Not a chatbot. Grounded in your firm’s documents, systems and decisions, inside your permissions, with a record of everything it does.
For firms of 15–150 staff in Australia and Singapore whose value sits in the work they have already done.
Just not safely, and not on your data. Every document pasted into a public tool is a confidentiality question you'd rather not have to answer.
It hasn't read your files, your history, or the way your senior people actually work, so it produces plausible work you can't rely on.
A feature bolted onto software you already pay for, trained on everyone's data but yours, priced per seat forever.
We own the whole stack: the data pipeline, the approval and audit layer, and the deployment. That’s why we can answer the questions other vendors deflect.
We start with your files: the shared drives, the system of record, the decade of work nobody can find. We clean it, structure it, and model it as the things your firm actually deals in (clients, matters, projects, decisions), so it can answer what position did we take and why, rather than returning nine documents that contain the phrase. Permissions are mirrored throughout: if someone can’t open a file today, the co-pilot won’t surface it to them tomorrow.
Every action is gated by an approval tier you set, and written to a signed, tamper-evident log. You can show a client, an insurer or a regulator exactly what happened, when, and who approved it. Most AI vendors can’t do this. It’s the reason we exist.
On-premise, or private cloud in your region. We scope the right deployment during the audit and put the data flow in writing before you commit. Your data is never used to train anyone else’s model, and the monthly cost is fixed within your deployment tier, so finance can plan.
An illustrative exchange, not a client’s. The shape is the point: a question in your own words, an answer that cites the files it came from, an honest gap where the files don’t support one, and a record you can hand to someone else.
“What position did we take for this client in 2019, and why?”
In March 2019 the firm advised against the restructure, on the basis of the financing covenant in the client’s facility agreement. The reasoning is set out in the file note from the same week, and the position was reaffirmed at the November review. No later document on file departs from it.
On your second question, about the 2021 amendment: there is nothing on file that supports an answer. Rather than guess, it says so.
Reading and answering needs no approval. Anything that leaves the firm does.
Exportable, and readable by someone who has never seen the system.
Built from the shape of the questions firms actually ask us about, not from any client’s files. Yours would run against your own corpus, in your own house style.
Twelve weeks, five steps. You’ll know what’s happening at every one of them.
We map your data, rank the use cases and quote a fixed price. You own the report either way.
The unglamorous part, and the reason most AI projects fail. We clean, de-duplicate and structure your documents.
One workflow, done properly. Retrieval over your corpus, permissions mirrored, tuned to your house style.
We test against real work your team has already done, and show you the accuracy numbers before anyone relies on it.
We host it, monitor it, keep it current, and add workflows as you find them.
Somewhere between 60% and 70% of this work is data preparation: duplicated files, three versions of the same document, scanned PDFs nobody ever extracted text from, folder structures that made sense to someone who left in 2019. It’s the work that decides whether the co-pilot is useful, and the work most vendors skip because it doesn’t demo well. We do it first, and we tell you what we found.
Per-seat AI tools look cheap until you multiply them. At $50–500 per user per month the bill grows every time you hire, and you’re paying it forever for something that still hasn’t read your files.
A private co-pilot is a build cost and a flat monthly fee within your deployment tier. Tiers step with corpus size and workload, never per user, so it doesn’t get more expensive as the team grows, the improvements accrue to you rather than to a vendor’s other customers, and the asset (your prepared, structured corpus) is yours to keep.
Somewhere around 30–40 users, the maths stops being close.
Illustrative at $150 per user per month. Your actual comparison goes in the audit, with your licence costs.
We’d rather you found out in thirty seconds than after a sales call. Nothing is sent anywhere; this runs in your browser.
Answer honestly. A no is more useful to both of us than a maybe.
The problem is the same everywhere: decades of knowledge sitting in documents, expensive people spending their day looking for it, and confidentiality obligations that rule out the tools everyone else uses.
Working papers, client history, compliance
Client history, mandates, suitability and review files
Precedent retrieval, matter summarisation, drafting
Policy interpretation, claims history, assessment
Standards, past projects, specifications
Not on the list? The test is simple: if your work product lives in documents, your data can’t go into a public tool, and your senior people spend their days searching, then it fits. Ask us.
We’re probably not the right fit if you have under 10 staff, no real document history, you want a demo this week, or what you need is everyday productivity: email, meeting notes, ordinary document editing. The AI in your existing subscription does that well. Come to us for the work it can’t touch.
Three weeks. A$12,000, fixed. You get a data map, three ranked use cases with the hours and dollars attached, a security architecture your IT provider can review, and a fixed-price quote for the build.
The ranked use cases are the point: estimated hours saved and dollar impact, scored against effort and risk, so what you are holding at the end is a business case rather than a workshop summary.
No obligation to proceed. If you do proceed within 90 days, the full audit fee comes off the build.
Book a Readiness AuditScoped and built by James Storrier and James Coffey. There is no delivery team behind us you have not met. Who we are.
No. Not ours, not anyone else’s. It’s in the contract.
It depends on the deployment we scope: inside your own infrastructure, or in a dedicated single-tenant environment in your region. We document the exact data flow in the audit, before you commit to anything.
The system answers from your documents and cites them. Where it can’t find support, it says so rather than guessing. We measure this before go-live against work your team has already done, and we show you the numbers.
No. We read from what you have. Replacing core systems is not the project.
It’s for a different job, and most of our clients keep using both. Those tools are built for everyday productivity (email, meetings, ordinary documents), and they’re good at it. We’re for the work where the knowledge is confidential, specific to your firm, and where you need to prove afterwards exactly what the system read, produced and was approved to do. If everyday productivity is all you need, use what’s already in your subscription and save your money.
Audit A$12,000. A first co-pilot typically lands between A$45k and A$85k depending on data condition and workflow complexity. The managed service starts at A$3,000 a month, which is the entry tier: tiers step with corpus size and workload, never per user, and the audit tells you which one you land in. The audit gives you a fixed number before you commit to the build.
About twelve weeks from audit kickoff to a live pilot group, assuming reasonable data access.
Usually one of three reasons: it didn’t know anything specific to the firm, it didn’t fit the way people already work, or nobody was accountable for adoption after go-live. The audit is designed to catch all three before you spend.
You own your data, your prepared corpus and your configuration. You can take them with you.