How AI is Changing the Art & Science of CPM Scheduling Dr. Dan - - PowerPoint PPT Presentation

How AI is Changing the Art & Science

• f CPM Scheduling
• Dr. Dan Patterson, PMP

BASIS CEO

• Two decades of analytics: bettering project plan integrity
• Evolved CPM through risk-adjusted scheduling & critiquing
• The next step, driving plan realism
• We’ve actually been implementing AI for a long time

BASIS Legacy

Critical Path Method

• 1956 – CPM invented
• Dupont/Remington
• UNIVAC-1 computer
• Today - same algorithm
• 15 lines of code
• Generates dates & float
• Dates are NOT inputs

The Shortcomings of CPM

It It’s not exe xecution that is letting us down…

• CPM plans are overly optimistic, best case
• Don’t encourage sound use of building blocks
• Industry breeds schedulers not planners
• Gantt chart has not evolved in 100 years

The Original Gantt Chart 1795 Harmonogram

Artificial Intelligence

“A “Any device e that per percei eives es its en environmen ent & ta takes acti ctions

• ns th

that t maxim imiz ize its its ch chanc nce of

• f

suc success ss at so some me goal.” AI is simply the next step in the evolution of computing power

Examples of AI

How many of us use AI in our personal lives? What about in our professional lives?

• Siri & Alexa
• GPS Guidance
• Waze or Google Maps
• Uber and Lyft
• Machine learning to optimize driver

delivery

• Commercial Airlines
• Only 7 minutes of a flight on a Boeing

airplane involves ‘human-steered’ flight

• Credit Card Fraud Protection
• Uses a neural net to predict fraudulent

transactions

• IoT & Connected Devices
• Fleet Management
• Equipment Uptime Optimization
• Cybersecurity

Wh Why aren’t we using it to build a more realistic plan?

• Ar

Artificial Intelligence: ability for computer to perform tasks that normally require human intelligence

• Au

Augmented Intelligence: supplements human thinking rather than replacing it

• Makes our lives easier by performing

tasks faster and with greater efficiency

• Still requires human intelligence,

reasoning, and expertise

Artificial or Augmented

“What at a computer is is t to m me is is it it’s t the mo most re rema mark rkable tool that at we’ve ev ever come up with, an and it’s s the eq equivalen ent of a bi bicycle e for our mi mind nds.”

Augmented plan building Active benchmarking Knowledge capture & re-use Consensus-Based Achievable Plan Incorporate team expertise Consensus analysis Review cycle & plan commitment

Artificial Augmented Intelligence Human Intelligence

Knowledge-Driven Planning with BASIS

What about all that data?

Source: PwC 2017 Global Digital IQ Survey

63 63%

Average percentage of organizations that believe they do NOT effectively utilize the data they capture to drive business value

That begs the question(s)… 1. Does AI Planning really need ‘big data’ to be effective?

• 2. Can I trust the suggestions made?

Expert (Knowledge-Based) Systems

Knowledge base + inference engine Rule-based, e.g., IF…AND…THEN… Domain-specific, e.g., planning

Neural Networks

Learn by example/pattern, e.g., face recognition Not task (domain) specific Requires history or supervised learning

Knowledge Library Inference Engine Learning Facts

BASIS Expert System

Expertise

BASIS Neural Network Weighted Inputs Hidden Layer BASIS Planner or Team-member Updated Weights

BASIS AI Approach

• Uses Expert System to make suggestions, Neural Network to learn
• Makes planning suggestions based on rules
• Automatically adjusts ‘weighting’ of each rule
• Doesn't require ‘big data’

Wh What is it?

