Company · Business plan
Legal AI your clients' confidences can survive.
Cogent Systems builds the IQ platform for Australian legal practice: research with citations you can check, practice management built around the matter, a curated legal corpus, and security monitoring that watches it all. Firms decide where their data goes, down to a mode where nothing leaves their own server.
No product is generally available and we have no customers yet. Every capability below carries its status: built and in alpha testing, in development, or planned. The live roadmap shows the detail.
Vision and mission
Every Australian lawyer should be able to use AI without gambling with a client's confidence.
Vision. Legal work in Australia done faster and checked more easily, on systems that keep each client's information where the firm decided it should be.
Mission. Build one platform for research, practice and security that a sole practitioner can run on one server and a larger firm can test and audit through its API and command line: cited answers rendered to the Australian citation standard, matters at the centre, Australian residency or a fully local mode, and security monitoring that defends and never strikes back.
How we work. Truth over polish. We publish what is built and what is not, we hold our engineering to the Australian Signals Directorate's Information Security Manual, and every safety rule the software must keep is an automated test.
Who it is for
Four kinds of legal team, one platform.
Each block says what the reader gets and where it stands today.
-
For law firms
Small and mid-size firms carry the same confidentiality duties as the largest, with a fraction of the IT budget.
- Choose how confidential the firm is, firm-wide and matter by matter Built, in alpha testing
- Matters with permanent codes, confirmed deadlines and import from your current system Built, in alpha testing
- The AI cost of every matter, drawn from metering rather than estimates Planned
- A place in the beta programme, with a direct line to the people building it In development
Every claim on this site carries its status, and the live roadmap shows what is built.
-
For in-house counsel
In-house teams want research and matters in one place, data that stays inside the organisation, and something to show the board about security.
- One application for research and matters, on one matter code Built, in alpha testing
- Data kept inside your organisation in the local mode Built, in alpha testing
- A monthly threat report for your organisation from AegisIQ In development
- Export on request Built, in alpha testing
- Retention review and legal hold, every step decided by your firm administrator In development
PracticeIQ's matter, research, drafts and documents screens run on our test server; mail and the AI cost view are planned.
The problem
Lawyers want AI. Their duties make most of it hard to use.
The profession's own research shows the gap between interest and safe use.
-
36.7%
of Victorian lawyers reported using AI tools in practice in the 2025 census; barristers (14.6%) and government lawyers (11.2%) far less. Legal Services Research Centre, Victorian Legal Services Board and Commissioner, 2026
-
Almost half
of non-users gave data, privacy and security concerns as their main reason for not using AI. Legal Services Research Centre, Victorian Legal Services Board and Commissioner, 2026
-
95.5%
agreed that lawyers have a duty to ensure AI use complies with professional obligations. Legal Services Research Centre, Victorian Legal Services Board and Commissioner, 2026
-
81
data breach notifications to the OAIC in 2025 came from legal, accounting and management services, in a record year of 1,205. Office of the Australian Information Commissioner, 6 July 2026
Courts have issued practice notes on generative AI and have referred practitioners who filed authorities that do not exist to their regulators (see Lessons from the industry). Regulators expect lawyers to keep confidential information out of public tools and to verify what AI produces. Law Society of NSW, Legal Practice Board of WA, Victorian Legal Services Board and Commissioner, 6 December 2024
Most firms are small: 78% of private practices are sole practices. Urbis for the Law Society of New South Wales, 13 June 2025 (data as at October 2024) They carry the same duties as the largest firms without a security team. The tools on offer ask them to send client material to someone else's cloud, or to give up AI altogether.
The products and how they fit
One application in front. Specialist engines behind it. A secure platform underneath.
-
PracticeIQ, the one application
The only screens a practitioner or firm administrator uses: matters with permanent codes, confirmed deadlines, calendar and mail, and research, corpus and security views inside it.
Built, in alpha testing matters, codes, deadlines, import, the matter link to LexIQ, and the research, drafts, documents, members and roles, firm settings and calendar screens · In development the security notices and monthly report · Planned mail, calendar sync with Microsoft 365 and the corpus views
-
Behind it: LexIQ, CodexIQ and AegisIQ
LexIQ researches and drafts with AGLC4 footnotes. CodexIQ curates and versions the legal corpus. AegisIQ monitors and defends the platform. Each is complete through its own API and command line, so larger firms and their IT providers can script and audit them directly.
