Not every process needs AI
The most common mistake in AI adoption is trying to apply it everywhere at once. That leads to scattered pilots, diluted focus, and nothing reaching production.
The smarter approach is surgical: find the 3-5 workflows where AI will deliver disproportionate value, prove it there, then expand. But which workflows? That is what this module gives you — two practical methods for identifying the ones that matter.
How does your firm currently identify processes for AI integration?
Method 1 — Follow the Document
Every knowledge worker's job ultimately produces documents. Memos, models, reports, presentations, emails, filings. The "Follow the Document" method traces the lifecycle of your firm's most important documents to find where AI can accelerate, improve, or automate steps along the way.
Pick a document your team produces regularly — a research report, a DD memo, a client update, a regulatory filing — and trace its journey:
- What triggers it? (A new deal, a quarterly deadline, a client request)
- What inputs go into it? (Source documents, data pulls, prior reports)
- Who touches it and in what order? (Analyst drafts, VP reviews, MD approves)
- Where does time get consumed? (The first draft? The data gathering? The formatting?)
- Where do errors or rework happen? (Version control issues? Inconsistent formatting? Missing data?)
Every point where time, errors, or rework concentrate is a candidate for AI integration.
Method 2 — Follow the Hours
The "Follow the Hours" method is simpler and more direct. Ask your team one question: Where do you spend the most time on work that feels mechanical?
Not strategic work. Not relationship work. The mechanical work — the hours spent gathering data, reformatting documents, cross-referencing information, writing first drafts of routine deliverables, checking compliance against checklists.
This is where AI delivers the fastest, most measurable value. Not because it replaces judgment, but because it compresses the mechanical work that currently consumes 40-60% of a knowledge worker's day.
In your experience, where does the most mechanical work happen at your firm?
Financial services workflow categories
In financial services, most AI-eligible workflows fall into four categories. Thinking in categories helps you ensure you are not missing opportunities in a part of the firm you do not personally touch.
Deal & Research — Sourcing, screening, due diligence, research production, market analysis. These are high-volume, document-heavy workflows where AI excels at compression and synthesis.
Client & Reporting — Client communications, portfolio updates, pitch materials, quarterly reports. These are recurring, template-driven workflows where AI can produce strong first drafts.
Compliance & Risk — Regulatory monitoring, KYC/AML checks, policy review, risk reporting. These are rule-based, high-stakes workflows where AI improves both speed and coverage.
Operations — Data reconciliation, trade support, document management, internal reporting. These are high-frequency, low-judgment workflows where automation has the clearest ROI.
Which workflow category consumes the most time at your firm?
The workflow documentation template
Keep it simple. For each workflow you identify, capture these five elements:
| Element | What to capture | Example |
|---|---|---|
| Trigger | What kicks off this workflow? | New deal enters pipeline |
| Steps | The major steps from start to finish (5-8 is enough) | 1. Receive CIM 2. Extract key financials 3. Build comp table 4. Draft memo 5. Review cycle 6. Final approval |
| Output | What deliverable does this produce? | Investment committee memo |
| Time | How many person-hours does this consume end-to-end? | ~20 hours per deal |
| Pain points | Where does time, rework, or error concentrate? | Data extraction (4 hrs), first draft (6 hrs), formatting (3 hrs) |
That is it. No process-mapping software required. A table in a document or a row in a spreadsheet.
Your exercise — document your top 3 workflows
Using the two methods — Follow the Document and Follow the Hours — identify and document your top 3 candidate workflows. These should be workflows that are:
- Recurring — they happen regularly, not once a year
- Time-intensive — they consume meaningful person-hours
- Document-heavy — they involve reading, producing, or transforming documents
- Structured enough — there is a repeatable pattern, even if it varies
For each one, fill in the five elements: Trigger, Steps, Output, Time, Pain Points.
These three workflows are the raw material for Module 4, where you will score them using the VALUE framework and prioritise which one to tackle first.
How many of your firm's high-value workflows do you think are formally documented today?
Module 3 — Final Assessment
What is the main risk of applying AI to every process at once?
The 'Follow the Document' method identifies AI opportunities by doing what?
Which characteristic makes a workflow a strong candidate for AI integration?
Why is workflow documentation a prerequisite for AI integration?
