Skip to content
Scott McQueen

AI implementation · Workflow automation · Operational transformation

Twenty years running complex operations. Now I build the AI systems that change how the work gets done.

I spent more than twenty years leading security and operations for Amazon and Brink's across Asia Pacific, the Middle East, Europe and North America.

Today I design and implement AI-enabled workflows for businesses and for my own products. I start with the operating problem, build the system around it, test it against real conditions and stay with it until people use it every day.

See selected workLinkedIn
Scott McQueen
Scott McQueen, Melbourne, Australia
400+ sitesin one global Digital Operations Center (DOC) I drove at Amazon100+ managers, 3 regionsmoved onto one finance workflow at AmazonUS$3.6M a yearsaved by redesigning the Pacific Northwest escort workflow at Amazon3,987 dictationsthrough my own self-improving AI dictation system in 30 days, to 1 Oct 2026

What I do

  1. Step 1

    Understand the operation

    Map what is really happening: where work stalls, where information gets lost and what the people doing it need.

  2. Step 2

    Design the workflow

    Decide the future process: where AI belongs, where plain automation belongs and where a person must stay in control.

  3. Step 3

    Build and integrate

    Turn it into a working system with AI, software, APIs and the platforms the business already runs on.

  4. Step 4

    Make it stick

    Test it, measure it, fix how it fails, and work with the people using it until it's part of normal operations.

Transformation at enterprise scale

The transformation work came before generative AI. At Amazon I was already redesigning workflows, operating models, data capture, operations reporting and governance. Technical teams built the software. I defined the problem, drove the requirements and got it into use.

Digital Operations Center (DOC)

Turning 400+ sites of operational data into one global decision system.

A Digital Operations Center (DOC) for Corporate Security, one view from a single site up to the Chief Security Officer, so failing systems are found fast and decisions rest on the same numbers.

400+ corporate sites in one Digital Operations Center

Read the case study

Before

  1. Sites compile monthly reports by hand
  2. Each region assembles its own view
  3. Leaders review late, inconsistent numbers

After

  1. Access, CCTV, alarm, incident and audit data flow in
  2. One view: global, region, cluster, site
  3. Thresholds raise alerts and managers act daily

Global security finance workflow

Replacing fragmented regional finance processes with one controlled workflow.

More than 100 security managers in three regions moved from spreadsheets and email onto one finance operating model, with regional and global portals behind it.

100+ security managers across three regions on one workflow

Read the case study

Before

  1. 100+ managers send figures on regional templates
  2. Headquarters copies and pastes them together
  3. Forecasts arrive late and inconsistent

After

  1. One operating model agreed across three regions
  2. Regional and global portals with standard inputs
  3. Automated consolidation and monthly reporting

Vendor escort operating model

Making an unmeasured workflow measurable, then redesigning it.

Security escorts for vendors in Amazon's Pacific Northwest buildings went from phone calls nobody could measure to a structured workflow, saving US$3.6M a year in guard costs.

US$3.6M a year in Pacific Northwest guard-force savings

Read the case study

Before

  1. Teams phone for guards separately
  2. Only the start of an escort is logged
  3. Extra guards on 4 or 8 hour minimums

After

  1. Requests through one Slack workflow, bot and form
  2. Start, end, location and secure areas recorded
  3. Demand combined, guards reused, staffing resized

AI systems I build now

I build with AI coding tools, mainly Claude Code. I define the problem, the workflow, the system design, the acceptance criteria and the controls. I direct and review the build, push back on technical constraints, test against real conditions, decide what ships and work with the users until it's adopted.

HeartTold · Co-founder · Conversational AI product

Building an AI system around long-term human memory.

HeartTold interviews people over time by phone or in the browser, asks follow-up questions from what they've already said, preserves their stories and makes them searchable later.

  • 1,000+ h of my own time on the product
  • 674 commits from March to September 2026, 575 of them mine
  • 284 automated back-end tests
Read the case study
Home page of hearttold.com

Verbatim and DayLog · My AI chief of staff

An AI that hears the conversation, sees the screen and follows up.

Verbatim began as a private meeting recorder on my Mac. I added screen capture and live call coaching, so AI can combine what's being said with what's on screen and guide me while the call is happening. DayLog carries it through the rest of my working day.

