SWITCH Agentic AI: The task-oriented agent built for urban mobility

Featured Content Partner

Share this article

Italian tech startup SWITCH has launched an Early Access Program for its brand new Agentic AI platform, with signups now open.

The SWITCH AI Agent combines simulation, forecasting and real-time decision-making to automate planning, routing and operations for urban mobility and logistics fleet operators.

“This technology was developed as a direct response to the challenges facing firms across the new mobility space,” Matteo Forte, CEO and Founder of SWITCH, tells Zag Daily.

“During the development phase, we worked directly with shared micromobility operators, car sharing platforms and food delivery services. That gave us firsthand insight into what tools the industry actually needs.”

How it works

Prior to the official launch of SWITCH in late 2022, Forte had been working on the creation of a car sharing consumer app.

After observing the evolution of the new mobility industry, he shifted focus to a business-oriented Agentic AI platform.

“We saw major challenges around both operations and planning, and asked ourselves how we could solve them. These obstacles include unpredictable demand, idle vehicles and wasted energy, inefficient dispatching, scattered data across tools and decisions made on gut feeling.”

So what exactly is Agentic AI? 

“Agentic AI refers to AI systems with a certain level of autonomy that can make decisions to complete tasks,” Forte explains. “It’s intentionally a broad definition, because these agents can have varying degrees of autonomy – from very low to very high – depending on what they’re designed to do.”

In the case of SWITCH, its AI Agent can forecast demand, optimize field operations, simulate strategic changes, and analyze performance in context. 

“It’s built to understand an operator’s goals, connect them with real-world data, and take meaningful action,” says Forte.

“You can communicate with the tool via chat, and it works alongside operations, product and planning teams in real time to reduce complexity and speed up decision-making.”

While the interface leverages advanced large language models (LLMs) tailored for fleet operations, SWITCH has built domain-specific tools for forecasting, optimization, analytics and simulation. This is in addition to connectors and semantic layers for integration with existing platforms like fleet management systems and Slack.

“The idea is for the SWITCH AI Agent to function like a team member – one that can answer questions and take actions based on the specific context of your operation.”

Tried and tested on the ground

Before launching the Early Access Program, SWITCH partnered with Milan-based new mobility operator Wayla, which launched its on-demand vanpooling service in October 2024.

Wayla turned to SWITCH to help tackle early strategic decisions, like fleet sizing and service area design.

“Using forecasted data about demand for the service, we simulated the launch ahead of time,” Forte explains.

“Wayla then rolled out the launch based on the recommendations from that simulation, including how many vehicles to deploy and where. When we reviewed the results four months later, the simulation engine was over 92% accurate.” 

Another successful use case came from SWITCH’s work with Italian micromobility operator Elerent.

Over a three-month period, Elerant used the Urban CoPilot tool to help manage the operation of its fleet of e-scooters.

“After just a few months, Elerant reported a 25% improvement in operational efficiency,” Forte recalls. “That means less manual tasks and more accurate forecasting of when and where vehicles will be needed.”

The Early Access Program

While SWITCH’s Agentic AI tools have been utilised to great success so far, Forte acknowledges that they’re still in their early stages. 

That’s why SWITCH is now inviting companies to join its Early Access Program, with the goal of refining the platform in close collaboration with the industry.

“The tools our clients have been using are now evolving into an agentic system,” he explains.

“So we’re giving selected companies white-glove onboarding and support. That includes helping integrate our tools with their internal platforms and combining both internal and external data streams.”

The Early Access Program is expected to run for six to eight weeks, with the first seven days dedicated to onboarding and setup.

“It’s completely free to join; our goal is to understand the real-world value this platform can offer. Meanwhile, participants will get early access to new tools and the opportunity to shape their development.”

Applications for the programme are now open, with Forte expecting to select 10 to 15 firms to take part.

“Companies operate on different timelines, but most want to get a working product in three months,” Forte says.

“We plan to start onboarding within the next couple of weeks and then spend the remaining time focused on real deployment. We’re especially excited to partner with firms that move fast and operate fleets of all sizes – from emerging mobility startups to established global players like ride-hailing companies, micromobility operators, car subscription platforms, and fleet management technology providers. 

“We’ve already seen applications from companies of all sizes, from those managing small fleets of vehicles to large-scale operators managing hundreds of thousands.”  

Share this article

Photography by