Front page
InterviewCX AI7h · 14:37 BST · 5 min read
Introducing Harmony: The New Voice AI Sales Platform from monday.com
Harmony has launched as a new voice AI platform to help sales teams follow up with leads, qualify opportunities, and manage customer conversations at scale. Head of Harmony, Vitali Margolin, explains why the company is betting on specialized models rather than general-purpose AI.
Voice AI has quickly become one of the most visible applications of enterprise AI. But while automated voice tools can handle simple, repeatable interactions, sales conversations place much higher demands on speed, judgment, and control.
That is the premise behind Harmony, a new voice AI platform launched by monday.com. Built using technology from an acquisition, Harmony is designed to help revenue teams manage the calls that are often missed when lead volumes rise, sales teams are stretched, or customers need a response outside traditional working hours.
Speaking to CX Today, Vitali Margolin, Head of Harmony, said the platform was created as monday.com explored where voice could fit into the next generation of CRM.
What Is Harmony, and What Makes It Different?
Harmony is a voice AI platform designed to help revenue teams manage more customer conversations, without positioning AI as a straight replacement for salespeople.
The platform grew out of monday.com’s acquisition of voice AI company OneAI. Rather than simply embedding that technology into its existing CRM product, monday.com chose to launch Harmony as a standalone platform, which can connect with the wider system over time.
Its initial focus is enterprise sales. Harmony’s agents can support prospecting, discovery, lead qualification, objection handling, and warm handoffs to live reps. The platform also includes call center capabilities such as predictive and power dialing, inbound routing, branded caller ID, consent and compliance controls, and native CRM write-back.
But Margolin says its core value is giving businesses more capacity at the moments when human teams cannot scale fast enough.
“I’m not here to replace your team. I’m not. My AI will not close a $50,000 deal for you. It’s not what I’m here for. But it will help you achieve your potential, the full potential of your business.”
Harmony is not presenting voice AI as a substitute for high-value, relationship-driven sales conversations. Instead, it is designed to prevent revenue loss from slow follow-up, limited coverage, or the inability to contact every lead and customer at the right time.
For Margolin, that is particularly relevant during periods of high demand. A sales leader cannot simply add thousands of trained reps when a campaign, promotion, or seasonal peak generates more interest. AI agents, by contrast, can be deployed as demand changes.
“There’s no dial. I can’t just click a button and then get 50,000 more sales reps. But with AI, you can.”
Why Does Sales Need a Different Type of Voice AI?
Harmony’s other point of difference is its sales-specific design. Margolin argues that sales conversations cannot be treated like standard support interactions.
A customer service agent may need to retrieve information, route an issue, or complete a routine request. A sales agent, however, may need to identify the right contact, qualify their interest, respond to objections, discuss a competing provider, and recognize when a live representative should take over. As Margolin highlights:
“Having a sales conversation is very different than having a support conversation.”
Harmony uses smaller, task-specific models intended to handle specific parts of that process. The company says these models can help agents recognize voicemails, navigate gatekeepers, qualify prospects, capture data, and respond to objections during live calls.
Why does this matter? Margolin explains that the distinction becomes clearer once an agent leaves a controlled demonstration and enters a real customer environment.
A sales agent may call a prospect but reach an assistant, for example, or hear that the business is already using a competitor. In those moments, the system needs to understand the situation and adapt rather than simply follow a fixed script.
“You need to do all those things when you’re doing a sales conversation, inbound or outbound,” Margolin said.
Why Does Speed Matter in an AI Voice Conversation?
Harmony says its proprietary models can produce a response in around 200 milliseconds from audio input to audio output. That measure may sound technical, but it directly affects whether a voice interaction feels natural.
Long pauses can make an automated agent sound unnatural, particularly in a sales conversation where a prospect is deciding whether to keep listening. Margolin shares that using a large general-purpose model for every element of a live call can create excessive delay:
“It needs to feel real. You can’t use large language models that are very slow in a real-time conversation.”
Harmony’s approach is to split a call into smaller, more defined tasks. An introduction does not require the same level of reasoning as handling a detailed objection. But each task still needs context and a model trained to perform it reliably.
The company is therefore aiming for a balance between low latency and the contextual awareness needed to move a conversation forward. For a revenue team, the intended result is a call that is responsive enough to sustain engagement and structured enough to produce useful information or a meaningful next step.
What Comes Next for Harmony?
Harmony is launching with a focus on voice-based revenue workflows, but Margolin said the longer-term ambition is to create a shared system in which AI and human sales activity can inform one another.
That could mean expanding to channels such as SMS, while bringing human and AI conversations into the same environment. In time, the company wants its agents to learn from the calls handled well by human sellers, without managers having to manually identify and document the best examples.
“I want the AI to learn from it without anyone saying, ‘This was his speech.’ I want it to happen automatically.”
For Margolin, the goal is a feedback loop that improves AI agents through human interactions, while also giving business owners clearer insight into what is happening in every customer conversation.
That wider vision remains ahead. But Harmony’s launch is an early example of how sales technology is shifting beyond assisting individual reps and towards increasing the total volume of conversations a business can sustain.
The real test will be whether voice AI can achieve the balance Harmony is promising: the capacity to scale, the speed to feel natural, and the controls needed to use automation safely in commercially sensitive conversations.
Check out more interviews on customer experience & AI here.
rate this storyhelps rank stories across CX Today
Read nextordered by techtelligence · every pick explained
same beat · CX AI
The End of Automation-Only CX: Why Resolution Is the Real Goal
8hsame beat · CX AIHow Customer Simulations Turn Research Into Valuable Insight15 Oct 2026same beat · CX AIWhy Resolution Is No Longer Enough for Customer Experience Leaders15 Oct 2026
More from the CX AI desk
Karthik Suri, Chief Product and Corporate Strategy Officer at Sprinklr, explains why AI customer service must move beyond containment and automation, helping CX leaders balance efficiency, trust, governance, and more personal customer outcomes at scale
Rob Wilkinson · 8hInterview
How Customer Simulations Turn Research Into Valuable Insight
Customer insight has long been limited by traditional research methods, leaving teams to make decisions with an incomplete picture of their buyers. Gather CEO & Co-Founder Mayank Mehta discusses whether AI simulations can make that insight more immediate, without losing the human perspective.
Sean Nolan · 15 Oct 2026Sponsored · AWS
Why Resolution Is No Longer Enough for Customer Experience Leaders
Pasquale DeMaio, Vice President at Amazon Connect, explains why mediocre service quietly erodes loyalty, how AI can reveal what customers really experience, and why customer experience leaders must move beyond deflection to build more connected journeys.
Rob Wilkinson · 15 Oct 2026