Built, in alpha testing LexIQ Alpha 1 and CodexIQ curation · In development AegisIQ detection and alerts
-
Underneath: the IQ platform
Identity, a TLS 1.3 gateway, one database per firm, metering of usage, one place to set up each firm, and the agent tier, designed never to give an agent more access than the person it acts for.
Built, in alpha testing the gateway, sign-in between services, a database per firm, metering and firm set-up · In development agents limited to the person's own access, and running across two sites
Three release editions, each with a provisioning application
-
Linux edition
PlannedA Linux server becomes a working site from an installer, with a provisioning application to set up the firm, its people and its keys.
On-premises installer for Linux servers on the roadmap for December 2026
-
Microsoft and Azure edition
PlannedFor organisations that standardise on Microsoft: the same software in their own Microsoft cloud, with a provisioning application.
Date not set
-
Windows self-host edition
PlannedFor firms and counsel who want everything on their own Windows machines, with a provisioning application.
Date not set; a Windows server installer follows the Linux one
The cloud profile runs the same software in an Australian region. No edition and no production hosting is available yet.
The market
A large, fragmented profession with rising duties.
The detail, the landscape and where we differ are on Industry and competition.
97,500
practising solicitors in Australia in October 2024, up 47% since 2014. Urbis for the Law Society of New South Wales, 13 June 2025 (data as at October 2024)
16,793
private law practices (excluding the ACT, whose size data was unavailable); 79 had 21 or more principals and employed 21% of solicitors. Urbis for the Law Society of New South Wales, 13 June 2025 (data as at October 2024)
64%
of solicitors work in private practice; 17% in corporate legal and 14% in government legal. Urbis for the Law Society of New South Wales, 13 June 2025 (data as at October 2024)
$39.2bn
estimated Australian legal services industry revenue in 2026, across 23,656 businesses (industry estimate). IBISWorld, updated September 2026
Customers and segments
Small and mid-size firms first. On-premises for those who need it.
Then in-house counsel and the Bar.
Small and mid-size firms
The bulk of the profession: sole practices and firms of two to twenty principals that want AI without a security department. They get confidentiality levels, matters at the centre and, planned, costs per matter.
Firms that must keep data on their own servers
Firms whose client agreements or practice areas rule out anyone else's cloud. The local mode can be tried on our test server in the beta; the self-host editions follow as they ship.
In-house counsel and barristers
Corporate legal teams want one place for research and matters; barristers want checkable authority and independence from any firm's systems.
Business model
Simple to buy, honest about cost.
We will publish pricing with the beta. Until then we do not quote numbers, and the models below are indicative until confirmed.
Subscription per firm and seat Indicative
A firm subscription for the platform and the products it chooses, plus a seat for each person who uses them.
Metered AI at a published rate Indicative
Every API call is metered today and other kinds of work are being added, so AI use can be charged at a published rate and, planned, shown per matter, not hidden in a bundle.
Implementation and provisioning Indicative
Importing matters and documents from the firm's current systems, and setting up on-premises and self-host editions.
Support retainers Indicative
For firms that run their own editions and want a named response time and upgrades handled with them.
Go-to-market
Earn trust one firm at a time.
The beta programme
Registration is open on this site for firm trials and individual testers. Invited trials open as the release gates pass.
Pilot firms
We plan to invite pilot firms across the segments above, including firms that run on their own servers, with a direct line to the people building the product.
Professional bodies
We intend to work with law societies, bar associations and practice management groups on safe AI use, starting from the guidance they have already issued.
Partners (planned)
IT providers who would install and support the self-host editions for their law firm clients, and integrators who would build on the open APIs.
Roadmap summary
Where each product is now and what comes next.
Drawn live from the same file as the roadmap page. Dates are planning targets and can move.
- IQ platform Architecture and security foundations In progress 19 Oct 2026 to 30 Oct 2026 Next: Two-site sync and on-premises installer 14 Dec 2026 to 25 Dec 2026
- LexIQ Alpha 1 In progress Now Next: Beta 1 Date not set
- PracticeIQ M2: calendar and mail In progress 4 Jan 2027 to 15 Jan 2027
- CodexIQ M1: first discovery Planned 14 Dec 2026 to 25 Dec 2026
- AegisIQ M1 Watch In progress Target 26 Nov 2026 Next: M2 Contain Target 10 Dec 2026
What makes it defensible
Hard to copy because it was built this way from the first line.
