Scribd
Upload
31 views5 pages

Benchmark_Dow (Gartner CSIG) (1)

Uploaded by

rahulspaliwal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
31 views5 pages

Benchmark_Dow (Gartner CSIG) (1)

Uploaded by

rahulspaliwal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 5

OUTSIDE à IN

Commercializing external perspective

BENCHMARK DOW
TAGS AI/ML Use Cases in Supply Chain

DATE 12/3/2024

ATTENDEES Dave Jackett


Jeffery Tazelaar, Dow Chemical
20+ companies in a round table format

SUMMARY Dow has been on a multi-year journey to digitize its supply chain. They have a Central
Integrated Supply Chain team who helps orchestrate digitization. One team leads
capability development and SC analytics in a POD-like agile structure. They are highly
focused on AI and ML applied to decision making. They are building AI agents to augment
decision making. They are highly focused on business process and speak about the
importance of People, Process and Technology. While they are advanced in the modeling
space, they lack integrated platform like BPM and Palantir to accelerate process
orchestration, automation and make integrated decisions.

Note: Key functions in their supply chain report to different leaders.

NOTES & INSIGHTS

Community Special Interest Group – Gartner

Jeff Tazelaar @ Dow (11 years at Dow)


Runs a central team of Analytics supporting Supply Chain
Entrepreneurial team started 6 years ago (early focus on IOT, Blockchain)
2 years ago took the team offline to focus on team skills, modeling

From: Inside out SC lacking standard process, low tech adoption and limited automation…firefighting
To: Outside-in, proactive SC aligned to the enterprise strategy with a roadmap to standardize, adopt and automate

Integrated SC: Need orchestration since major functions of the supply chain report thru different groups.

Dow 33k vendors


45mm tons shipped
300 whse/terminals
4k deliveries per day
34mm tons of raw material
Digital Supply Chain Design
Digital Twins x3
1. Physical Flow of Raw Material thru Mfg to Customer
2. Demand Signal flow thru Systems of Record
3. Financial Flows of cost thru the Supply Chain

Good view of agents for people and agents working with each other.

SC Decision Support Platform


Integrated Data Hub on Microsoft Azure with Graph on Neo4j
Focus on business process execution
10 levels deep on the knowledge graph…vectors are material transition. Raw material supplier to end customer.
10 second to id all of the flows, impact etc.
Can overlay carbon impact, profit impact etc.
Example: Can answer questions around the impact of a change on profit

Using a stack of technology today

SC Decision Support Platform

HS Code Classification…example of agent to help folks make decisions


Focus on building expert level knowledge into the system. Have humans in the loop for regulatory reasons.
Have input and output trace…now codified vs. word doc
From 60 minutes to 3 sec per material ID w/ accuracy
Master Scheduling Automation

Reinforcement Learning
Used synthetic data to simulate 100,000 years of examples
Love-hate project…drove infrastructure, but hard to deploy into production. Now in pilot.
Challenges getting operational data and getting the models to run in “real-time”. Easy to model, hard to deploy.
Optimizes for margin, reduces “off-grade”
Innovating on the edge of planning systems vs. replacing functionality (similar to OMP + Palantir)

Digital Assistants

Using GenAI to build a LLM based assistant for customer service representatives.
Launched shipper-assistant…Microsoft highlighted it at Ingite
Look at freight costs, assess leakage/opportunity
Built in Microsoft.
Can’t do full automation today (technology). Building knowledge of which decisions are accepted and why…
comfort with decisions to ready for automation.
Lesson’s Learned

Modeling the process is hard…need to focus on process standardization.


Easy to test, hard to scale.
Why it’s hard to scale: People, process, technology…need psychological safety and culture of learning.
AI is not appropriate for every need…think thru readiness, may be overkill
 Need Data, Process, Sponsorship, Culture, Value proposition
Use Agile, but require a business Product Owner or Activator to go.
Have data stewards on his team and on each product
Focusing next on Workforce Planning

You might also like