Our products and client work.
AI products we've built and run, independent projects, and work for clients in regulated industries.
We keep client work general. Our clients trust us with sensitive systems, so we don't name them or share their costs, figures or other specifics.
Our own product
LifeLoom
An AI biographer that phones people for easy, unhurried chats about their life, and turns their stories into a printed memoir.
- Live and selling in Australia since August 2026
- Guided phone calls, a printed hardcover book, and an online memoir for the family. The storyteller just answers the phone.
- Designed, built and run by us, on top of specialist voice, telephony and print partners
- Sector
- Consumer, family
- Services
- AI Build, AI Infrastructure
LifeLoom is our own consumer product. People sign up to tell their own life story, or give it to someone they love. An AI biographer then phones the storyteller at times that suit them for easy, unhurried conversations, and their stories become a memoir in their own words: a printed hardcover book, plus an online memoir the family can read and listen to.
We designed, built and run it, on top of specialist voice, telephony and print partners: phone conversations with Australian landlines and mobiles, conversation design for people who have never used an app, the pipeline that turns hours of conversation into chapters, and the print-ready book at the end. Scheduling, consent, payments and support all run in production.
It is a demanding brief for AI. The person on the call is sharing things that matter deeply to them, so the conversation has to be patient and warm, and everything they say has to be handled with care.
An Australian business in a regulated industry
AI review of regulated sales calls, calibrated by the compliance team
An AI pipeline that reviews recorded sales calls against the client's compliance standards, so problems can be caught and fixed early, before they turn into complaints.
- Led by Damien Healy as the client's fractional Chief AI Officer
- AI findings now agree with the client's own compliance reviewers far more often than the system it replaced
- The compliance team sets the rules, and the AI never overrides a human ruling
- Sector
- Regulated consumer sales
- Services
- AI Expertise, AI Build
The problem. Every sale the client makes by phone is a regulated conversation, and it has to be right. Reviewers could only listen to a small sample of calls by hand, and an earlier AI system wasn't accurate enough to rely on.
The work. As the client's fractional Chief AI Officer, Damien led the build of a pipeline that transcribes each call, removes payment and identity numbers, and checks the conversation against the client's compliance standards and the facts of the product that was sold. Every finding points to the moment in the call it refers to, and results flow straight into the client's own systems.
How it was approached. Accuracy was measured against calls scored by the client's own reviewers from the start, so "better" was never a matter of opinion. The compliance team stays the authority: each of their rulings becomes a written rule and a test, so the same disagreement is less likely to come back. Every significant change is challenged by AI models from different providers before release, and the client's own IT team runs production.
What it changed. Reviewers now check flagged evidence instead of listening to whole calls, and advisors get specific, actionable feedback. The focus is fewer complaints and better outcomes for customers, not catching people out.
An independent, non-commercial project
AI Inquiry Atlas
A guide to every public submission to the Parliament's Joint Select Committee on AI, with AI-written summaries shown beside submitters' own words.
- All 571 public submissions on one map: 6,275 pages including attachments
- 8,678 recommendations recorded, each shown with the submitter's own words
- Nearly 12,800 quotes checked by a program to appear word for word in their source
- Built and published in three days, and improved since
- Sector
- Public policy
- Services
- AI Build
AI Inquiry Atlas is an independent, non-commercial project by Damien Healy, Qanara's founder. No one commissioned or paid for it. It is unofficial and not connected with the Parliament of Australia or the committee, and its summaries are written by AI and can be wrong.
The problem. The Joint Select Committee on Artificial Intelligence received 571 public submissions, about two million words, and held 11 days of public hearings. It is a large public record of what people and organisations told the Parliament about AI, and hardly anyone will read all of it.
What was built. A guide at jscai.au that places every submission on one map and records every recommendation with the submitter's own words and a link to the original. Recommendations that AI models judged to seek substantially the same thing in three or more submissions are grouped into 189 shared asks, and the guide shows which of them an AI model matched to passages in the hearing transcripts. It is built entirely from the public record on aph.gov.au, as downloaded on 13 November 2026.
How it was built. AI models read every submission and wrote the summaries, key points and recommendations, and AI coding agents wrote most of the code under Damien's direction. For 70 submissions from individuals, the guide publishes only the person's own words. Because the analysis is written by AI, much of the work went into testing it. A program checks that every quote appears word for word in its source. AI systems from two other developers reviewed samples in five rounds, and after each round AI models corrected the whole set. A later pass by another developer's model tested every statement the guide attributes to a submitter, 13,702 in all, against the original. The site publishes that record, errors found included.
Why it's here. It is Damien's own project, not client work. It is listed because it shows an approach: a large body of public documents turned into something people can use within days, with the checks published and the limits stated plainly. Damien made a submission to the inquiry himself, which the site discloses.
An Australian business in a regulated industry
Taking over a monthly commission-invoicing cycle with AI agents, fully reconciled
A supervised AI pipeline that turns commission statements into partner invoices, revenue reports and payment files, delivered on deadline in its first month.
