AI Training Should Change How Finance Teams Work
The real opportunity is not better prompting. It is better financial analysis, supported by AI. Surojit Chakraverti of Third Wave Capital on training the workflow, not the tool.
Key takeaways
- The real opportunity is not better prompting.
- It is better financial analysis, supported by AI.
- Surojit Chakraverti of Third Wave Capital on training the workflow, not the tool.
Written by the team at
Third Wave Capital
London, United Kingdom
4.9(149 reviews)
Third Wave Capital delivers practitioner-led corporate finance and investment training in financial modelling, three-statement modelling, LBOs (leveraged buyouts), DCF and valuation, M&A analysis and execution, private equity, leveraged finance, credit analysis, ECM and DCM.Led by a former Citi and
- Graduates struggle to apply finance theory at work
- Financial models lack accuracy and commercial insight
- Teams lack confidence in valuation, M&A and LBOs
The real opportunity is not better prompting. It is better financial analysis, supported by AI.
For: Heads of L&D, team leaders and AI adoption leads in banks, funds and corporate finance
Reading time: 4 minutes
Case type: Illustrative workflow (not a client engagement)
Author: Surojit Chakraverti, Founder, Third Wave Capital
AI is already changing what finance teams can achieve. I have seen significant improvements in the speed, quality and consistency of work when people learn to use it properly. What matters is not how many prompts someone knows, but whether they can use AI to make their everyday work better.
That means building AI into the work itself: preparing for results, reviewing filings, comparing transactions, testing assumptions and putting together analysis that colleagues can actually use.
The opportunity goes well beyond access
In a 2024 survey, 75% of firms responding to the Bank of England and FCA said they already used AI, while skills remained a significant obstacle [1]. The technology is reaching desks quickly. The bigger opportunity now is to help teams use it well.
Used thoughtfully, AI can reduce time spent finding information, make reviews more thorough and leave analysts with more time to think commercially. Those are meaningful gains, not simply productivity statistics.
Train the workflow, not the tool
I do not train people to memorise prompts or become experts in one product. I start with a real task and show how to build an AI-enabled workflow around it: what to give the system, what to ask it to do, where the analyst steps in and how the finished work gets checked. The aim is to make AI complementary to the person doing the job.
Better work still needs judgement
AI can review documents, compare disclosures and help with first drafts at a pace that changes how teams work. It is not infallible. An experiment with BCG consultants found substantial gains on tasks well suited to AI, but worse performance on a task beyond its capabilities [2].
That is a reason to train people properly, not a reason to avoid the technology. In finance, checking figures, spotting what has genuinely changed and understanding the commercial implications remain part of the job.
What this looks like on results day
Illustrative workflow
A simplified equity research example, showing how an analyst could organise an AI-enabled results review.
Before. The analyst reads the results release, filing and transcript, searches for what changed since last quarter, updates the model and writes a note. Much of the time goes on gathering and comparing information.
With AI in the workflow. The analyst starts with the investment questions, then uses an approved tool to compare disclosures and management commentary. The tool highlights changes in guidance and performance, with sources attached. The analyst checks the material figures, works out what matters for the investment case, updates the model and produces the note. AI speeds up the groundwork; the analyst does the analysis.
The improvement comes from combining faster retrieval and comparison with an analyst who knows the company, challenges the output and can update the model sensibly. The same approach applies to transaction screening, covenant reviews and many other recurring tasks.
Build checks into the process
Teams should work in approved systems, protect confidential and non-public information, and trace important figures back to source. These are not reasons to keep AI separate from the job. They are part of designing a workflow that people can use confidently.
Measure what has improved
Choose a few recurring tasks and compare the work before and after training:
Speed: is the team producing a useful first draft sooner?
Quality: are figures checked and key changes identified more consistently?
Rework: are fewer rounds of correction needed?
Adoption: are people still using the process after 30 and 60 days?
Capacity: is there more time for judgement, discussion and higher-value analysis?
I have seen better outcomes when training is organised this way. It gives people a practical method they can take back to the desk, rather than an interesting demonstration they soon forget.
A practical place to start
Pick one task your team repeats every week. Rebuild it with AI supporting the steps it can do well, and the analyst responsible for verification and judgement. To see the approach in practice, explore Finance OS on L3VLUP, alongside our financial models and practice tools. These are built to make financial analysis more accessible, practical and repeatable.
About the author. Surojit Chakraverti is the founder of Third Wave Capital and L3VLUP, and a former investment banker at Citi and Rothschild & Co. He designs and delivers practitioner-led training in financial modelling, valuation, M&A and practical AI-enabled workflows for finance teams.
Sources
[1] Bank of England and Financial Conduct Authority (21 November 2024). Artificial intelligence in UK financial services - 2024. bankofengland.co.uk/report/2024/artificial-intelligence-in-uk-financial-services-2024
[2] Dell’Acqua, F. et al. (2023). Navigating the Jagged Technological Frontier. Harvard Business School Working Paper 24-013. hbs.edu/faculty/Pages/item.aspx?num=64700
Ready to find your training provider?
Search our marketplace of 8,000+ training providers.




