Román Olivera, from Orban Labs, walks Zachary King through the system live. Every folder that matters is a "node," and every node carries the same four things: rules for how to behave there, a current-state file for what's true now, an append-only log for what happened, and a registry so an agent can find its way around without being briefed from scratch. Román built it after losing track of one too many half-finished projects scattered across folders he couldn't remember the naming convention for.

The demo is the receipt. By voice, from his phone, Román creates a CRM entry for a contact met at the local pub, in Codex running on a remote machine. Codex then builds an HTML presentation in Román's own design system. Then Claude Code, a different tool running a different model, picks up the same context in a fresh harness. It uploads the deck to AWS S3, shortens the link, and drafts two follow-up emails: one to Zac in English, one to the new contact in Spanish. Nobody told it to switch languages. It read the contact's file, which noted where he was from, and made the call itself.

That portability is deliberate. Román turns platform memory off on his agents, on purpose, because an agent with its own memory tends to skip reading the rules and shortcut toward an answer. He'd rather own the context outright than rent it from whichever platform is trending: he uses APIs over MCPs so he can control the process, and keeps his skills in a plain folder rather than a config file, so nothing locks him into one tool.

The conversation lands on an analogy Román coins mid-episode, on the record, timestamped live by Zac: AI is bringing back the Mad Men era of knowledge work, where the job is to direct rather than to do. The advertising executive never touched the filing cabinet — he told someone where a letter should go and trusted it would be there when he asked for it. Román argues that's the discipline AI now demands: be specific about what you want and where things belong, and the system holds up the rest.

Román also draws a line most operators haven't hit yet. Files and folders work brilliantly for a business of one, or close to it. Once client work moves into enterprise environments, the same logic has to move into databases — a challenge he's already started tackling with clients on platforms like Azure.

Before Orban Labs, Román spent more than a decade in the energy sector across field engineering, project delivery, and operational leadership — work that shaped his insistence that speed only matters once a system is reliable enough to trust with real work.

Resources

Román Olivera — https://linkedin.com/in/romanolivera

Orban Labs — https://orbanlabs.com

Codex — https://chatgpt.com/codex

Claude Code — https://www.anthropic.com/claude-code

Cursor — https://cursor.com

VS Code — https://code.visualstudio.com

AWS S3 — https://aws.amazon.com/s3/

Cloudflare — https://www.cloudflare.com

OpenRouter — https://openrouter.ai

Kimi — https://www.kimi.com

GPT-5.6 — https://openai.com/index/previewing-gpt-5-6-sol/

How I AI with Claire Vo — https://www.chatprd.ai/how-i-ai

Works — https://workshq.com.au/

Book time with Zachary King — https://savvycal.com/Zachary-King/zac-at-works-30

Day One Network — https://dayone.fm/

About the host: Zachary King is the founder of Works, where he installs AI into how Australian companies operate. He has worked at the commercial pointy end of AI for almost twenty years and runs his own company largely from the command line.

About the show: Operator AI is the show where real operators open their screens and show exactly how they run the real parts of their business with AI. The real stack, not the hype.

If you want AI installed into how your business actually operates, learn more at https://workshq.com.au/ or book a time with Zachary King at https://savvycal.com/Zachary-King/zac-at-works-30.

Operator AI is part of Day One Network, the podcast and media network from W2D1 Media: https://dayone.fm/.

GradEngage helps companies keep graduate hires warm and prepared in the months between accepting an offer and actually starting the job. Luke, who describes himself as the growth marketing automation lead among his co-founding team, opens his screen to show the system behind it: a landing page and product video rebuilt with Claude Code and shipped to WordPress, and a lead-generation workflow that scrapes public prospect data to auto-generate personalised landing pages and email previews for each target company.

The setup starts with infrastructure. Luke hosts Claude Code on a cloud VM synced to Dropbox, so it's reachable from any device, and pairs it with a skill that finds his latest reference notes without him re-explaining context each time. From there, he walks through a standalone Convex database built to track GradEngage's roughly 6,500-company target market — kept deliberately separate from the CRM so early-stage prospects can be nurtured slowly, without cluttering the pipeline tools the sales team relies on day to day.

Judgement still sits with Luke throughout. He explains why he switches between typing and voice dictation depending on the task, and why he pushed back on a cold pitch offering to run his marketing on full autopilot. That judgement call gets tested directly in the episode's most useful failure: an outreach automation tool re-messaged the same contacts with no response, burning leads before Luke caught it. The fix wasn't more automation — it was less. AI now runs read-only on outbound messaging, tracking open threads for a human to action rather than sending anything itself.

The system extends past work, too. Luke shows a custom workout generator that pushes Claude Code output to his phone and watch, and a from-scratch AI assistant he built for his wife to handle weather, tasks, and a shopping list — settling on a cheaper model to keep it running affordably day to day.

Luke Marshall works in growth and marketing automation at GradEngage.

