1. Automation Roadmap
Automation is not an all-or-nothing leap; itβs a progression
Start smallβpick one process to automate.
Build on successesβ¦
by Parikshit Padture
2. Purpose
Understand the Technology
Understand the Roadmap to Implement
Technology
ChatGPT, Copilot, Gemini
Prompt Engineering
Decide Actions
β’ Identify Pilot Use Cases for AI
β’ Decide AI Tools
β’ Upskilling and Change Management
3. Digital and Physical Process
Transformation
Digital Processes- in office
β’ Digitization of workflows with apps
β’ Integration of apps
β’ Centralized data
β’ Automate data entry by collecting real
time data from sensors
β’ Analyze data and make predictions and
decisions
Physical processes- in plants
β’ CNC Machining
β’ Robots for Cutting, Welding
β’ Robots for Material handling
β’ Optical vision for inspection
β’ Robots for floor sweeping
4. Level Description Example
1. Process Optimization
Remove redundant steps; streamline
routing; eliminate bottlenecks.
AS-IS process mapping and TO-BE
process design
2. Standardization
Use standard formats, checklists,
masters, and norms for consistency.
Create SOPs
3. Digitization
Implement applications for CRM,
purchasing, inventory, and finance,
etc
Tally, Inventory management, QMS
4. Portability Access apps on mobile/web.
Mobile/ Web app for production status
tracking and DCR-NCR authorization
5. Collaboration Tools
Shared workspace, document
sharing, Citrix licensing.
OneDrive/SharePoint, GoogleDrive for
live project documents
ERP system for Order-to-Cash,
Technology Implementation Roadmap-
DIGITAL
5. Level Description Example
7. Data Analysis & BI
Actionable insights from central
data.
WIP and Project Closure
dashboards
8. IoT Data Entry
QR/barcode scanning for quick
data capture.
Scan QR codes on stock items to
fetch details, productivity data
from data loggers/ sensors
9. Computer Vision CCTV for remote supervision. View shopfloor from office.
10. OCR & RPA
Automate data
entry/reconciliation.
Three-way matching of PO-GRN-
Invoice
11. Rule-Based
Automation
Automated actions (e.g., PO
generation).
Auto-generate PO at reorder
levels.
Technology Implementation Roadmap-
DIGITAL
7. Progressive technology implementation
β’ Each stage of automation builds on the previous one
β’ Lower levels rely on process standardization and workforce training (you
cannot automate chaos β processes must be well-defined first).
β’ Mid-levels require systems integration investment (connecting machines,
collecting data).
β’ Highest levels demand advanced tech adoption (IoT, AI) and change
management, ensuring the organization can utilize and maintain these smart
systems.
β’ Generative AI in a Nutshell - how to survive and thrive in the age of AI
8. What is AI?
β’ Now computers (AI) understand human languages and no more needs coding to
give instructions.
β’ Till now humans learned computers, now in the age of AI- computers will learn
humans.
β’ AI predicts the next step in a sequence based on patterns in past data.
β’ AI needs to be taught like a fast-learning kid but its output needs to be checked.
AI Tools
1. AI Assistant- helps in performing a task: ChatGPT/ Gemini
2. Integrated AI assistant: Copilot
3. AI based workflow automation- Performs complete tasks: n-eight-n,
Zapier
9. AI Tool Selection
1. Affordable, subscription-based, or open-source tools that require minimal setup
and no specialized consultants, making them perfect for quick wins.
2. AI assistants
β’ ChatGPT and Gemini free is available for all employees
β’ All HODs who have MS-Teams will get Copilot in Office 365
β’ No need to copy and paste
β’ AI integrated within the MS-Office suite
β’ Other HODs will get ChatGPT Plus
β’ Deep Research
β’ Custom instructions- do not repeat instructions
β’ Create custom GPT
2. AI Agent
12. Role-Playing Prompt Format
Recipe for a Good Prompt = Be Specific + Provide Context + Set the Tone (formal or casual, direct or
indirect)
β’ Act as a [ROLE]
β’ Context: [CONTEXT / BACKGROUND / MOTIVE]
β’ Inputs: [INCLUDE INPUTS / FILES / DATA SOURCES HERE]
β’ Task: [ACTION VERB] [DESCRIBE REQUIRED OUTPUT]
β’ Format: [SPECIFY OUTPUT FORMAT β e.g., plain text, paragraph with heading, bullet points, table, image]
Example:
βAct as a design engineer. Context: Need to prepare cost estimate. Inputs: Heat Exchanger BOM.
Task: Calculate total cost. Format: Table.β
β’ Ask AI to fact check!
13. What is Reasoning in AI?
Reasoning means the AIβs ability to:
β Analyze information logically β understand relationships and
implications.
β Break down complex problems β step-by-step thinking.
β Draw conclusions β even from incomplete or conflicting information.
β Handle instructions with multiple steps β not just one command at a
time.
ChatGPT-Paid: Selecting correct AI
model
19. AI Use Cases
Category Examples
Search and
Research
Search information from internet, attached files, pre-trained stds and books, subjects.
Deep search on competitors, patents, procurement strategies, etc.
Text
transformation
Translation
Fact verification
Summarization
Grammar check
Format change
Text/ document
generation
Content writing from topic/ short prompt
Write story, poem
Spec to TDC->Review MTC with TDC-> prepare Check Test Plan for discrepancies
SOPs, user manuals, legal contracts
Formats/ Templates for ERP dashboards
Storyboard for videos
Job descriptions
Checking and
Challan and Invoice with PO
Fabrication drawings with datasheet
20. Category Examples
Image Generation Create images of designs for presentations.
OCR
Extract Text from images
Extract Text from drawings
Calculations Get ball-park price estimates, basic design calculations
Data Analysis
Forecasting and Decision making based on past data
Analyze sales, purchase, stock data from ERP
Programming Get macro for repetitive task in Excel
AI Use Cases
21. Pilots
β’ Prioritize use cases by feasibility and ROI
β’ Input data must be clean
β’ Identify AI pilots
β’ low risk
β’ high impact
β’ Quick-Win
β’ Eg: Generate Excel PTR (equipment summary) from GAD
β’ Generate TDC from specifications
β’ Budgetary quote from heat exchanger datasheet
β’ Extract special requirements from project specifications
22. Change Management
β’ Leadership to:
β’ Assign an AI champion in each AI tool and department.
β’ Identify AI use cases
β’ Get hands on experience
β’ Drive AI usage by regular monitoring and celebrating success stories
β’ Establish a centralized data strategy with clean, structured, and accessible
data.
β’ Digitize key data (production logs, design files, maintenance records).