All About Titan

What is Titan?

Titan is customized AI technology. It produces quality first drafts of legal documentation in a fraction of the time it takes to do manually. It also helps other functions produce core documentation, such as proposals, press releases, role descriptions and policies.

What’s the story behind Titan?

Titan was created to make lawyers’ lives easier. We know this because our founder and creator, Connor James, developed Titan to help him in his work advising energy companies on their legal and regulatory issues.

What technology sits behind Titan?

Titan is based on a combination of the GPT4 and Anthropic large language models. Users benefit from the power of GPT 4 to deliver accurate results quickly, combined with the finesse of Anthropic, which generates output in language that is more natural than that generated by GPT4 alone.

Exploring Titan

Can I see Titan in action?

Yes – we’d love you to!

Join one of our demos, which run on Fridays every week. Book your spot here.

You can also view short (<90 sec) videos of the various features of Titan in action. Take a look by clicking here.

Can I trial Titan?

You sure can!

Click here to register for a 7-day free trial.

Alternatively, register to attend one of our demos and sign up after that.

Using Titan

How do I subscribe?

Go to titan.compliancequarter.com.au to register.

You can register yourself and you can also choose to register others in your organisation at the same time.

How could I use Titan?

Lawyers can use Titan to generate first drafts far more quickly and accurately than using manual work alone. Titan can analyse contracts, summarise cases, draft policies or determine logic from an Act. It can compare documents as well as create the best version of a document from several different sources.

Titan doesn’t only help with pure legal work – it assists with business processes, too. It can summarise meetings and create proposals, timeline and project plans. Some lawyers record all first client meetings (with permission) and use Titan to turn the recording into a proposal for work, winning more work than they did previously purely because they’re able to submit proposals swiftly.

Titan also assists marketers, HR professionals, project managers and regulatory professionals. It can create advertising copy and press releases, role descriptions and interview questions, project plans and timelines.

The best way to understand what Titan can do is to try it for yourself, or take a look at our short videos.

How should I start using Titan?

The best way for beginners to start is by using one of Titan’s pre-set prompts under each role heading. Upload or cut and paste the information that you would like Titan to analyse and click to get it working.

More experienced users might like to enter their own prompts, which they can do under ‘Text query’ on the right-hand side of the screen.

How do I include my branding in Titan’s export-to-word feature?

You can upload your branding when you register for an account.

Do I need to check the outputs that Titan produces?

Yes. You need to check everything produced by Titan that you intend to use or rely on. Titan is a sophisticated tool, but it is not infallible. Like any AI tool, the results Titan generates need to be checked by a human.

How do I check the outputs Titan produces?

Titan provides links to the sources it uses to generate results, so you can check the foundations on which an output is created.

Do I need to be an experienced prompt engineer to use Titan?

No.

Part of the beauty of Titan is that the sophisticated prompt engineering is done for you. Titan knows what to ask to get the best results for each type of search, so all you need to do is click on one of the pre-set queries.

Am I able to create my own prompts in Titan?

Yes. You can create your own prompts under the ‘Text Query’ section on the right-hand side of the screen. This is useful for specialist work or for those who are more experienced at creating prompts.

Can I use Titan in live consultation with clients?

Absolutely. There is no reason why Titan should stay in the background, and we know of many clients who find it helpful that their lawyers can surface answers to their queries rapidly using Titan.

The same rules apply whether using Titan live with clients or in the background – all outputs must be checked by a human before being used or relied upon.

How do I store useful prompts that I have applied in Titan?

If you use the ‘Text Query’ function, you’ll be inputting your own prompts. Since Titan is entirely private to your organisation, it does not use or store prompts. If you want to keep a record of particularly useful prompts, you will need to actively store them yourself.

We suggest creating a central repository for helpful prompts that everyone within your organisation or team can access and contribute to.

How do I store the outputs Titan creates?

Since Titan is entirely private to your organisation, the outputs it creates are not stored within Titan. You will need to store them yourself.

The best way to do this is to store outputs in your document management system or to your hard drive before you move on to the next query and/or before you exit Titan. This works just the document management system you’re used to – if you create a document you want to keep, you need to save it before you close your system. The same applies to Titan.

Do I need to disclose to my client that I am using Titan?

Client disclosure in the use of AI is an evolving area and if you are interested in this topic, we recommend you follow the work of the Centre for Legal Innovation which is looking at the legal practice implications of the use of LLMs.

Ultimately, you will need to satisfy the regulatory requirements imposed on you in your home jurisdiction and it is a matter for each lawyer or organisation to understand what is required of them. However, we recommend as a matter of best practice that private practice lawyers include a clause in their costs agreement / client care letter that informs the client that you may use a generative AI tool that is private to your organisation, to assist with your work.

