AI x Sustainability (3): Supply Chain Compliance with AI

Supply chains are complex systems where every link—from raw material sourcing to final product delivery—ideally adheres to stringent standards of efficiency and compliance.

Reality does not look always quite as ideal though, as we all know. And that’s one of the origins, or rationales, for the existence of supplier audits as well as H&S and environmental legislation etc.

Figure 1 below illustrates how complex supply chain can turn out to be. Indeed, it may be more suitable to talk of a ‘supply network’ rather than a ‘supply chain’, because few if any of the links between partners are in the bigger system scheme of things purely unidirectional.

Many of the desirable optimisations and desired improvements in such a network are far from trivial to achieve. They require the acquisition, management and alignment of large amounts of not so easy to obtain data, as well as complex trade-off assessments …

At the same time new technologies, amongst them distributed ledgers (block chain) and real time data and sensor monitoring, create ever more data, and ever more complexity, ever more ‘under exploited potential’.

Reverting back to the illustration on AI maturity levels by Unit8, supply chain management can benefit from AI notably at step 3 of the maturity scale. Further, and while still early in the development, first Step 4 type of applications are becoming available.

In the remainder we’ll be looking at a small number of areas where AI could offer relevant benefits to supply chain management, many of which are ultimately related to sustainability-relevant efforts.

• Supply Chain Risk Assessment and Resiliency
• Resource, Process, Partner Optimisation Support
• Synergies with (other) cutting edge IT Technologies

Supply Chain Risk Assessment and Resiliency

Managing risk as part of supply chain management is a rather broad field of needs and requirements. Hence the potential risk application areas can be diverse: some may be directly linked to e.g. labour standards, while others are related to e.g. trade regulations or indeed part of the financial audit realm. The fundamental goal is of course to pro-actively manage the risk.

• Compliance Monitoring: AI application can of course, and I have mentioned this in the previous blog post on reporting, be designed to take on some of the compliance monitoring against regulatory standards and internal policies.
• Support Supplier (Sustainability) Risk and Capability Assessment: The same capability as above used for monitoring are not very far off to conduct assessment for the ‘above and beyond [compliance]’ component of ESG, drawing from all available resources, both internal, external, open source or proprietary. Be it a supplier’s reports, certifications, and public disclosures (if they exist) or indeed carbon emissions, water usage, and audit outcomes. I
• Help prioritize efforts to mitigate high-impact issues and related risks: the identification of the areas in need of the biggest improvements including a first proposal on remediation can be a help to human expert to not get overwhelmed by and lost in the abyss between too much detail and too much data.

Resource, Process, Partner Optimisation Support

Supply chains are inherently always being optimised. Be it because consumer demands shift, because materials are harder to get hold off, because quality slips, partners are non-collaborative in regards to compliance or best practice approaches, or because import/export taxation schemes change. AI applications do already exist in selected areas to support and take some heat off the teams.

• Inventory & Planning Optimisation: At the base of historical sales data and patterns, market trends, lead times, storage capacities, and market trends, AI can help to determine what ‘suitable’ inventory levels look like. In doing that, it could support the teams to find the ‘best possible’ point for reordering goods, or indeed make the case for phasing out entire products or product lines.
  This can push the envelop by integrating risk insights and allow for scenario based planning support – hence enabling companies to hedge their bets for when things can and do go wrong.
  Far from this being futuristic, exiting applications like Project44 or PredictHQ are already on the case to do exactly that.
• Logistics Optimisation: Logistics have numerous variables that play a crucial role. Transportation routes (with fuel and carbon costs, vehicle write depreciation), order and delivery windows, locations for warehouses and DC to just name a few. Every such decision has a considerably complex number of knock-on effects. AI applications can help discovering patterns, evaluate different options against one another – also dynamically, i.e. ‘on the go’ – and gear the outcomes towards pre-set goals (e.g. reduced carbon emissions, lower warehouse-to-customer shipping time, …).
  And as logistics is a key ingredient in all circularity efforts, AI can even help here: by identifying opportunities for material reuse, recycling, and waste reduction; or by analysing product life cycles to suggest opportunities for recycling and reuse.
  There are already different AI support tools that offer some of those options:
  Coupa for the supply chain design aspect, Route4Me for the ‘Warehouse to Customer’ route planning.
• Partner Evaluation and Selection: Using risk assessment insights (see above) and combining it with goal trajectories, legal / contractual aspects, product design and costing data, and sustainability performance data, there is opportunity to score and grade supply chain partners, or run and evaluate a roster of existing and potential new ones against industry as well as internal benchmarks. The beauty of this is: Tier 1 is what teams typically already are dealing with, if still very manually. By adding the power of tech, it is becoming easier to cover further supply chain tiers, hidden deeper and deeper in the maze of Figure 1.
  Companies trying to tap into such a market are GEP, Jaggaer, or Icertis