- Led by Damien Healy as the client's fractional Chief AI Officer
- First month delivered on deadline, fully reconciled, with every line accounted for
- The process is now documented and repeatable
- Sector
- Regulated consumer sales
- Services
- AI Expertise, AI Build
The problem. Every month, the client receives commission statements from its product providers, each in its own format, and has to turn them into accurate tax invoices and payments for its partners by a hard deadline. The process relied on spreadsheets and know-how that hadn't been written down.
The work. As the client's fractional Chief AI Officer, Damien led the takeover, directing AI agents alongside the client's own finance, operations and data teams. The rules were recovered from evidence, by replaying past months against the client's own records, and every rule was documented with its source. The new process ran alongside the existing one before taking over a live cycle, and every file that went to the client's team was independently checked by AI models from different providers.
What was built. A pipeline that reads each provider's statement, matches every line to the right sale, applies the right partner agreement, and produces the invoices, reports and files the finance team needs. Anything ambiguous is held and listed for a person, never guessed, and nothing leaves until every line is accounted for and the numbers tie out.
People hold the money levers. The pipeline never changes the client's systems. The client's finance team reviews, sends and pays, and stays in control of every dollar.
Our own product
ozzai
An Australian AI platform that's easy for anyone to use, and gives power users models from the leading AI labs: Anthropic, OpenAI, Google, xAI and Perplexity.
- A Simple mode for everyday use, and an Advanced mode where power users choose from every available model
- Models from Anthropic, OpenAI, Google, xAI and Perplexity on one account, including web-search answers and image generation
- Account and chat history stored in Sydney
- AI models run by overseas providers (mainly US and EU)
- Sector
- Consumer
- Services
- AI Build, AI Infrastructure
ozzai is our own AI platform for Australians. Instead of juggling separate apps and subscriptions, one account gives you models from the leading AI labs, Anthropic, OpenAI, Google, xAI and Perplexity, including answers from web search and image generation.
It is built for two kinds of user at once. People who aren't technical can just start typing in Simple mode, while power users switch to Advanced mode to choose the exact model for the job, from Anthropic, OpenAI, Google, xAI and Perplexity. Your account and chat history are stored in Sydney. To answer you, your prompts and files are sent to AI providers overseas, mainly in the US and EU, as ozzai's privacy policy explains.
We designed, built and run the platform: connecting several AI providers behind one account, the accounts and credits, the payments, and the infrastructure underneath.
An Australian business in a regulated industry
Ongoing monitoring of third-party marketing claims, with every finding human-reviewed
An ongoing pipeline that checks what third parties say about products in the client's market, and turns it into evidence-backed findings for a compliance reviewer to decide on.
- Led by Damien Healy as the client's fractional Chief AI Officer
- Every finding backed by the exact source text and a saved capture
- Every decision made by a person, and carried forward over time
- Sector
- Regulated consumer sales
- Services
- AI Expertise, AI Build
The problem. In a regulated market, what third parties say about products matters, and a single unsupported claim can mislead customers. Checking it by hand meant starting again each time, with nothing carried forward and no complete record.
The work. As the client's fractional Chief AI Officer, Damien led the build of a system that regularly captures the relevant websites, ads and videos, then uses AI to identify claims that may need attention. The AI has to quote its source exactly, and anything it can't support is rejected. Findings are consolidated into a clear report, and anything that couldn't be checked is reported as a gap, never as a clean result.
People decide. Nothing the system produces is a judgement on its own. The client's compliance reviewer accepts, rejects or queries every finding, and that decision carries forward, so the same item isn't reviewed from scratch again. Nothing goes outside the business without a person's approval.
What it changed. Monitoring that used to be a repetitive manual chore is now a running, evidence-backed record that shows why every item was raised.
Our own product
FuelPick
A free public service that brings live fuel prices from four states into one place, with AI that suggests when to fill up.
- Free for everyone, with no ads and no tracking cookies
- Prices from the NSW, Queensland, WA and Victorian government fuel feeds, in one view
- AI that reads local price cycles and suggests when to fill up
- Runs in the browser on any phone, with no app to install
- Sector
- Consumer, motoring
- Services
- AI Build, AI Infrastructure
FuelPick is our contribution to Australian drivers. It's free, it carries no advertising, and we don't make money from it. We built it because fuel is a cost every household feels, and the information to save on it should be easy to use.
Fuel prices in Australia move in cycles, and each state publishes its prices differently. FuelPick pulls the official feeds from New South Wales, Queensland, Western Australia and Victoria into a single model, so drivers can compare stations near them on one map.
On top of the data sits an AI fuel advisor. It reads each area's price cycle, predicts whether prices are likely to rise or fall, and suggests when it is worth filling up.
We built and run the whole thing: ingestion from four different government sources, the mapping and location search, the prediction layer and the hosting behind it.
Have something you want built?
Tell us what you're trying to achieve, and we'll tell you honestly whether AI is the right way to get there.