Ashley Moser, Co-Founder and CCO of MelodyArc – Interview Series

Ashley Moser, Co-Founder and Chief Commercial Officer of MelodyArc, leads the company’s commercial strategy, working with customers to translate ambitious operational goals into measurable outcomes. She brings a background spanning operations, product strategy, logistics, sales, and financial services, including leadership roles at Wonder and Walmart’s Jetblack, where she worked across supply chain, continuous improvement, strategy, and logistics. Earlier in her career, she served as Sales Director at Stella Connect, later acquired by Medallia, worked with mobile retail platform NewStore, and spent more than four years at J.P. Morgan in institutional equity sales.

MelodyArc is an enterprise AI orchestration platform designed to help operations teams coordinate data, business rules, AI systems, and human decision-making within a single governed environment. The platform is built around executing complex operational processes end to end, allowing teams to define workflows, apply business logic, deploy AI operators or other agents, and escalate cases to people when greater judgment is required. Its Point Engine evaluates potential paths for resolving each case while working within organizational rules and recording each step, giving businesses greater visibility and control over AI-powered workflows across areas such as customer service, merchandising, site messaging, supply chain, and assortment.

You’ve had a diverse career spanning finance at J.P. Morgan, enterprise SaaS at Medallia, and operational roles within Walmart’s Jetblack. What specific insight or frustration led you to co-found MelodyArc, and how did those experiences shape the company’s core vision? 

We started the company because my co-founder, James McHenry, and I had both been operators and technologists before. We’d seen firsthand, and in many instances, helped build great technology for operations teams. But we kept running into the same gap, point solutions that either addressed a narrow slice of what operations actually did, or didn’t match how the work actually happened on the ground.

With that in mind, we set out to build a platform that could bring new technologies to frontline operations teams in a way that they could actually use. Today, that’s manifested as human-in-the-loop orchestration.

MelodyArc is putting AI directly in the hands of frontline operators. How does this approach differ from traditional top-down enterprise AI deployments, and why is this shift happening now? 

With MelodyArc, once the platform is in place, we put the ability to spin up new workflows, ideas, and human-in-the-loop processes directly in the hands of teams, with little to no impact on the broader roadmap. That means teams can improve their own processes using AI without having to go up and down the chain of command every time. It’s a fundamentally different way of deploying this technology.

This is especially clear in our largest enterprise deployments. At one client, we started with a single use case on one team. Today, we’re live with more than five use cases across multiple teams.

We’ve also seen teams build an AI-assisted process in a single day, something they’d previously deprioritized because it never cleared the ROI bar for what would have been an engineering-intensive build.

Many organizations are investing heavily in AI but still struggle to operationalize it. What are the most common gaps you’re seeing between AI strategy and real-world execution?

Two big things stand out. First, when companies look at a process they want to apply AI to, they often take an all-or-nothing approach. They either automate everything about a focus process, or don’t touch it at all. We encourage companies to instead look at the composition of a process and identify where automation fits, where AI fits, and where people still need to be involved, not necessarily in that order. That approach opens up far more processes as candidates for AI, and it also unlocks more automation downstream, because you’ve been deliberate about where people are looped in.

Second, from a strategic standpoint, companies often have hundreds of eligible use cases to choose from. The key is to evaluate them on real business impact, dollars, hours, and concrete metrics, rather than going after what’s flashiest or easiest. Attacking the biggest, highest-impact areas first, even if they’re harder, is where companies get the most value. This also shows a positive flywheel of value that AI projects can create in organizations.

MelodyArc combines AI with human workflows rather than replacing them. How do you determine the right balance between automation and human decision-making in enterprise environments?

MelodyArc helps enterprises achieve their automation goals by finding the right balance between AI and human-in-the-loop decision making. When MelodyArc begins the implementation process, an organization’s end goal may be for automation to complete 90% of the work, with the remaining 10% relying on team member expertise. But rather than prioritizing speed alone, MelodyArc leverages human-in-the-loop involvement from the very beginning. For example, starting with a 50-50 split between automation and team member input, then incrementally automating more of the process. This approach gets enterprise-level workflows into production faster, because we’re not trying to perfect the workflow before integrating AI.