• Knowledge-driven planning
• Guided plan creation/validation
• Analyzes realism
• Incorporates team consensus

Result: A more achievable plan BA BASIS Appr pproach

Sketch

• Top-down
• Timelines

Plan

• Detailed plan
• CPM-based

Markup

• Feedback
• Expert Opinion

Consolidate

• Consensus
• Buy-in

All four modules designed together to make planning behavior be better r & more re efficient

BASIS Software Introduction

1) 1) Sketch ch

• Top-down planning
• Planning packages: timelines
• Set deliverable/contract dates
• WBS dictionaries/templates

2) 2) Plan

• Detailed planning
• Work packages & tasks
• Detailed logic/calendars
• Full CPM analysis

Building Schedules in BASIS

A Smarter, More Natural Way to Plan

A.I. used to suggest timelines & benchmarks A.I. used to detail schedules

3) 3) M Markup

• Unique markup layers
• Simple experience
• Durations, dates, risks, etc.
• Visualize impact

4) 4) Con

• nsoli
• lidate
• Review contribution
• Analyze consensus
• Flatten into plan
• Visualize impact

Building Schedules in BASIS

A Smarter, More Natural Way to Plan

H.I. used to capture expert opinion H.I. used to drive consensus

• Offers real-time suggestions
• Uses powerful inference engine
• More than just a search engine
• Understands context
• Progressive emphasis
• Confidence score in assessment

How Does BASIS Offer Suggestions?

BASIS Inference Engine

• Offers real-time suggestions
• Uses powerful inference engine
• More than just a search engine
• Understands context
• Progressive emphasis
• Confidence score in assessment

How Does BASIS Offer Suggestions?

BASIS Inference Engine

How Does BASIS Learn?

Artificial Intelligence

Se Self Learning

• Progressive emphasis
• Confidence impacted by emphasis

Hu Human Tea eaching

• Influence suggestions through rules
• Project or corporate rules
• Establish patterns without natural

matches

Both drive a more natural planning approach

Te Team Member Markup

• ‘Sandbox’ (markup layers)
• Determine buy-in & consensus
• Consensus is key
• It’s okay for team to push back as long

as there is consensus on changes

• Buy-in without consensus reflects

‘chaos’ in your project

Ma Markup Review & Plan Consensus

Planned Duration: 40d TM1 Markup: 60d TM2 Markup: 60d Planned Duration: 40d TM1 Markup: 35d TM2 Markup: 60d

Human Intelligence Input

Plan Validation

Demonstration

Step 1: Sketch

Manually add new planning packages. Drag/drop to create logic links between planning. Use Waypoint analysis to benchmark your plan. Automatically create multiple planning packages in sequence or in parallel. Interrogate the Knowledge Library to import predefined scope, planning packages, & benchmarks.

Step 2: Build Plan

Manually create activities and milestones. Drag/drop to create logic links between activities and milestones. Track plan alignment through BASIS dates. Define planning windows, e.g., time windows when specific scope or work must be completed. BASIS indices track alignment, detail, & how continuous activities are

• ver the

duration of the project.

Step 3: My Markup

Understand the impact of your markup. Drag/drop activities that you believe need resequencing or different durations. Identify risks, issues, action items, etc. BASIS will suggest common risks, etc., as you review. ‘Accept’ or ‘Change’ each line item that you have been asked to

• review. Markup

individual activities or work packages as a whole.

Step 4: Consolidate Markup

Determine the impact on durations and dates. Consensus analysis shows how much alignment there is between team members. Review all contributions for each activity & choose either the consensus or a specific team member’s opinion to commit to the plan. If team members have changed start dates, these can be modified by either a constraint (default)

• r lags. Lags will give

a more free-flowing schedule than constraints. Analysis determines how much buy-in versus push back & how team member

• pinion impacts the

plan.

A. A.I.

• Access: Organization’s knowledge

is accessible & useful

• Speed: Plan creation is

accelerated

• Quality: Plans are based upon

standards, benchmarks & history

• Completeness: Plans inclusive of

total of scope H. H.I.

• Ownership: Promotes buy-in &

plan acceptance

• Feedback: Quantify team

contribution & inclusion

• Closed-Loop: Refine Knowledge

Library with validated plans

• Efficiency: Minimizes need/time

required for interactive planning sessions

So How Does Intelligent Planning Help?

A Smarter, More Natural Way to Plan

For the latest updates & BASIS content subscribe @

www.basisplanning.com/subscribe