Professional Documents
Culture Documents
for AI in the
Supply Chain
Who We Are
AltaML is a leading North American applied AI company. With offices
across Canada, our mission is to elevate human and business potential
through ethical applied AI. To do this, we build AI Products that solve
industry problems in partnership with industry organizations. We also
deliver custom AI Solutions with a focus on the following industries:
Healthcare, Power and Utilities, Financial Services, Energy and Natural
Resources, Supply Chain, Public Sector, Insurance and Agriculture and
Forestry. To date AltaML has served more than 80 customers with more
than 300 unique AI projects delivered.
01 Introduction »
08 Conclusion »
PART 01
Introduction
Artificial Intelligence in
Supply Chain and Logistics
They have already proved their worth in gaining visibility and insights into
complex structures of vast scale within other industries.
Its forecasting ability is only as good as the acquired data. It also relies on
human interaction to query the involved data and is dependent on human
understanding to interpret it, which has its natural human limitations.
Predictive analytics also may not account for unknown variables and is
restricted by the confines of the stated algorithm.
Take, for example, ordering supplies for a repair job. Standard forecasting
for inventory and service order planning is driven by static algorithms that
use historical data averages (such as lead times and parts used for a repair
work order) to plan the work orders.
Greater Visibility
Visibility within the supply chain is essential. A recent survey revealed
that 83% of supply chain executives believe a lack of end-to-end visibility
within their organizations was a barrier to growth. Supply chain visibility
today entails the ability to track a widget as it makes its way from the
factory to the assembly plant, wholesale distributor and retailer or
customer. It also includes all the individual components of that widget
as well as its sub-assemblies. That’s a tall order, which is why Gartner
predicts that 50% of global product-centric enterprises will have invested
in real-time transportation visibility platforms. These platforms will
involve fleets of IoT devices, producing sensory data that can improve the
transparency of a component’s route through the supply chain network.
Overall, better visibility and planning drives cost savings, mitigates risks,
and improves customer relationships. By leveraging AI-based solutions,
companies can implement automated systems that can rapidly analyze
troves of actionable data to aid in this kind of decision-making.
Improved Safety
In 2020, the US Bureau of Labor Statistics reported 4764 workplace
fatalities and 2.7 million injuries and illnesses in private business. The
National Safety Council puts the cost of these incidents at a hefty $170
billion in 2019, including wage and productivity losses plus medical and
administrative costs. That doesn’t even include the reputational impact,
which can be costly in ways that are less easy to quantify. AI machine
learning can identify risk patterns that jeopardize the health of warehouse
and delivery workers, ensuring compliance and a safer health environment
that also reduces costs. The integration of computer vision, intelligent
robots and sensors can monitor and identify potential risks, thereby
reducing the need for personnel in risk-prone environments.
Contract Management
AI and ML models can both increase the efficiency of contract reviews
while mitigating risks throughout the contract management lifecycle.
AltaML has developed models using Natural Language Processing (NLP)
to expedite the review of contracts. In addition to ESG applications
identified above, NLP will identify opposing terms and conditions as
well as highlighting key risk areas for consideration at the start of the
contract review cycle. As the contract moves through the lifecycle from
managing to closeout, ML models can predict which suppliers will not
meet their contractual obligations, such as service level agreements, and
automatically identify pricing discrepancies. Take for example a large
services contract for a construction company. ML models can quickly assess
if the invoices match the contract terms and conditions, progress updates,
and labor reports to allow operations and finance to manage by exception.
Supply chain forecasting efforts of the past have relied on historical sales
to estimate future demand. Another way to describe this process is “best
guess” forecasting. AI-driven systems sever the dependency on the past,
extrapolating and unifying internal and external data drawn from within and
outside of the organization in real time using demand sensing rather than old
school predictive analytics. Besides macro forecasting, insightful correlative
data from this kind of demand sensing allows for granular forecasting by
departmental managers. For instance, AI can be leveraged to predict demand
using sentiment analysis so the inventory is proactively redeployed closer
to the customers’ location within the network. ML can also recommend
stocking levels and reorder points at the SKU level for a specific location that
automatically update using internal and external sources on the frequency
your organization determines. Processes are simplified when they are
automated. Supply chain management is no different.
C A SE STUDY
After implementing the final project, the company was able to reduce their amount
of forecasting errors by between 10-20% when compared to the former ERP
system. This not only improved margins, but reduced the company’s required
working capital. The AI driven solution was also able to identify underlying
patterns, insights, and anomalies that the ERP system had failed to recognize,
allowing the company to further optimize the company’s operational performance.
Here we had a large energy company that did not have full visibility on their spend.
The company’s platform, SAP, gives them the ability to complete detailed spend
analytics on the items and services procured using their transactional item and
service master data. The problem stemmed from the fact that there was significant
maverick spend in purchased goods and services that are not procured off an item
or service master, but more haphazardly, using free-text. Determining the spend in
a specific category was made difficult because of all the free-text items.
The use of free text became a real disadvantage when it came to category
management, spend consolidation, budgeting and forecasting. The free text items
couldn’t be consolidated into a category. Without knowing what the spend was
in a given category, the company was unable to effectively leverage their buying
power. This handicapped their ability to negotiate new contracts to attain pricing
considerations and improved terms.
The objective here was to extract the text from these various unstructured
fields, classify the material into the appropriate category and provide a
consistent structure so that the company had visibility into the total demand for
a given category of material.
Conclusion
Thanks to the digitally connected world we reside and operate in, delivery
windows are rapidly compressing as order fulfillment expectations
continue to rise. What’s more, changing consumer trends have elongated
the typical supply chains and global commerce has resulted in complex
supply networks. Because of that, the need for visibility and data driven
decision making to ensure greater efficiency now exceeds the capabilities
of legacy solutions that rely on human intervention.
That’s why optimizing your supply chains and logistics can serve as a
power differentiator and be a competitive advantage for your company.
Contributor:
Brent Willett
Regional Director, North
America West and Supply
Chain Leader
brent.willett@altaml.com