  • 88 meetings recorded, transcribed and summarised
  • 88.6 h of conversation transcribed
  • 11,191 screen captures kept alongside 70 meetings
Read the case study

Before

  1. Notes typed while trying to listen
  2. Detail lost, actions remembered or not
  3. Follow-up written by hand afterwards

After

  1. Both sides recorded on my Mac, with screen captures
  2. During the call, AI suggests what to raise next
  3. Recap, owners and next agenda go out automatically

AI compliance platform · Lead product and AI engineer · Pre-launch

Designing controlled AI workflows for a regulated environment.

One controlled record of a firm's obligations, evidence and approvals, built privacy-first, with every finding traced to its source and every rule tested on history before release.

  • 10,896 real emails replayed before a single flag was released
  • 0 leaks across 200 real messages and 1.6 million characters through the pseudonymiser
  • 147 checks passing across 5 suites, 19 of them deliberate-break tests

Pre-launch and under NDA. Details are limited to what's shown here.

Read the case study

Before

  1. Scope, fees and promises live in email
  2. Drift is noticed late
  3. Evidence is hard to produce on request

After

  1. One record built from the signed agreement
  2. Checks that quote their source and show the arithmetic
  3. Overdue actions raised, evidence ready

Other systems shipped

Client work goes through my consultancy, LuminHive Labs. The rest are my own products and businesses.

How I put AI into real operations

Start with the workflow, not the model

I don't automate a bad process. The US$3.6M escort saving at Amazon came from measuring and redesigning the workflow first.

Measure before release

Before releasing a compliance flag, I replayed it over 10,896 real emails. One rule would have fired on about 1 in 6 outgoing messages, so it was redesigned before anyone saw it.

Keep people accountable

AI drafts, advises, analyses and flags. A person approves anything material. The car-detailing follow-up texts stop once someone from the business has phoned the customer.

Build for privacy and control

My dictation, meeting and DayLog transcription runs on my own machines. The compliance platform codes names and contact details before any text leaves the machine.

Design for failure

I ask what happens when the model is wrong, an API fails or the data is incomplete. Verbatim's microphone track was quietly losing 1% of its audio. Every recording now reports how much it kept.

Measure adoption

A tool nobody uses is a demo. 3,987 dictations in 30 days, and Verbatim on every call I take.

Recently shipped

The experience behind the build

More than twenty years running security and operations at Amazon and Brink's. The work was the same as it is now: understand a complex environment, find where it fails, redesign the process, set the controls, bring the stakeholders along, measure the result and make sure people use it.

Amazon

2018 to 2026

  • Global Operations Manager, Amazon Corporate Security, Seattle
  • Head of Corporate Security, Asia Pacific and Middle East, Tokyo
  • Regional and Area Security Manager, Tokyo

Brink's Global Services

2004 to 2018

  • Head of Operations and Security, Singapore
  • Director, Security, New York
  • Director, Global Operations and Security, London
  • Senior Manager, Risk, Security and Compliance, Japan

Highlights

  • Managed a capital and operating budget of over US$100M across more than 20 countries.
  • Point of contact for a US$380M global guard-force procurement.
  • Ran the return-to-office security strategy for more than 450,000 employees, briefing the Chief Security Officer daily.
  • An access-control transformation with projected savings of more than US$15M a year.
  • The 400+ site Digital Operations Center, the three-region finance workflow and the US$3.6M escort redesign above.

Education

  • MBA, Australian Institute of Business
  • BA, Language and Linguistics (Japanese), Griffith University
  • Certified Chief of Staff, Chief of Staff Association
  • Speaks Japanese

Technical capability

AI and models
Claude, OpenAI, Whisper, MLX Whisper, SpeechBrain, Vapi, ElevenLabs, Deepgram, Embeddings and retrieval, Vision models
Application and back end
Python, FastAPI, Next.js, React, TypeScript, SQLite, Supabase
Integration
Stripe, Shopify, GoHighLevel, MYOB, Microsoft Graph, Urable, Slack
Delivery and monitoring
Vercel, Railway, Cloudflare, Sentry, PostHog
How I build
Claude Code

Hiring for AI implementation, workflow automation or operational transformation? Let's talk.

LinkedIn