Before You Automate, Draw the Process on Paper
A practical method for digital transformation: map the real process, find the friction, redesign with the team, then choose automation or AI.
Technology
Business process and intelligent automation
The first automation in Rohit's career was a set of Unix shell scripts that removed a few hours of repetitive work each week. The principle carried forward: automate the dull, repeated, well understood task first, and leave judgement to people.
Quick answer
Quick answer
What is intelligent automation?
Intelligent automation combines ordinary workflow rules with AI for the parts that need reading and judgement of language: classifying a request, extracting details from a document, drafting a reply. It works best as a partnership in which the system handles the predictable majority and routes the unusual minority to a person with the context already assembled.
What to automate, and in what order
- Rules and workflows where the logic is clear and stable
- AI for reading, classifying, extracting and drafting
- Human review for exceptions and consequential steps
- Logs that show what happened and why
- A way to switch automation off safely
The order matters. A process is drawn on paper, simplified, and only then automated. Steps that can be removed are worth more than steps that can be made faster.
Measuring whether it worked
Automation is easy to celebrate and hard to prove. The honest test is a baseline taken before anything changes: how long the task took, how many items came back for rework, how many exceptions appeared in a week. After launch the same numbers are measured again. If they have not moved, the automation is either in the wrong place or sitting on top of a process that still needs redesigning.
The second test is quieter. Do the people who do the work still understand what the system is doing on their behalf, and can they explain it to a customer? If not, the automation has taken away more than it gave.
Insights
Related essays
Where AI Agents Help, and Where They Should Stop
A practical framework for agentic AI: which tasks suit AI agents, how to set levels of autonomy, and why people should approve consequential actions.
What Telecom Billing Taught Me About Data Quality
Lessons from implementing telecom OSS/BSS systems for MTNL on mediation, reconciliation and data quality, and why they matter for AI and analytics.
Questions
Questions about business and intelligent automation
Does automation mean fewer people?
In the organisations Rohit works with it usually means the same people doing work that needs a person. The tasks that automate well are the ones nobody wanted: retyping, copying between systems, chasing status.
Where does automation most often go wrong?
When it is added on top of a process that was never simplified, and when there is no safe way to switch it off or to see what it did.
Conversations
Have a question about this?
A short message with some context is the best way to start.
- Business and intelligent automation