AI?at Corvus IT

All progress is seemingly intertwined with AI at the moment. For an innovative technology based Service Provider that relies on deep expertise from its employees, AI is an area we have been watching very closely for the past 4 years.

AI impacts Service Providers and we will discuss this. We use AI at Corvus IT for outcomes that we can achieve ourselves without AI, but in a faster timeframe. We do not use AI for document generation, but some images accompanying text may be used. We also encourage that the inclusion of any AI output is declared as such, not masked to be from the hand of the author.

In a nutshell, that is AI for documents. In regard to the professional work we perform, AI takes a back seat. Its¡¯ only use is seldom and largely validation. An example might be to craft a complex Linux bash command for summarising the list of file extensions in a directory. That command is then copied and tested. Artificial Intelligence is also useful in our industry on how to undertake a particular challenge, what high-level steps to follow. In this capacity, AI serves as a guide, not the engine.

In summary, the use of AI at Corvus IT is light. Our blogs are not AI generated, some of our images are (I think they are obvious). Our website content is all hand-written by ourselves with the assistance of a local copywriter. No AI. Our proposals date back to 2019 and we don¡¯t use AI, but in some cases, we do ask for a re-write of some paragraphs, and the proposed changes are often not as good as the original text.

We have had Co-Pilot introduced to us and it is now throughout our office applications and the changes range from subtle to branding updates. We discovered (and changed) that our standard M365 subscription permitted users to start a Full Co-Pilot trial without Administrator approval. I am sure many will find out this after they realise that are charged A$47 per month per user after the trial completes. Problem avoided.

The Challenge of AI

Artificial Intelligence is a wonderful frontier of new capabilities. We are careful to not let it replace what we do, because if that does occur, we will either dilute or lose the core strengths in deep expertise that we currently cherish. If your database is damaged, you need an expert who will try many different approaches to ensure the best outcome. Agentic AI is nowhere near that capable even though there will be evangelists sprouting that very capability.

The above may read that we are fearful of AI or anti-AI. Neither is true, it is very important that any consumer of AI understands where it is best used to avoid long-term harm. There are many lessons to be learned from the adoption of social media that we can apply to the new frontier of AI.

Oracle Apex with AI

Corvus IT has a strong local Oracle Apex practice. We are building AI solutions internally integrated into Oracle Apex. Being database experts, we are very careful about where data goes and are very mindful of both data protection and sovereignty in general. We use Oracle¡¯s OCI based AI models in conjunction with Oracle¡¯s Autonomous Databases. A safe and great combination that ensures the data is protected and confined to our Tenancy. We have an internal Apex demonstration application that showcases the fantastic capability of Oracle Apex. We have now integrated AI into some parts of this application, and whilst we need to build a larger data set to see the true value of AI, we are now getting real-world glimpses of how AI can be transformative.

We are focusing on human enquiry at the moment. Rather than a structured enquiry where you step through multiple fields and potentially screens, we surface data in a user-friendly format based on a question. The results are promising. Some people think that would be obvious.

What is evident is that AI could easily be used to bypass user-security of data. In our simple system we are asking it to identify contractors that we should target proactively to ensure insurance renewals are performed in time. That same capability could be used to also find out how much they are paid, information that the user is not entitled to under normal application/data security. Hence why this capability is still a work in process.

A lot of applications rely on data not being available to a user merely by excluding access to a report or screen via the user¡¯s menu. The underlying data is available to view if the user is given access ¨C that is why SQL Injection was such a concern because it relied on data not being secured to the user. AI amplifies the lack of data guardrails, and it has become rather obvious in our demo system.

Evangelists will quote theory, but we are grounded in a world of reality. This is giving us pause to reflect. We know how it can be resolved, but it is not a simple undertaking. We want to arrive at a generic approach to employ AI data guardrails so that the introduction of AI into applications does not create new problems. You will hear more on that at a later time.

AI?for Data Generation

One successful venture that we have made with Artificial Intelligence is the generation of fake data to assist with data masking of databases. We have a small number of clients with a particular set of needs that would be difficult for an excellent masking tool like DataVeil to fulfill. We have used AI to generate large data sets that are extremely realistic that we are using to map into sensitive production clones to aid the testing and development in lower environments. AI has been a huge time-saver in this regard.

A?New Era of AI

The title of this post includes 2025. That is relevant because it reveals that we are at a recent dawn, and this post is likely to become obsolete quicker than other articles we have created. AI is introducing velocity to the rate of change, but that presents danger if combined with pressure to use AI to maintain a competitive edge within the market. As a Service Provider, we have a responsibility to our client and this involves caution. We can¡¯t take a step forward unless we know that client data confidentiality is maintained, client security is maintained or improved, and that true value is being added as opposed to perception only.

We are facing the new era of AI with optimism and care, we are in an industry that relies on deep expertise with extreme agility. This is currently not under threat from AI, but opportunities in using AI are there if the correct governance is applied. We hope this little insight into AI at Corvus IT has been of some interest to you and we welcome any discussion you would like to have about your own challenges in this space.