CLOUDLESS LABS

A free 30-minute call, then a fixed quote

AI consulting and agent development in Sydney

Agents that do real work in a real system, with a human owning the design and the results. And teams that ship faster because they've learned to work with them.

What we do

  • Agents in your product

    Features built on Claude and open models, with the costs per task, the guardrails and the data residency worked out before launch, not after the first bill.

  • Systems handed to agents

    An existing system made safe for agentic maintenance: specifications, tests and review gates, so routine changes are done by agents and checked by an engineer.

  • Claude Code and Codex for your team

    Coaching engineers to ship with coding agents: specs, skills, review habits and the workflow that keeps quality up while the pace goes up.

  • Forecasts and models in production

    Research code and notebooks turned into production systems: data lakes, models, accuracy measured against reality, and apps people use.

Who it's for

Small businesses and larger companies that want AI to do work, not demos: a product that needs an AI feature that pays for itself, a system that costs too much to maintain by hand, or an engineering team that wants to ship with coding agents without losing control of quality.

What "in production" means here

An agent is only as good as the specification it's given and the checks around it. Every system I hand to agents has a written spec, tests that prove a change works, telemetry that explains failures, and an engineer reviewing what ships. That's how a forecasting platform I built by hand is now maintained through agents, with a better model and no downtime, and how a partner's crop-risk platform shipped in a week with Claude Code.

Choosing the model is part of the job. Price per token decides little; the cost per task does, and so does where the data is allowed to go. Which model, at what price and the AI entries in The Stack are the comparisons I use with clients.

What it costs

The first call is free, and so is the written proposal that follows it. The proposal breaks the work into pieces, each with a fixed price set from its scope: what gets built, the systems it touches, and the risk in changing them. You approve each piece before it starts, and the price doesn't move once it's agreed. That's our default, not a rule: if a retainer or hourly billing suits your project better, we'll work that way.

How an engagement works

  1. A free 30-minute call. What the AI should do, for whom, and what it's worth to the business.
  2. A written proposal with a fixed price. The design, the expected running cost per task, and the price for each piece of work.
  3. I build it. Specs, tests and telemetry first, then the agent, measured on real tasks.
  4. Your team keeps it. Documented, in your accounts, with the coaching to run it.

Questions

Do you build on OpenAI, Anthropic or AWS Bedrock? On whichever fits the task and the data: the Claude API directly, open models through OpenRouter when the cost per task matters most, and Bedrock's Australian routing when the data must stay in Australia. The three side by side, with prices: Anthropic API, Bedrock or OpenRouter?. The rest of the trade-offs are in The Stack.

We tried an AI pilot and it went nowhere. Why would this be different? Most pilots fail on the specification and the measurement, not the model. I start from the task and how we'll know it's done right, then pick the model.

What does it cost? The first step is the free call, then a written proposal with a fixed price for each piece. You know the cost before you commit.

See it in practice

  1. 06 Forecasting a weather-driven risk for farmers, from a research paper to agents An Australian agriculture-technology company with a large IoT sensor network · Agriculture technology A sensor network, a white paper from an established Australian scientific organisation and some research code became a production system that forecasts when field conditions turn risky: data lake, models, accuracy reporting and an installable app. Built by hand, then modernised with Claude Code, which now runs it.
  2. 10 Crop-risk scores from a partner's proprietary model, shipped in a week with Claude Code An Australian agriculture business · Agriculture A platform that tells growers how exposed their crop was to an environmental event, and how likely the harvest is to be damaged. The data ingestion and the app are ours; the science stays with a specialist partner's proprietary model. Built entirely with Claude Code, shipped in a week.
  3. 09 Rescuing a stalled modernisation, then handing routine changes to AI An established Australian organisation in the agriculture sector · Agriculture · In progress An organisation consolidating its legacy systems into one portal hired a large outsourced team, and got a late, fragile, insecure platform. I took it over alone, am cleaning it up, and am now building the process that lets non-technical staff ship routine changes through AI.
  4. 13 HappySkills, the package manager for AI agent skills Our product · AI developer tools My own product: a registry and CLI that gives AI agent skills what code has had for decades, dependencies, a lock file, sync and team sharing, all driven by talking to your agent.
  5. 05 Deep neural networks that forecast the weather An Australian agriculture-technology company with a large IoT sensor network · Agriculture technology Machine-learning models and deep neural networks, trained on the company's own sensor data with GPUs on Amazon SageMaker, to predict the weather where its customers farm.

Talk about your project.

Tell us where the business needs to go. You'll talk to the engineer who will design and build it, not a salesperson, and get an honest answer on whether we're the right fit.

Talk to Nic

Contact Us

For general inquiries, please contact us at:

nic@cloudlesslabs.com

For e-learning capabilities, please contact:

lin@shirleenwong.com