We hold no patents and claim none. Our advantage is how the system is built and governed.
The engineering standard
Written decisions with rejected options, tests before code, safety properties as tests that are never retried to pass, and a register tracking our controls against the Australian Government's Information Security Manual. Built, in alpha testing
Australian residency and compliance posture
An Australian cloud region by design, a fully local mode, and a register of every Australian law and standard that applies, with the gaps stated. Built, in alpha testing
Local models and per-firm learning with consent
A local model the firm controls, and learning that stays inside each firm's boundary and is shared only with the firm owner's or administrator's consent. Planned
The security product
AegisIQ is part of the suite, so the platform is watched by a tool that understands it, and firms are told about threats to them. In development
The headless architecture
Every engine works without a browser through versioned contracts, so new screens, integrations and editions do not need new engines. Built, in alpha testing
Team
Small, and organised to stay honest.
Founder and system administrator
Sets the product direction, approves every architecture decision and system change, and runs security operations for the platform.
Engineering
AI coding agents working to written plans, decision records, test gates and review rules, under the founder's direction.
Independent review
Legal review of our customer terms, privacy documents and this plan, and independent security testing, are open items before the beta.
Risks
What could go wrong, and what we do about it.
| Risk | What it is | How we manage it |
|---|---|---|
| No customers or revenue yet | Every product is pre-release. Nothing has been sold. | A beta programme with invited firms before any general release; pricing is published with the beta; release gates in every hosting profile. |
| AI accuracy and the courts | Courts in Australia and overseas have criticised practitioners who filed authorities that do not exist, and practice notes now restrict how AI may be used. | Footnotes rendered by code from the sources actually retrieved; a research aid label on every answer; respect for each court's rules (see Lessons). |
| Open security items | Immutable segregated backups, malware scanning in every upload path and independent testing are not yet in place. | No real client data until they close; each item is listed on the Security page with its milestone. |
| Reliance on hosted AI providers in the hosted levels | At the Standard and Enhanced levels, questions and passages go to a hosted provider outside Australia under that provider's terms. | A Maximum level in which nothing leaves the firm's server; a warning naming the provider's region and retention before a firm chooses a hosted level; a register of each provider's data terms that the code checks before any call. |
| Engineering scope for a small team | Two-site sync, several hosting editions and four products are a large surface. | Logical tiers first and physical splits only when load requires; a fault-injecting simulator before any sync code; scope cut from the end of a sprint, never the tests. |
| Key-person dependence | Product and security decisions rest with the founder today. | Decisions are written down as numbered records with rejected options, runbooks are kept for operations, and every rule the software must keep is a test. |
| Content rights | The curated corpus needs the right to copy, index and serve each source. | A rights review of each source's terms before the corpus grows beyond samples. |
| Regulatory change | Privacy reform continues: automated-decision transparency duties start on 10 December 2026, and a further privacy bill is in exposure draft. | A register of every Australian law and standard that applies, with the gap and who is responsible for each row, reviewed at least yearly. |
Work with us
Help build legal AI the profession can rely on.
-
Firms and legal teams
Register for a firm trial and tell us about your practice. You will hear from us when invited trials open.
-
Partners
IT providers, integrators and professional bodies: register your interest and say how you would like to work together.
-
Investors
If you share the view that legal AI must be built for confidentiality first, follow our progress through our product updates.
This page is general information about Cogent Systems and its products. It is not an offer of securities, a prospectus, an information memorandum or financial product advice, and nothing on this site should be relied on for an investment decision. Any investment discussion will take place under the applicable Australian law, with the appropriate documents.
Sources
Where the outside facts come from
Each source was read on 20 Oct 2026. Where we could not open a page directly, the entry says how it was read.
- 2024 National Profile of Solicitors
- Legal Services in Australia, industry summary page
- Generative AI Use in the Legal Profession: Findings from the 2025 Victorian Lawyer Census
- Statement on the use of artificial intelligence in Australian legal practice
- Data breach notifications increase to all-time high in 2025, new NDB stats show
Be one of the first firms on the platform.
Register for a firm trial, or leave your email for product updates. We will contact you when invited trials open.