How does the underlying technology work?

Database and Knowledge:

Our applications are trained on a vast range of publicly available Australian legal data to enhance accuracy.

Machine Learning

We train large-language models to understand complex legal terminology and domain-specific information.

AI Integrations

Our applications can analyse, synthesise and retrieve information from a large volume of documents instantly.

Smart Learning

Over time, our applications will get better at prioritising the information and questions that are most valuable to lawyers.

Frequently asked questions:

If you can't find the question you want to ask below, please reach out to us and we will be happy to explain our technology and integration process further.

Why is our tool different to existing AI research tools?

Existing AI tools are trained by sweeping the entire internet, which limits their use for specialised use cases such as law because they are trained on a vast corpus of knowledge. In contrast, because our tools are trained on a comparatively smaller corpus of Australian legal data, this greatly increases the accuracy of our results. In turn, our technology reduces the likelihood of 'hallucinations' or fabricated results, which is a huge risk associated with leveraging existing AI foundation models.

Can firms train models and conduct research using their proprietary data?

Various firms have expressed interest in training models on proprietary data internal to the firm, with the desire that a particular writing style can be emulated or that our information retrieval methods can be applied to internal documentation. While development of our core product suite is our primary focus, we are very open to conducting these integrations with firms. However, we note that there is a strong legal imperative that client confidentiality be maintained, and firms should ensure that lawyers are not incorporating private client details in their prompts to our search tools.

Will this tool be targeted towards use cases such as discovery and due diligence?

Our initial focus is on building AI tools for conducting legal research into relevant precedents more efficiently, as well as drafting legal advice. However, our underlying models have transferable value for engaging in use cases such as discovery and due diligence over hundreds of thousands of files. As we progress, we intend to cover a wide range of legal use cases and incorporate a suite of legal tools.

Does our tool cite the sources used to inform AI-generated answers?

Yes. We think that sourcing and attribution is deeply important in understanding how an AI tool has produced a certain conclusion to a legal question. This is because a lawyer can trace the 'logical path' which has been used to produce an answer, and think about whether they want to reformulate their chosen prompt, so that our AI tools approach the problem in a different way.

How secure is your tool when it comes to protecting sensitive information?

Security is one of our top priorities - and clients have the option to run their systems locally, or integrate with existing security frameworks. Data and questions that are inputted as prompts will never be shared with third parties or competing firms.

Can your tool provide up-to-date and accurate legal research across multiple jurisdictions and areas of law?

At present the database of information our tools are trained on spans up until and including January of 2023. Every month, we make efforts to expand this corpus of knowledge to include the most recently published case law. Again, this gives us an informational advantage over tools like ChatGPT, which has limited knowledge of events after 2021.