What I learnt using and working with AI

Exit from Oracle due to AI

Towards the tailend of last year – September 2025 onwards – was when I started using AI (ChatGPT) for my work. It was the internal ChatGPT approved by Oracle, so now and again I would ask it a questions related to my work.

After a couple of questions, I asked it this question: Will AI take over and put me out of a job? Knowing it could not really lie and I was curious to see what it said with the follow up question: What shall I do now to prepare for when AI makes my role redundant…

The answer then as it will be the same but a little less specific if I asked these questions now – that is, it told me it was hard to say – it all depends on what my job is now, and it listed the jobs/roles that would be most affected by AI. As for the “what should I do in preparation – for when AI takes my job” – it told be to become an AI evangelist!

September 2025 was bad at Oracle (and it has been ever since) – there were threats of mass RIF (reduction in force) and the threats did materialise into lots of fellow Oracle employees in India and the US getting laid off. The UK was spared, but the impending doom of RIF was demoralising and people did not, for one moment, thought they were safe as it was all over. Personally, for me I was not hopefully – in fact more the opposite as I had been laid off from Cisco Meraki previous to getting this role at Oracle, so I needed to do something about it now.

I wasn’t going to leave it to chance to be laid off twice, so I starting looking for new roles enabling me to leave before the next round of redundancies. I applied for roles in the SRE and Observability area especially as I liked working as a Observability SRE with Cisco Meraki before being “reduced” prematurely! By the way, Oracle started the RIF process to raise capital expenditure (CapEx) for their AI expansion – the Abilene DC in Texas. They made redundancies where they can to reap the most amount of CapEx – there is no other reason why an individual is made redundant apart from raising as much money from their departure as possible…

To my surprise, two such roles appeared – one at Graphcore and one at Nscale. Both of these companies have a close but differing relationship with AI. I interviewed with both using standard and usual preparation techniques with no help from AI. I was offered a role by Nscale but was rejected by Graphcore. I accepted the role at Nscale and once the contract was signed and handed in notice at Oracle – serving a 1 month notice period before joining Nscale in mid-December of 2025.

Use of AI at Start-ups like Nscale

It is of no surprise that the use of AI has been adopted by small companies and start-ups who need to move fast, launch products, and perform support, sales and marketing tasks fast. AI allows you to do this. One moto at Nscale is to move fast and that good is good enough – don’t let perfection slow you down or halt your progress. With the use and help of AI – in all areas of a start-up company – it will allow you to do this with minimal resources and little time.

I found, on joining Nscale, all employees had access to ChatGPT and could ask for access to Claude, and were encouraged to use AI for all aspects of our work. The company were also using modern applications with AI built-in or enabled so were also encourage to use that AI. For example, traditional applications such as Jira and Confluence were replaced by Linear and Notion. These had AI native functions and behaviour which would speed-up or automate your work. They also integrate with each other and other applications such as Slack and Gmail to enable you to combine AI queries across multiple sources of information.

I joined as an Observability Platform Engineer expecting, as part of my roles, to be creating dashboards and alerts with my skills and experience. But no, AI replaced all of this such that any engineer (without o11y skills or Grafana experiences) could ask AI to simply “create me a dashboard to show the latency of X, Y and Z and associate an alert when X, Y, and Z crosses a threshold of A, B or C” – for example. AI would be able to have a very good attempt at doing this very fast (minutes instead of hours). It seems like I was already out of a job before I even started…

Not surprisingly, as the weeks rolled on at Nscale, the use of AI was very apparent, with each Engineering Weekly meeting having a demo that sang the praises of how AI helped with creating a useful or needed feature or solution in a short amount of time. Later, as I used AI to create applications, I recognised or realised that ALL if not most of the Nscale applications, interfaces, and features were created using AI. The UI to their console is a big give away:

No human would write their code on a a few lines!

All Nscale employees were “faking it until they made it” and using AI to help them do so fast. I found that apps like Notion enables you to use AI to find info, detail and documentation really readily and will also summarise and condense information for digesting in a short amount of time – no more tl;dr – get AI to summarise and read the pertinent snippets…

Addictive Nature of AI

First lession learnt after using AI for work (and also for anything else) is that it is addictive – once you’ve used it and found it helpful – it is hard to not use it and go back to how you use to do things albeit a slower and more laborious.

It’s like using a calculator – if you use it for everything, you lose the skill of doing mental arithmetic and come to depend of it to do the all the calculations. If you imagine AI as a super super magical calculator that can help you solve and perform all your work tasks – it will become addictive and the more you use, the more skills you will lose and the more dependent on it you become…

How does an employee get appraised if AI is doing all the work? It comes down to your manager, and unfortunately, after 2.5 months at Nscale, I was transferred to a new manager who didn’t like the look of me and extended my probation period – setting me up to fail so that he could easily dismiss me during this probation period without causing an issue with the company. So after 4.5 months with Nscale, I was let go.

AI Addiction becomes a habit

Once I knew how to use AI and take advantage of its features and limitations – it was hard to not use it. There are so many areas where it could lighten the burden of tedious work and speed up tasks 10 to 100 times. The first thing to use AI for after been made unemployed is to find a new role and/or job.

 role = what you actually do and how you do
 job = formal title and place in the company
 ATS = applicant tracking system

I started out seeking a new job by setting up a spreadsheet to keep track of my job applications – I knew the job market was going to be a lot tougher than the previous two period of unemployment, so I named this sheet appropriately!