For anyone who has played around with ChatGPT or Midjourney, it’s now clear that AI has the potential to be an incredible tool for enhancing efficiency and productivity. Like all innovations, however, developing technology as complex as the human brain requires time and significant investment.Here, too, the journey of AI began as far back as 1956 at a workshop on Dartmouth College’s campus in the U.S and has endured several cycles of investment booms and busts. Each cycle promised to deliver software that boosts productivity but most fell flat, at least in terms of broader rollout It wasn’t until OpenAI’s introduction of ChatGPT that the general public truly became captivated by the practical applications of a functional AI model.

It’s been about two years since OpenAI publicly released ChatGPT. Amidst this boom, some veterans of the AI industry question whether this cycle will be different, especially as some corporate entities report disappointing returns on their initial investments. Even MIT released a study in January suggesting that AI is too costly to replace human roles that require visual input.

As someone passionate about efficiency and technology in logistics, but not an expert in AI, I was initially very excited about AI’s potential. Like many, I envisioned substantial cost savings and productivity boosts for the logistics industry, which is often hampered by non-standardized formats and requires high adaptability. Historically, we were able to use technology to wrangle standardization for things like pricing and rating, helping ExFreight emerge as the first digital forwarder.

With that in mind, I wanted to-defluff AI and share where I see the initial low-hanging fruits of AI integration for freight forwarders like myself. Logistics technology isn’t about hype and flash, it’s about making this massive, complex industry more efficient. So you’ll see that the use cases aren’t ones that are going to blow anyone away; they are low to mid-sophistication tech that can each help us get a few percentage points more efficient.

So without further ado, here’s how Exfreight has successfully leveraged AI in cost-effective ways to address some compelling use cases:

Customer-Facing Chatbot:

Logistics is a service business. We’ve implemented commercially available chat tools integrated with base LLMs like ChatGPT, all fine-tuned to our business operations. Our customer-facing chatbot assists customers with inquiries about service levels, required documents, and basic tracking updates.This is faster, on-demand service, all based on a backend of structured data that we’ve developed over years, that make our customers happier and free up our employees to work smarter and more efficiently.

General Customer Service Email Filtering and Responses: Speaking of external customer service – with an LLM, there’s no need to differentiate between email and chat. So we are currently experimenting with using the trained LLM developed for our chatbot, combined with filtering tools, to automate responses to various email inquiries. This system would efficiently handle routine questions, freeing our staff to address more complex customer needs. Again, this takes careful and meticulous training. In logistics, the cost of an error or exception is always where problems start to balloon.

But it’s certainly not only about how we work with customers. Here’s a few ways AI is shaping our operations internally.

Internal Training Chatbot:

Knowledge transfer is always a challenge. So looking inwards, we’ve used a modified version of our external chatbot to provide a resource for employees and new hires to ask complex questions about scenarios not covered in their initial training or SOP guidelines. Once trained with our internal SOPs, the chatbots handle straightforward queries, escalating more complex issues to supervisors as needed. The system learns from these interactions, continuously improving its responses.

While the technology still has limitations, such as occasional inconsistencies in responses (“hallucination effects”), there are many areas where AI can make a significant impact. These use cases are still in early days on our side but are ones that we are actively looking into:

Accounts Payable and Commercial Invoice Digitization: Automating the conversion of emailed invoices into digital formats reduces manual data entry. Whether it’s commercial invoices, invoices or other documents we see on a regular basis, there is a prime opportunity here to reduce manual error and speed up the transition into structured data that, in turn, feeds the chatbots mentioned above.

Automated Phone Attendants and Full Truckload Tender Negotiations: This may be the most futuristic idea we’re toying with. AI-driven systems that can work quickly and with strong natural language processes can handle phone queries from truck drivers and negotiate truckload rates in real-time (ironically, potentially with a trucking company’s AI on the other side). Additionally, they can perform check in calls for status updates which automatically update TMS systems.

ERP System Decision Making: Integrating AI into ERP systems could automate purchasing decisions based on real-time data on shipping rates and capacity. The industry has shifted towards more digital procurement already, with instant pricing, rating and even eBooking; this could take it one step further.

In conclusion, the role of AI in logistics is expanding, and the notion that AI could entirely replace human jobs is as exaggerated as the fears once associated with computers entering the workplace. AI will enable us to automate mundane tasks and refocus our human resources on more strategic activities. Modern forwarders do not need to force AI into every aspect if their business but if there’s one takeaway I’ve had from the past ten years of tech, it’s that embracing the right tech at the right time and in the right place can have a huge impact on businesses going forward.