By Shahid Malik Mohammed and Maria Augusta, edited with real-world augmentation.
A Balanced View of AI in Finance
The rise of AI tools like ChatGPT, Claude, and Deepki has sparked both excitement and
skepticism in the finance world. Can these systems truly assist with deep financial work?
Should analysts feel threatened—or empowered?
Our experience shows clearly: AI is not here to replace the analyst – or at least yet. It’s here
to enhance them.
Where AI Shines
1. Summarisation at Scale
AI can digest hundreds of pages in seconds and generate concise summaries, saving analysts
hours.
Real-world example:
An investment analyst reviewing a 200-page offering memorandum can use Claude or
ChatGPT to extract and summarise key lease terms, exit strategies, and IRR sensitivities.
This accelerates review processes during tight due diligence windows. It also works when
the language is not one that you master. You can ‘chat’ with the documents and ask specific
questions, like ‘what is the net asset value in the year 2018?’
2. Structured Storytelling
From building outlines to generating charts and bullet points, AI adds structure to ideas that
are often scattered.
Real-world example:
Using ChatGPT, a junior analyst can convert raw deal notes into a pitch-ready investment
memo or board report. Combined with Excel, the analyst can generate GPT-powered
commentary on financial metrics such as cash-on-cash returns or debt service coverage
ratios. You can add ChatGPT into your financial model in Excel through API without the
need to send the whole dataset to ChatGPT – and risk data breaches.
3. Speed and Productivity with Workflow Automation
AI significantly enhances productivity by automating repetitive tasks and streamlining workflows—allowing analysts to focus on higher-value judgement and interpretation. Rather than replacing expertise, AI tools extend it by generating accurate code snippets, templates, or workflows in seconds, reducing manual effort and errors.
Real-world example:
A real estate debt analyst running monthly covenant compliance checks on a portfolio of 25 loans can use ChatGPT to generate a VBA script or Python code that pulls debt service data from Excel, calculates DSCR (Debt Service Coverage Ratio), LTV (Loan-to-Value), and ICR (Interest Coverage Ratio), and flags loans breaching thresholds.
Instead of manually calculating and checking 75+ ratios across different spreadsheets, the analyst automates the process and reduces the time spent from 4–6 hours to under 15 minutes per review cycle. The AI-generated code includes conditional formatting and auto-generated summary tables that can be shared directly with the credit committee.
This accelerates reporting, ensures consistency, and reduces the risk of oversight in covenant monitoring workflows.
Where AI Needs Guidance
1.Independent Generation Can Mislead
AI sometimes generates incorrect figures or misclassifies terms like liabilities or income. Without human checks, this leads to misinterpretation.
Solution:
Use AI only as a first-pass generator. Always double-check figures, ideally by integrating
your own knowledge into the AI output. That’s where the role of human knowledge really
comes into play. AI can generate a lot of IRR / Sensitivity / Cash Flows that don’t make any
sense.
2. Lacks Financial Context
AI does not inherently understand the implications of a debt covenant, regulatory note, or
equity dilution.
Pragmatic fix:
Train custom prompts or upload company-specific documents (e.g. term sheets or debt
agreements) into Claude or ChatGPT. These can provide context-specific answers but must
be framed clearly and confirmed manually.
3. Hallucination Risk
AI might fabricate projections or cite figures that aren’t in the source document.
Preventive steps:
- Use Retrieval-Augmented Generation (RAG): Upload actual documents and allow AI to
reference directly. - In tools like ChatGPT Enterprise, activate the “quotes with sources” feature to ensure all
claims are traceable.
The Right Relationship: Analyst + AI
Used well, AI is the analyst’s strongest co-pilot. It helps structure analysis, highlight blind
spots, and rapidly test different scenarios.
But the analyst brings domain knowledge, context, and judgement—the very things AI lacks.
Final Word
AI is not your competitor—it’s your competitive edge.
The analysts who master AI won’t be replaced by it—they’ll be the ones leading the future
of finance. The difference is not whether you use AI—but how strategically and responsibly
you integrate it into your workflows.
“Don’t just prompt AI. Design smart workflows where AI supports your thinking, not
replaces it.”
Our course ‘AI Application for Real Estate’ builds this exact knowledge, you learning not
only to prompt but also to use the power of AI to build smart workflows with python, VBA,
javascript with no prior coding knowledge needed.