Impact of AI - Skills Churn in Trading Technology

July 30, 2025 · 7 min read · ai trading technology ·

How are roles likely to change due to AI?

I firmly believe AI will change the way we act and operate on a daily basis in our professional lives. Technology teams in capital markets are not immune to this change. It is probably wise to think about what is likely to change and what we, as employees and employers, need to do to ride the wave. Let's look at some key activities performed as part of the following roles and understand how they are likely to change.

Business Analysis:

Mining for institutional knowledge: Most documentation systems already have some capabilities for AI-driven synthesis: Microsoft Copilot and Atlassian Intelligence spring to mind immediately. The next step in the evolution here would be multi-agent co-ordination (either utilizing A2A or MCP, or both) to yield insights faster than ever possible.
Process modelling: Converting conversations into diagrams and documentation are becoming easier. Excerpts from conversations, or even messages in chat tools can be turned into diagrams or pages on your wiki, with the help of AI-enabled tools and chatbots.
Prototyping: For a business analyst or a product manager, this is excellent news; AI-guided coding, a.k.a. "Vibe coding", can quickly allow for a PoC to be built, circumventing the need to requisition in-demand engineering resources for a sprint or two. To be clear, this does need guardrails, tight controls and verification systems. Vibe-coded apps should never be anywhere near production¹. Nevertheless, this can become extremely useful in value discovery, especially in the highly dynamic world of capital markets. Watch this space.

Technical Operations:

Observability: AI and ML for IT Ops are emerging capabilities in the observability space. Open telemetry is the standard now, and software construction will continue to evolve to support these capabilities.
Troubleshooting issues: Technical specialisms, like expertise in protocols, will be augmented with AI. AI will increasingly become the new specialist there. For instance, when dealing with FIX, it is entirely plausible to imagine a FIX MCP Server being able to decipher the missing elements in a given FIX message. It could also be used to determine the cause of FIX message processing errors by looking at source data.
Problem and incident management: Chatbots and AI tools are going to play an increasing role in problem and incident management. MCP tools that integrate with alerting systems, escalation and comms are likely to be commonplace in the near future.

Software Engineering:

Software engineering and DevOps will change drastically over the next few years. This is where we are seeing maximum traction today and I can see the allure. Guido Appenzeller emphasized the enormous market potential of AI augmentation in software engineering, resulting in trillions of dollars worth of productivity gains². There is a grain of truth in such wild claims. Here are some simple cases where AI could be utilized.

Coding co-pilots already provide context-specific recommendations within your IDEs, which is a significant improvement on existing IDE functionality which is largely driven by static code analysis. Moreover, custom agents will provide recommendations based on your personal/team-specific preferences and idioms. Large-scale refactoring, however, is still not quite as successful, but this is the next evolution.
DSL conversions: In essence, everything is a DSL today. AI-guided tools will make it easier to generate code that translate from one DSL to the other. Think gRPC message to SWIFT. Or a VCON message to a REST API call. Or an order record in the database to a formatted email. The list goes on. Development tasks such as these will start getting easier and easier, as AI agents in your IDE bear more of this burden.
Generating infrastructure as code becomes a lot easier. Instead of internal developer platforms and UIs that act as an abstraction layer over Terraform, generating that Terraform becomes a lot easier with AI-driven agents that respond to user intent.

In general, AI-driven coding copilots and IDEs abound today and will continue to get better at taking on complex software development tasks. They will also understand/infer/remember non-functional requirements more and more. They will be ever-present, and they will learn at a faster rate than ever seen before.

Will AI decimate graduate jobs?

The rate of change here is terrifying, so this is an extremely valid line of inquiry. Comments like those made by Satya Nadella³ and Masayoshi Son⁴, do make us think about its impact on job markets. After all, isn't AI a very good junior software engineer today? Will we need these many software engineers in the future?

The truth, though, in my eyes, is slightly more nuanced. Recent job market downturns look to be more due to macroeconomic factors (post-Covid repositioning and global economic uncertainty) than simply due to AI⁵. I don't believe there's actually going to be a net reduction in demand for technology personnel. But there is a skills churn that is taking place. I believe the next generation of AI-native software engineers will differ from the existing generation in a few ways:

Marco Argenti said every individual has to learn three leadership skill: Explaining with clarity, delegation and supervision. AI-native engineers are growing up with these skills and will only continue to get better at them⁶.
AI-native engineers will learn the 'craft' of software engineering much sooner than others in recent years. The ability to go from junior engineer to staff/principal engineer has so far largely relied on personal contributions and learning. AI-native engineers will be able to leverage other people's contributions and learning, thanks to generative AI tools and techniques.

Several other engineering leaders seem to echo such sentiments. Farhan Thawar, Head of Engineering at Shopify wants to hire a thousand interns⁷. This might be an indicator of how employment in technology will change in the future. I don't intend to minimize how difficult it is for graduates to get hired today. I'm simply saying that becoming AI-literate enhances everybody's chances of employment.

What isn't changing?

Personal accountability. Guardrails, risk analysis, verification systems, supervision and governance are critical elements when one considers adoption of AI in financial services. Achieving reliability and consistency in AI solutions continues to be a dark art. Human-in-the-loop (HITL) approaches continue to be the best way of using AI responsibly. Debugging, testing and verification skills, instead of fading away, become even more crucial in an agentic-AI driven world.

AI is not a one-size-fits-all answer and is likely to be another tool in your technology teams' arsenal to address business problems. We need AI-literate technologists to be able to wield this tool effectively.

How can employees get ready?

Embrace the change. Ignoring AI doesn't serve any purpose, except for appearing fashionably skeptical. It would be wise to adopt a growth mindset and broaden our skills.

Here are some things employees can do in their teams today:

Learn Prompt Engineering.
Investigate AI tools being adopted by your firm and gain proficiency in those tools. Tools like M365 Copilot are becoming increasingly commonplace. Additionally many firms are rolling out their own (air-gapped) implementations of ChatGPT.
Find some low-hanging fruits to explore and practice AI usage within your firms. Could AI help with how you run meetings? Could AI help how you find information within the firm? We need to ask and explore questions like these.

How can engineering leaders help?

The primary challenge here is shifting the culture towards greater adoption of AI. The principles here are the same as creating a generative, innovation-friendly culture. Here are some specific changes one could make in their organizations for AI.

Give access to AI tools for people to start experimenting in safe and non-regrettable ways.
Make AI learning mandatory as a development goal for your teams.
Work with your HR colleagues to incentivize and gamify AI learning.
Encourage innovation using AI and remember to celebrate achievements.

I hope that was useful. I will provide some links for learning materials in future articles. Until then, adios!

===================