Another strength of this approach is that as we launch into production, we collect process data that tells us exactly which parts of the process to automate next. This allows MelodyArc to keep incrementing toward automation goals faster and in a way. In practice, this often means a company sees automation improve continuously, without having to disrupt the workflow for fixes or adjustments.

With your background in customer experience and operations, how do you think AI is redefining what “great service” looks like in an environment where responses can be generated instantly?

I believe “instant” is going to become table stakes and essentially commoditized. Great service has never been about getting a customer an answer quickly, but rather getting them the right answer with the least amount of effort. AI gives us new ways to reduce that effort, whether it’s executing on behalf of customers, suggesting next steps, or auto-filling fields – anything that makes an experience snappier and more delightful. Generating an answer instantly means nothing if it’s not actually relevant, and I’d argue that chasing speed alone is how companies end up optimizing for the wrong thing.

Enterprise AI often fails at the final stage, where adoption and usability become critical. What design principles have been essential to ensuring MelodyArc is actually used by frontline teams?

We start from the understanding that there’s a lot of tacit knowledge living with our teams, and we’re very clear about what we can and can’t automate. Being upfront about that distinction makes the experience much more intuitive. It’s not a black box where teams are unsure what’s happening or why.

Usability comes down to the idea that technology works with frontline teams, rather than becoming something they have to work around. That ties back to why we started the company in the first place, too many point solutions either addressed a narrow slice of operations, or didn’t match how the work actually happened on the ground. We found that’s a big reason why many enterprise AI tools sit unused months after launch.

As AI agents become more autonomous, what governance and control challenges should enterprises be preparing for before scaling these systems?

Observability in any enterprise environment is essential, especially the ability to see what decisions were made and why. For processes enterprises could automate but don’t feel comfortable automating outright, I recommend keeping a human in the loop. It’s critical not only to maintain oversight, but to trial new systems to best understand how they work and their impacts, both positive and negative.

From a commercial perspective, how has selling AI into enterprises evolved over the past few years as buyers become more informed but also more cautious?

Everyone has gotten a lot smarter, both buyers and vendors. We’re about four years into what’s really a multi-decade- cycle, so we’re all still learning.. As AI becomes more capable and more available, we’re rapidly iterating on what we can do for clients at the same time they are learning what they can do within their own organizations, which is genuinely exciting because there is a lot of value exchange.

Organizations are focused on information security more than ever, on preserving what makes them distinct, and on the actual efficiency AI delivers. They’re approaching projects more rationally too, willing to increment into adoption rather than go all-in at once. Build versus buy is still very much a live calculation internally, and increasingly we’re seeing hybrid approaches such as build and buy, or “build via buying,” rather than one or the other.

You’ve worked across both high-growth startups and large organizations like Walmart. How does building AI for enterprise operations differ depending on company size and organizational complexity?

There are really two dimensions here: the breadth of the organization, and the historical depth of its tech stack and processes.

At high-growth startups, you don’t have as many people or as much organizational complexity, and you’re not working around systems built 20 years ago or teams that have done the same thing for a decade. The larger and more established the organization, the more you need to rethink ways of doing things from the ground up. That means deciding how you want to operate in the future, then incrementing your way there in a way that fits the organization. Sometimes that also means running independent experiments at the edge, somewhat separate from or in parallel with existing processes.

Looking ahead, how do you see the role of frontline workers evolving as AI becomes embedded into everyday workflows? Are we moving toward augmentation, orchestration, or something entirely new? 

For everyday workflows, an orchestration layer is critical. Getting all your people and technology working together natively is the direction we’re headed. But implementing this type of technology can be complex as it’s not a simple silo where organizations can automate one step then hand it off to an associate. It’s a combination of AI, automation, and an associate working in tandem.

There’s a lot of opportunity for frontline teams to implement workflows. Every time we speak to clients, there are dozens, if not hundreds, of projects they’d love to tackle but can’t, whether it’s budget, tech roadmap, team size, or ROI holding them back. That’s exactly where AI and orchestration come in, it helps  teams do a lot more with what they already have while lowering the cost to launch. This ultimately leads to a positive ROI for future projects.

Thank you for the great interview, readers who wish to learn more should visit MelodyArc.