The 3-Step Framework to Cut Manual Work by 80%

There was a time in Investment Banking operations where my team and I were the automation.

We were human robots.

In Collateral and Margin operations, we lived inside spreadsheets, macros, rule engines, inboxes, and queues. We invested years optimising manual processes using old technology, fragile logic, and heroic effort.

At the time, it felt advanced.

In reality, it was a workaround.

We were using macros and rule-based systems to paper over structural problems that could not scale and quietly increased operational risk.

That experience is why I built and now teach a different approach.

It is called ACI. In April 2024, the Founder & CEO of intellimation.ai, Har Pulak Bahadur, taught me this three-step process, which has been engrained in me as an AI Product Director.

The Reality of "Efficiency" in Banking Operations

In Collateral operations, the goal was always the same.

• Reduce manual work.
• Increase throughput.
• Avoid breaks.
• Survive the day.

So we:

• Built Excel macros
• Added more rules
• Created exception queues
• Hired smart people to babysit broken processes

It worked.

Until it didn't.

Every new product, counterparty, regulation, or volume spike exposed the same truth.

Rule-based automation does not scale.

It becomes brittle, opaque, and dangerous.

Why Old Automation Creates New Risk

Traditional automation assumes the world is stable.

Banking operations are not.

Rule-based systems:

• Break when data changes
• Cannot reason about ambiguity
• Push humans back into full review mode
• Create silent failures that only appear during stress

This is how operational risk creeps in quietly, hidden behind "automation".

What we needed back then was not more rules.

We needed intelligence with control.

Introducing ACI: Built by Operators, Not Theorists

ACI is not a consultancy slogan.

It is the framework I wish we had when we were running Collateral desks at scale.

• Automation removes the mechanical work.
• Control protects the firm.
• Intelligence learns from humans instead of replacing them.

Each layer exists because of real operational pain.

Step 1: Automation – Stop Treating Humans Like Processors

In operations, humans should not be:

• Extracting data
• Re-keying values
• Sorting emails
• Applying obvious rules

That work belongs to machines.

Automation in ACI focuses on:

• Data extraction from documents, emails, and feeds
• Classification and routing
• Straight-through updates to systems
• Removing keystrokes, not people

This is the easy part.

It is also where most firms stop.

Step 2: Control – Define What Actually Needs a Human

In Collateral, not everything is an exception.

It just looks that way when systems are dumb.

Control is where ACI changes the game.

Instead of reviewing everything, we ask:

Control is designed through:

• Confidence thresholds
• Business tolerances
• Exception definitions
• Audit and traceability

This creates queues with intent, not panic.

Step 3: Intelligence – Turn Exceptions into Training

This is the step we never had in legacy operations.

Using RHLF – Reinforced Human Learning Feedback, Vertical AI operates with two queues.

Here is the difference.

Every human decision trains the AI.

Not next year.

Not in the next project.

Immediately.

Why This Would Have Changed Collateral Operations Forever

In old operations:

• Humans fixed the same issues repeatedly
• Knowledge lived in people's heads
• Exceptions never reduced

With ACI and RHLF:

• Exceptions shrink over time
• Straight-through processing increases
• Humans move from checking to supervising
• Operational risk decreases as volume grows

This is how you cut manual work by up to 80% without increasing exposure.

Why Macros and Rules Could Never Do This

Macros do not learn.

Rules do not adapt.

Spreadsheets do not scale.

They lock knowledge into brittle logic and force humans to compensate.

Vertical AI does the opposite.

• It captures judgement.
• Learns continuously.
• Improves under pressure.

That is the difference between automation theatre and operational leverage.

Where ACI Works Best

ACI thrives where I spent most of my career:

• Collateral and Margin
• Finance operations
• Regulatory workflows
• Document-heavy processes
• High-volume, high-risk environments

This is not generic AI territory.

This is Vertical AI, designed for real operations.

Final Thought

We spent years turning people into machines because the technology could not think.

Now the technology can learn.

When Automation removes the grind,

Control protects the firm,

and Intelligence learns from human judgement,

operations stop being a cost centre and become a strategic asset.

This is why ACI sits at the core of The Vertical AI A.U.T.O.B.O.T™ Playbook.