The company that decided coaching shouldn't be a perk for executives - and built an AI named Nadia to prove it.
Somewhere right now, a newly promoted manager at Delta is typing into a chat window before a conversation she's been dreading. She is not talking to a chatbot that schedules meetings. She is talking to Nadia - an AI coach that remembers her last review, knows the politics of her team, and will role-play the hard talk with her until she stops bracing for it. When she logs off, no one in HR gets a transcript. That privacy is the point.
Valence, headquartered in New York, makes that moment possible at industrial scale. Its product, Nadia, is the most widely deployed AI-native coach in the Fortune 500. Across more than fifty global deployments it has run over a million coaching conversations - the kind of number that usually belongs to consumer apps, not enterprise HR software. The companies paying for it are not experimenting at the edges: roughly a third of Experian's global workforce uses Nadia, and Delta, Kraft Heinz, General Mills, Prudential, WPP, and Analog Devices have all signed on.
Note the pronoun. People say "her." When customers talk about software the way they talk about a new hire, something unusual has happened to the category.
Here is the open secret of corporate development: one-on-one coaching is one of the few interventions that reliably changes how people lead. It is also absurdly expensive. A human executive coach runs hundreds of dollars an hour, which is why coaching has always been rationed to the top of the org chart - the vice presidents, the high-potentials, the people already doing fine.
Everyone below them gets a slide deck and a half-day workshop they forget by Friday. The frontline manager, who actually touches most of a company's employees, gets the least support and carries the most weight. That is the tension Valence exists inside: the development that works can't scale, and the development that scales doesn't work.
The global enterprise AI coaching market was pegged around $20B in 2024 and is forecast to push past $118B by 2029. Valence didn't invent the category - it just got there before most people knew there was a door.
Valence started in 2018 as team-performance software - diagnostics and tools to help managers run better teams. Useful, unglamorous, and a long way from where it would end up. Its co-founder and CEO, Parker Mitchell, had an unusual resume for an HR-tech founder: he co-founded Engineers Without Borders, advised Fortune 500 companies at McKinsey, collected an Ernst & Young Entrepreneur of the Year award and a couple of honorary engineering doctorates, and - the detail that mattered - had worked on small neural networks and LLM precursors back in grad school.
So when ChatGPT landed in late 2022, Mitchell didn't see a novelty. He saw the missing ingredient. Large language models made it possible to deliver genuinely personalized coaching at near-zero marginal cost - the thing that had been economically impossible for a century. The team pivoted hard, started building in December 2022, and shipped Nadia in early 2023. The bet was simple and slightly reckless: that the best use of generative AI inside a company wasn't writing emails faster, but helping people get better at their jobs.
Nadia is not a search box with a friendly avatar. Valence describes it as four roles fused together: the wisdom of a mentor, the contextual knowledge of a manager, the support of an assistant, and the strategic read of a chief of staff. The features follow from that ambition.
Persistent memory means advice builds on your history instead of restarting every session - the difference between a coach and a stranger.
Role-play and simulation let you practice the layoff, the feedback, the negotiation before it's real and unrecoverable.
Multilingual support means a single deployment can coach a genuinely global workforce, not just the English-speaking head office.
Each deployment is tailored and rebranded to the company - "customized to all things Analog Devices," as one customer put it - and integrates with Microsoft 365.
Enterprise software is full of impressive-sounding adoption claims. The honest tell is whether anyone uses the thing twice. Valence's case rests on two stubborn figures: a million-plus conversations, and a Net Promoter Score that sits above 90 - territory reserved for products people actively recommend, and almost unheard of for HR software, which is usually endured rather than enjoyed.
Bessemer Venture Partners led the $50M Series B in September 2025. Partner Sameer Dholakia - a former CEO who also backed Anthropic, Abridge, and EvenUp - took a board seat. Total raised to date is around $75M.
Strip away the funding headlines and Valence's mission is almost old-fashioned: take coaching - historically a luxury good - and turn it into a utility that every employee can reach. The company frames the modern workplace's biggest strains - AI anxiety, burnout, constant reorganization - as human problems first, and argues that the organizations that navigate them best will be the ones that invest in people, not just platforms.
It's a tidy inversion of the usual AI sales pitch. Most vendors promise to remove humans from the loop. Valence is selling AI whose entire job is to make the humans in the loop better at being human - more reflective, more accountable, less alone before a hard decision. Whether that's idealism or excellent positioning is a fair question. The million conversations suggest it doesn't have to be only one.
Return to the opening scene. The manager closes the chat, walks into the conversation she was dreading, and handles it - not perfectly, but better than she would have an hour earlier. Multiply that by a million conversations and you start to see what Valence is actually building: not a smarter chatbot, but a floor under the quality of everyday management at companies with hundreds of thousands of employees.
The market it's racing into is enormous and crowded, and "AI coach" will get slapped on plenty of mediocre products before this is over. Valence's edge is that it started early, picked the unglamorous customer - the frontline manager, not the CEO - and built something people choose to come back to. The coaching that used to belong to the corner office now answers anyone who asks. That's the change. The corner office can keep the corner.