In our previous volumes we touched on topics like workflow solutions, hallucinations, etc., but we haven’t yet addressed how it will impact the human being sitting in the middle of this transformation.

The Skillset That Got You Here Will Not Get You There

What has traditionally made someone a strong analyst? Maybe strong analysis on data, or the ability to build complex financial models, or clean writing, maybe the discipline to check every number twice. These are real skills that took years to develop. But if AI can now do the mechanical parts — data extraction and analysis, model population, first-pass writing — then what happens to the analyst whose entire value is tied to those mechanical skills?

I have seen this firsthand. At SP2, the briefs we receive from clients have shifted noticeably. Two years ago: “We need an analyst who can build a three-statement model from scratch.” Today: “We need someone who can review AI-generated models, catch errors, and make judgment calls on assumptions.”

The task is effectively the same — a model gets built. But the human role has moved from creation to verification, from data entry to interpretation.

What the AI-Augmented Analyst Looks Like

If you are an analyst, here is the profile every firm will be hiring for within two years (my view).

You still need deep domain expertise. Professionals such as CAs, CFAs, or MBAs in Finance with strong analytical skills and a solid understanding of accounting standards and valuation frameworks are becoming more relevant, not less. You need the ability to work with AI outputs critically. Not just prompting — anyone can prompt. I mean looking at an AI-generated earnings summary and immediately spotting what is wrong. Trust me, a lot of AI work is not important enough to be shared with a client and it misses a lot of important items that you would like to focus on. Knowing that when an AI tool says “management guided for margin expansion,” you need to verify whether that was explicit guidance, or confirm if margin might really expand with changing scenarios, or was it just AI’s interpretation of a vague comment.

You also need workflow literacy — understanding how to work within structured AI workflow systems, not just chatbots. The previous generation of analysts needed to be Excel experts. The next generation needs to be AI workflow experts.

And finally, when the mechanical work is automated, communication and judgment become the core differentiator. The analyst who can synthesise AI-generated research into a thesis that a PM or investment committee actually acts on — that analyst is worth multiples of what they were worth five years ago.

What This Means If You Are Making Hiring Decisions

If you are a team lead or a managing director, three implications.

First, pure execution hires become less valuable relative to judgment hires. If you are still hiring juniors primarily to populate models and format pitch decks, you are hiring for tasks that AI already does faster. Hire juniors for their ability to verify, think critically, and grow into judgment roles.

Second, domain expertise matters more, not less. When AI-generated output increases in volume, the ability to separate signal from noise becomes the bottleneck. A qualified professional who understands the domain can do that. A generalist cannot. At SP2, we have always focused on CAs, CFAs, and MBA Finance professionals. AI makes that conviction stronger.

Third, you need people who can work alongside AI workflow systems. One of our clients, a mid-sized equity research firm, found something revealing after deploying a workflow solution. Their senior analysts initially resisted but, once they adopted it, their output quality went up significantly — they used AI-generated drafts as starting points and spent time on interpretation and thesis refinement. Their juniors adopted it immediately but struggled with verification — they trusted AI outputs too readily because they did not have the experience to know what to question. The lesson: you need both. Experienced professionals who verify and judge, alongside juniors who are fast, adaptable, and AI-fluent.

The Training Gap and the Headcount Question

Firms are spending significant amounts on AI tools and licenses. But almost nobody is investing in training their analysts to actually work with these tools effectively. Not a one-hour ChatGPT demo — structured training on verifying AI-generated financial data, spotting hallucinations, using citation-checking, and knowing when to trust the system versus override it. This gap is creating a two-tier workforce within firms, and there is no formal programme to close it.

At SP2, we have started building this into how we prepare our analysts before they join client teams. Our professionals are trained to work within AI-assisted workflows and verify outputs rigorously. We see this as non-negotiable in 2026.

And the headcount question everyone is thinking about: will AI reduce analyst headcount? For some execution-heavy roles, probably yes. But the firms furthest along in AI adoption are not firing analysts — they are redeploying them. The analyst who used to spend three days building a CIM from scratch now spends half a day refining an AI-generated CIM and the other two and a half days on client conversations, thesis development, and deal strategy. Output goes up. The work becomes more valuable. AI is not a headcount play if you know how to leverage it.

What Is Your Take?

Are you an analyst who has figured out how to work effectively with AI — and if so, what changed for you? Has your firm started rethinking what skills to hire for? Or are you a team lead struggling to get your team to adopt AI workflows? I would love to hear your experience.