Synergies with (other) cutting edge IT Technologies

Real time monitoring, IoT, Blockchain / distributed ledgers. The range of processing power enabled novel IT-tech applications has unsurprisingly proliferated over the last 10 or so years. But in which combination could AI add particular benefit in the sustainability sphere?

Two of those come to the forefront:

• Full-depth Digital Product Passport – with ‘real time option’: The digital product passport tracks and shares detailed information about a product’s lifecycle, from raw materials to end-of-life. This offers itself up to sophisticated approaches as follows: Use of real-time IoT sensors to ensure that each component complies with environmental and ethical standards, and blockchain integration to create an immutable record of every transaction. To illustrate this: A fashion brand could use AI in this way to monitor the water or chemistry footprint of its clothing throughout the supply chain, with the digital product passport provides consumers access to this data.
• Digital Twin: Digital twins are virtual replicas of entire supply chains, allowing companies to simulate various strategies and predict outcomes without disrupting real-world operations. By integrating sustainability into these simulations, companies can design supply chains that minimize carbon emissions, energy use, and resource consumption. Digital twins also enable predictive risk analysis, allowing businesses to forecast potential disruptions, such as extreme weather, and develop contingency plans. AI enhances these digital twins by analysing vast amounts of data to optimize decisions. For example, a retailer can use AI to simulate the impact of different shipping routes on both costs and carbon emissions, choosing the most efficient and eco-friendly option.
  An already existing software platform to do that is Microsoft Azure’s Digital Twins platform.

Collectively, these approaches enhance supply chain visibility and accountability, and support companies in getting to grips with the many details and components of a circular economy.

Challenges

Not all is plain sailing of course, and every gold medal has a flip side. This is no different when reflecting about the benefit (or lack thereof) of AI in supply chain management and related sustainability issues.

Interestingly, the challenges we encounter seem to be fairly consistent across all application areas of AI – or indeed novel technology at large:

• Data Quality and Reliability: AI’s effectiveness relies on high-quality, accurate data; poor data can lead to incorrect risk assessments and decisions. Or as mentioned in a previous post: ‘crap in = crap out’ …
• Integration with Existing System: Merging AI with current inventory and logistics systems can be complex and resource-intensive.
• Investment and Resource Allocation: Implementing AI requires significant financial and resource investment. Careful consideration of costs and benefits is therefore a not to be underestimated precondition.
• Supplier Data Availability: Reliable supplier evaluation depends on the availability and accuracy of data. In many geographies supply chain partners keep struggling with their basic infrastructure to even remain in business – never mind even thinking of more cutting edge approaches.
• Bringing People Along: Adopting AI often requires changes to procurement practices and obtaining buy-in from stakeholders. Bringing people along on the journey can be complicated due to the fear many have of either becoming redundant or failing to understand the practical use of the new tech at hand.

Conclusion

AI presents compelling opportunities for advancing supply chain sustainability by streamlining processes and leveraging data more effectively. Its potential to optimize manual tasks and enhance decision-making is significant.

However, many supply chain partners are still in the early stages of digital transformation, focusing on basic processes like recruitment and payroll. This will likely hinder reaping the benefits and rewards of AI anytime soon.

And last but not least: It is not tech that ‘does’, but ultimately people. Ensuring people’s well being and safety, mentally as well as physically, remains as essential as ever. When it comes to AI this is in as far more important than ever, as it is at the same time the key component that will ‘make or break’ this technology’s fruitful usage going forward.