Axion was privileged to be part of the Salesforce Ventures Analyst Summit panel - a discussion led by John Somorjai, President of Salesforce Ventures on how enterprises are adapting AI, with panelists Evan Smith, CEO of Altana, Axion’s founder and CEO Daniel First, and Diego Lammanto, Chief Marketing Officer at Writer. The following are the key highlights Daniel First’s panel discussion

1. What challenge is Axion solving for U.S. manufacturers?

“American manufacturing is booming, but product failures are a huge issue impacting many different industries. If you think about medical device recalls or electric vehicle recalls, you know, airplanes that are having issues in the hands of airlines, or data centers that are rapidly being built out but facing, you know, power and energy challenges—product quality issues are a major problem for manufacturers across industries. Which means that American manufacturers that are trying to be competitive by releasing high-quality products that deliver exceptional customer experiences are really challenged when the products end up having many failures.

And so, our mission is really to help American manufacturers bring the best products to market that customers love. The problem that manufacturers have is that, though they know who their customers are and they know what their parts are, they actually don’t know what issues are impacting their customers. And when issues arise, it can be months to years until they find out.

Our mission is to help American manufacturers bring the best products to market that customers love.

And so Axion—we are an operating system for customer-centric manufacturing that enables manufacturers to learn about emerging issues that are impacting their customers at the earliest warning signal. And so we work across aerospace with folks like Boeing and RTX, you know, different players in the data center supply chain, whether Cummins, power systems, Daikin HVAC, Schneider Electric, medical devices like Medtronic, and consumer appliances like SharkNinja to help them bring the best products to market that customers love.”​

2.  Why do 95% of enterprise AI pilots fail, while Axion's pilots reach production?

“There obviously was this very well-cited MIT study that 95% of the enterprise AI pilots are not making it to production. And at Axion, actually roughly 90% of our pilots do make it to production. And the reason is that we’re approaching enterprise AI really differently.

You know, many of the reasons that enterprise AI pilots fail is because there’s human knowledge that the AI models don’t have access to, and so users end up not trusting—rightfully so—the models. There’s process issues, which is that you often need process redesign in order to really have healthy adoption. And then there’s often a gap between insight and action, which is, you know, AI models could produce insight, but who’s actually supposed to take action on this?

And so, by virtue of us being hyper-focused on helping manufacturers learn about issues at the earliest warning signal, root cause those problems - why are there L2L3 escalations - and ultimately drive them to fix, we’ve been able to build out a number of specialized technologies that are able to overcome these challenges.

At Axion, actually roughly 90% of our pilots do make it to production.

3. What does ‘human-in-the-loop’ AI look like in practice?

​”We have a human-in-the-loop or expertise-in-the-loop layer in our platform that enables customers to—when they see the AI surfacing insights that don’t match their engineering expertise—they’re able to fix it.

For example, let’s say you’re an engineer at Harley-Davidson and you’re trying to figure out why are my motorbikes failing in the field. The AI might give you one explanation, but you actually have expertise about the way that the wire harnesses work or fail in this particular set of motorbikes. And so we’ve built human-in-the-loop interfaces that both internally at Axion—we have subject matter experts—but also at our clients, you’re able to embed your engineering expertise into the models in a seamless way so that the AI is able to learn from your knowledge of the products, and that’s been really critical for adoption.

4. How does Axion drive cross-departmental innovation?

“When we initially were going to market with our product, it really was used by quality and CX leaders who were trying to learn about the earliest warning signals that there’s an emerging issue on their products. But then for some of our customers, we started noticing that they were adding tens or even hundreds of users onto the platform from other departments. And, you know, so for example at companies like Daikin or SharkNinja, though initially we were used by the quality and by the CX team, we started noticing that there were people on the product strategy and the engineering team who were being added to the platform.

Initially we were used by the quality and by the CX team, we started noticing that there were people on the product strategy and the engineering team who were being added

And when we started speaking with them, we realized that, though initially we were helping the quality and the CX team react to and fix issues that were on the products that had already been released, the product strategy teams and the engineering teams were able to then use the data that was created there as a strategic asset to inform the strategy for the next wave of products. And so I think what’s great is that when you have the right partners, they often will lead you in the right direction in terms of the ways that the data that’s being created by the users using your platform can help other groups.”​​

Learn more about Axion and Daniel First, Founder& CEO here. 

Watch the full panel session below.