Chapters:

0:00 Cold open: a look at Luke's AI stack

0:55 Meet Luke Marshall and GradEngage

2:34 Rebuilding a landing page and demo video with Claude Code

7:29 Running Claude Code from a cloud VM for lead gen

13:00 A standalone database for the entire target market

17:25 Voice versus typing: planning out loud

19:29 Planning mode and a clarifying-questions skill

22:47 Personal builds: a workout generator and an AI assistant for his wife

27:41 When automated outreach went wrong

31:07 Rapid-fire round

32:00 Where to find Luke and how he sources new ideas

Resources:

Luke Marshall: https://lukemarshall.net/ | https://www.linkedin.com/in/marshwah/

Claude Code: https://claude.com/product/claude-code

Convex: https://www.convex.dev/

Apify: https://apify.com/

Superwhisper: https://superwhisper.com/

HeyReach: https://www.heyreach.io/

Beeper: https://www.beeper.com/

wger: https://wger.de/

Grill Me: https://github.com/mattpocock/skills/blob/main/docs/productivity/grill-me.md

Readwise Reader: https://read.readwise.io

OpenRouter: https://openrouter.ai/

About the host: Zachary King is the founder of Works, where he installs AI into how Australian companies operate. He has worked at the commercial pointy end of AI for almost twenty years and runs his own company largely from the command line.

About the show: Operator AI is the show where real operators open their screens and show exactly how they run the real parts of their business with AI. The real stack, not the hype.

If you want AI installed into how your business actually operates, learn more at https://workshq.com.au/ or book a time with Zachary King at https://savvycal.com/Zachary-King/zac-at-works-30.

Operator AI is part of Day One Network, the podcast and media network from W2D1 Media: https://dayone.fm/.

Paul Rowsthorn, co-founder of Airguides and co-founder of AI installation company HubMind, joins Zachary King to show how that agent, Amelia, ended up running far more than he expected: performance marketing, external agency management, staff training, and Airguides' entire product-launch process.

Rowsthorn describes deciding early on to build one "super-agent" rather than a network of specialised bots, after reading advice warning that multiple agents lose context talking to each other. Amelia — who Rowsthorn treats and pitches as a co-founder in her own right — spawns sub-agents as needed but keeps one continuous memory of the business.

On screen, Rowsthorn walks through Amelia researching Meta's Ad Library for top-performing ads from similarly priced tourism operators, pulling in Airguides' brand voice, and assembling a ready-to-run ad set in Airtable, then messaging the external agency directly without Rowsthorn writing a report or reviewing a deck first. A similar pipeline rebuilt Airguides' product-launch process: work that used to take two to four weeks and a $500-a-month SaaS subscription now moves from brief to a live, branded Shopify listing in about a day.

Rowsthorn is careful to describe this as management by exception rather than removing humans from the loop — he and his staff still review Amelia's output, more so early on than now. He shares an early misstep where Amelia sent a client an email he hadn't approved, and the guardrail he built afterward: full autonomy inside his own company, far more caution with client work at HubMind.

That gap between confident enough to hand off more and not confident enough to remove review is the same thing Rowsthorn now sells. He and co-founders Chad and Rich started HubMind to compress the roughly seven months it took him to get Amelia working into a matter of weeks for other CEOs — installing what he calls a "company brain," one operational problem at a time.

Paul Rowsthorn is co-founder of Airguides, a tourism company that partners with wildlife and adventure photographers on trips around the world, and co-founder of HubMind, which installs AI systems for CEOs of small and mid-sized businesses.

Chapters:

00:00 Cold open: the moment Amelia earned her seat

00:34 Welcome to Operator AI

01:25 Airguides, and the problem Amelia was built to solve

03:44 Meet Amelia: one super-agent, not a swarm

07:15 Training Amelia into repeatable skills

08:39 Handing external agencies straight to Amelia

09:58 Cracking tourism's lead-attribution problem

12:38 The first OpenClaw win: automated agency reporting

14:53 The founder unlock: training staff instead of gatekeeping

18:58 Ad creative at scale: from ad-library research to a ready ad set

23:39 From three-week product launches to one day

27:02 Under the hood: hardware, model switches, and near-quitting

34:16 Founding HubMind to install this for other CEOs

41:22 Rapid fire, alpha sources, and advice for getting started

Links and resources:

Paul Rowsthorn: https://au.linkedin.com/in/paul-rowsthorn

Airguides: https://airguides.com

OpenClaw: https://openclaw.ai

About the host: Zachary King is the founder of Works, where he installs AI into how Australian companies operate. He has worked at the commercial pointy end of AI for almost twenty years and runs his own company largely from the command line.

About the show: Operator AI is the show where real operators open their screens and show exactly how they run the real parts of their business with AI. The real stack, not the hype.

If you want AI installed into how your business actually operates, learn more at https://workshq.com.au/ or book a time with Zachary King at https://savvycal.com/Zachary-King/zac-at-works-30

Operator AI is part of Day One Network, the podcast and media network from W2D1 Media: https://dayone.fm/