Podcasts

Adiona Tech’s Richard Savoie on An Hour of Innovation: AI, the last mile and fleets that run 40% empty

November 19, 2026

Our CEO Richard Savoie joined Vit Lyoshin on An Hour of Innovation to talk about where delivery costs really sit, how Adiona uses AI, and what he's learned building the company.

An Hour of Innovation podcast episode artwork featuring Richard Savoie

Listen: Episode page · YouTube · Apple Podcasts · Spotify

How Adiona moved from commute optimization to logistics

Adiona started with a different problem. Richard's sister used to drive past three branches of the same grocery chain to get to the one she worked at, and Adiona's first optimization engine set out to fix commutes like hers. It proved hard to plug into HR and recruitment systems, so Richard and his co-founder, Quang Huynh who holds a PhD in optimization and computer science, turned the engine on logistics instead.

Why the last mile makes up 53% of delivery costs

About 80% of Adiona's customers focus on the last mile, because that's where the cost sits. Richard used a book as the example. Of all the transport it takes to get that book to your door, raw materials included, the final trip from the warehouse to your house makes up 53% of the cost. Cost and emissions move together, so it's 53% of the emissions too.

Commercial delivery vehicles also run at about 60% capacity on average, "meaning that most of the time it's traveling 40% empty." So Adiona's first step with a new customer is usually to right-size the fleet and do the same work with fewer vehicles.

How AI fits into route optimization

"AI is one tool in the kit," Richard said. Adiona combines it with historical patterns, live traffic data and constraint-based optimization, a pre-AI technique that has been around for close to a hundred years.

The catch with those older algorithms is computing time, which grows exponentially with every parcel you add. AI helps frame the problem and feed in the history, and Adiona is researching quantum computing to speed up the processing itself.

Simulating your delivery network before making changes

Adiona's customers found a use the team never planned for: simulating their whole delivery network. Richard described how an organization like Australia Post can model taking on a new customer with thousands more deliveries a day, including how many vehicles it would need and how many drivers should be staff or contractors. Since the US tariffs hit, customers have also run best, worst and middle-case simulations of their delivery costs.

One simulation delivered an awkward answer. An Australian alcohol distributor set up collection points in its stores so customers could recycle champagne and wine corks. Adiona's platform showed that shipping small satchels of corks to a consolidation center, and then overseas, cost more and produced more emissions than the recycling was worth. "That was a really tough lesson to learn for a company who was trying to do the right thing," Richard said.

Planning an electric delivery fleet rollout

Richard says Adiona built the first electric vehicle routing solution in the Australian logistics market. EV uptake has been slow, especially in North America, so helping fleets with their EV rollouts is a big focus.

He pointed out that postal services like USPS and Australia Post have moved to EVs faster than many private companies, because government backing lets them take more risk with capital. Adiona's simulations help fleets work out where EVs will pay off with the least risk.

Agentic AI for real-time delivery prioritization

Richard expects agentic AI to change how fleets set priorities on the day. He described a medical supplies company that used an early version during snowstorms, so critical-care supplies reached hospitals on time while other deliveries waited. People and traditional systems struggle to reprioritize like that once the vehicles are already on the road.

Automating dispatch amid supply chain staff shortages

The US has 450,000 unfilled manufacturing jobs, and around 60% of supply chain organizations say they're struggling to recruit. Richard says dispatchers spend about 80% of their time on manual routing and 20% on real problems, like a driver calling in sick or a customer chasing a missed delivery. His suggestion is to hand the 80% to automation and let people focus on the problems that need a human. He also reminds customers they have a duty of care to help long-serving staff move into that work.

Lessons for logistics tech founders

Richard's first lesson for founders is to "find a problem that people are willing to pay for", and to test it like a scientific experiment. Plenty of people say their company would pay for something, and the answer changes once the request goes up the chain.

He also says to "invest in marketing and branding from day one." Most of the people who sign off on a product aren't technical, and they care whether it delivers, not how it works. He also warned engineers about edge cases, "It's so easy to get buried in the edge cases and follow those rabbits down the hole."

Related on Adiona