Security and Privacy

Does Titan search the web?

Titan creates outputs based solely on the inputs you provide. Unlike ChatGPT, it does not automatically use content on the web to produce results.

Titan can use web-based content to respond to prompts, if you choose to input specific URLs, but it does not scour the web unless you ask it to.

Even when interrogating content within the URLs you provide to Titan, the nature of your prompt and the output created will remain private to your organisation. They will not become publicly available in the way that ChatGPT’s queries and outputs are available on the web.

Are my prompts and outputs public?

No – they are not public.

We have designed Titan specifically not to store or use your queries or the inputs you enter for analysis.  It’s entirely private to your organisation. Nothing you do or create is available to others, including to us here at Titan. We also do not use your queries and outputs to train Titan.

The best way to see this in action is to issue a prompt, generate a result and then click the refresh button. You’ll see that with one click of ‘refresh’, your query and output are gone.

Is Titan private?

Yes. Titan is private to your organisation. The only way your data comes into contact with the AI is when it is processed. Once your query is complete, it is gone forever and is not accessible publicly or by us here at Titan.

Please explain how Titan is private if it is based on GPT4

GPT4 for business is a private large language model (LLM). OpenAI, the developers behind GPT4, have created it to be used by corporates that require data privacy. GPT4 does not store your data or prompts and does not use them to train the system. Once it has processed your query, it is gone forever.  This is why organisations can use GPT4 for confidential work.

In addition to the security built into GPT4 (which many large organisations already rely on), Titan incorporates an additional feature. It removes identifiable data before a query is processed by GPT4 and re-inserts it before the output is presented to the user in Titan. This means that identifiable data doesn’t reach GPT4 and provides Titan users with an extra layer of privacy.

In contrast, ChatGPT, the chat version of GPT4, is not private. It scours the web and uses prompts and data to train the AI. It is not secure and should not be used for anything confidential.

Do you use my searches and outputs to train Titan?

No.

Technical

What types of formats can Titan analyse?

Titan can analyse all standard documentation, including Word, Excel and PowerPoint documents. It can also analyse audio files. Titan can analyse DOCX, PDF, PPTX, CSV, WAV and MP3 files.

Is there a limit to the number of words Titan can analyse?

Titan can analyse up to 65,000 words in a single document.

Could Titan link to my other systems?

Titan can link to other systems, such as your accounting or document management system, via an API.  If you’re interested in this, please contact us to discuss a custom package.

How fast is Titan?

Typically, Titan analyses 10,000 words in around 1 minute.

While Titan is always faster than a human by many orders of magnitude, speed can vary according to server performance.

We’ve found that Titan has been getting faster week-by-week.

Does Titan hallucinate?

Hallucinations are generated outputs from large language models (LLMs) that sound plausible but are factually incorrect or ignore the relevant context. Like human dreams, they resemble reality but are not. They occur because the LLM has no idea of the underlying reality of what it describes and therefore does not apply this context to its output. Hallucinations are seen as a current weakness of large language models such as GPT-3 and GPT-4 and give rise to the risk that users will rely on output that leads them astray. Titan’s creator, Connor James, has written about hallucinations and you can access his piece here.

Like all LLMs, Titan is at risk of its outputs containing hallucinations. This is a current weakness of all LLM-based systems. At Titan, we have, however, incorporated two things into Titan that help reduce the likelihood of hallucinations. First, we use Anthropic’s powerful 100K token model to bring context to the output Titan creates. Second, Titan undertakes past processing once an output has been produced and before it is seen by the user, to identify and discard ‘not true’ statements. 

No AI-produced output is infallible, which is why anything Titan creates should be reviewed by the user before it is used or relied upon.

Pricing

How much does Titan cost?

Titan costs $100 per user per month – less than 20 minutes of many lawyers’ billable time.

Do you offer discounts?

Yes. We aim to get as many many people using Titan as much as they possibly can, so we incentivize this through our discount structure and offer discounts for organisations with over 5 users. Please contact us for details.

Am I locked-in to using Titan?

No.

Our standard subscription runs month-to-month, with no lock-ins.

Alternatively, you can choose to commit to Titan for 12 months and receive your final month for free. Please contact us to arrange this.

Are there any set-up fees?

There are no set-up fees and no set-up is required.

Can I cancel at any time?

If you choose the month-to-month option, you can cancel at any time and your subscription will conclude at the end of the month in which you cancel. Just make sure you give us notice in writing at least 2 working days before the month end.

Do you offer a referral programme?

Yes, we have a programme for connecting Titan to people in your network. Please contact us to discuss.