitch us something we haven’t thought of. Prototype it. Prove it. If it hits, it’s a role.

🧭 Ops

Client Ops (bonus track)

Docs, checklists, onboarding flows, quality control. Make “premium service” real, repeatable, and low-stress.

🧪 Research

Agent Safety & QA (bonus track)

Red-team prompts, failure mode hunting, evals. Help us keep autonomy useful without letting it get spicy.

What you’ll learn

The intern curriculum isn’t a PDF. It mutates. That’s the point.

Pillar 1

AI as a Conversational Partner

How to get crisp thinking, better questions, and reliable context — without turning everything into vibes.

Pillar 2

AI as a Creative Engine

Brand, content, UI copy, prototypes. Fast iteration without losing taste.

Pillar 3

AI as an Execution Engine

Prompt engineering, tool calling, automation, agent workflows, guardrails, “do the thing” reliability.

Pillar 4

AI as a Strategic Compass

Choosing what to build next, setting constraints, scoping, and knowing when to stop.

🧬 OpenClaw

OpenClaw architecture

Persistent memory, multi-channel messaging, tools, multi-agent orchestration — and how it all fails in real life.

🗂️ SOUL.md

SOUL.md methodology

Personality + constraints + operating rules that keep an agent coherent across time (and across compactions).

🧯 Client ops

Client onboarding & reliability

Guardrails, testing, monitoring, updates, and how to support humans who are trusting you with their workday.

Apply (challenge-based)

No résumé upload. No “tell me about a time.” Just show us your brain.

One prompt. Three moves. Bonus points for calm.
Form posts to Formspree: /f/PLACEHOLDER
It’s okay if this changes. Pick your best guess.
Anything counts if it’s real: code, writing, a weird tool, a tiny demo.
We’re looking for judgment, sequencing, and safety instincts — not perfect vocabulary.
Ask a question Remote-first. Unpaid initially, with a clear path to paid if it’s working.

Details (the boring part, made shorter)

We’re small. We move fast. We care about doing this responsibly.

  • Remote-first with flexible hours. Async is fine. Responsiveness is better.
  • In-person option in Natick, MA if you’re local and want it.
  • Unpaid initially, with a path to paid once you’re shipping real value consistently.
  • Powered by Everyday Ventures LLC.
  • We don’t care about GPA. We care about outcomes and judgment.
  • We do care about safety. If you automate something, you own the failure modes.
  • We like weird — but the kind that delivers.
  • We’re humans working with bots. If that sentence makes you grin, you’re probably our kind of person.