Why I have a Secret AI Business

For the last six months I have been running a separate division of Cartland Law that has been ‘secret’.  It is a strange thing to keep a business secret. Normally we want to advertise and promote a new business venture. (But normal people don’t do tax law – please don’t judge). The business is a flow-on from another significant event: the quiet murder of my beloved Ailira. 

No Silicon heaven?  Where do all the Calculators go?

Two years ago I put to death my artificial intelligence Ailira, whom my wife described as my “primary girlfriend”.  The reasons for terminating an AI business that was otherwise profitable and growing, and at the cusp of an AI boom, are a story for another day (the story of Ailira is coming). Put simply, in my view, we are at the top of the AI hype cycle, and winter is coming. 

I didn’t tell anyone because, quite frankly, who cares?  I don’t see a need for aggrandising the celebration of failure.The reason I will be setting out the story of Ailira now is that I think that there are many interesting things to learn (including from my many failures) that have become relevant in the present AI boom and so I’d like to share those insights.

Do Sleeping Robots Dream of Electric Sheep?

In the time since I have not slept on AI. We have been continuously developing fun things purely for my firm and not as products for sale. How can a sole practitioner tax firm afford to build things while also paying for my expensive suit collection?  The secret is that the cost of implementing AI has dropped so dramatically no one would believe it. Ten years ago thing that would take months of development time and $100,000’s can now be done in days. Instead of a custom or propriety or patented systems, the existing environments such as Microsoft, Google or OpenAI can, if set up correctly, perform amazing feats.  There is no need to buy proprietary systems when much more reliable ecosystems can provide superior results when used properly. I haven’t slept on AI,  I have pushed it to its limits.

Never Tell Me the Odds

There was a short step from building AI for our own firm, to building AI for clients of our firm.  I have spoken at conferences about things that we’ve done such as custom AI builds for clients as part of tax disputes. My AI development team has been using the skill set that they developed on building for my firm and turned it to building for my clients.  Being an exceptionally busy Tax Law firm has meant that I have kept my AI team working on our current matters. So I have been running a secret second AI business that has been consumed wholly by building for existing clients.  We have had a number of other Law and accounting firms reach out to me for advice on how they could implement AI in their firm as I have done in my firm.  I have recently started contracting my AI development team out to professional colleagues. Given the demand for an unadvertised business, I have had to keep it secret.
 

Come With Me If You Want to Live

In order to meet the demand from clients, and also my own desire for more cool things, I have expanded the team and formerly set up a separate company, Cartland AI Pty Ltd. This stealth company has nearly a dozen staff. Now that capacity is expanding I am now able the reveal the business more broadly, and are now accepting new instructions.  

 Cartland AI will make the existing products that you are currently disappointed in actually work. We will show that you can get from AI results that are beyond the statements of the most bullish vendors.  But as part of Cartland Law, we also deeply understand professional practice and how to make technology fit in existing workflows, rather than replacing them. 

Just as your advices, documents, and financials are not off-the-shelf things that you rubberstamp, so too should your AI not be a pre-packaged product that dictates how you operate.  Leading AI platforms can easily be implemented to work for you and within your existing systems. 

The short answer of why I killed Ailira was because there is no need to charge exorbitant fees for a wrapper on another company’s base-model AI. It is faster, better and cheaper to simply set up those base models to work for you and your firm. This is what my new company does.