Ryan Millner and the Problem of Being Heard

Before Ryan Millner had a sales department, he had a deadline. In conversations with potential buyers for Unwrap, he would ask about their work, describe a possible product, and promise something to try in six weeks. Then, six weeks later, he returned with a working demo. A calendar reminder is a modest foundation for a company. It does, however, have the advantage of being testable.

Millner’s account of those first sales contains the usual startup ingredients: introductions, awkward asks, prices that needed revisiting. The revealing detail is the return visit. He was asking people to believe he understood a problem they lived with. Delivering something gave them a better reason to believe him than another conversation would have.

Today he is the co-founder and CEO of Unwrap, based in Santa Barbara. The company helps businesses make sense of customer feedback scattered across different channels. Its premise comes from an everyday professional embarrassment: being responsible for understanding people while knowing you have read only a fraction of what they are telling you.

Alexa had plenty to say. So did its users.

At Amazon, Millner worked on Alexa. Questions about the product were often quite concrete. Were its answers the right length? Was it missing something people wanted to ask about? Were the visuals on devices with screens useful? Each question could influence what a product team built next. None could be settled simply by admiring the technology.

The people using Alexa were expressing opinions in reviews, support tickets and online communities. Their words existed. The difficulty was assembling them into an understanding broad enough to guide a roadmap. A memorable complaint could get attention; a less memorable pattern could remain scattered across hundreds of messages.

That tension gave Millner and his colleague Ashwin Singhania an idea. Natural-language processing already helped their product interpret language. They could apply related techniques to the language customers were leaving behind. The ambition was practical: give product managers a way to learn from feedback at a scale that defeated manual reading.

A familiar face across the table

The partnership preceded the idea. Millner and Singhania had worked together at Graphiq, the Santa Barbara startup previously known as FindTheBest. Amazon acquired it in 2017, bringing both into the orbit of Alexa. Their eventual departure for Unwrap was a return to building a small company after years inside a large one.

Millner’s education was in mechanical engineering at Duke. His professional work took him toward knowledge graphs and product management: organizing information, making it usable, deciding what a product should do. At Graphiq he became a group product manager. At Amazon he led product work on the semantic language engine and managed knowledge engineers.

The next step took him through the AI2 incubator as an entrepreneur in residence. By July 2022, Unwrap was publicly launching. It had five employees and between five and ten paying customers at the time. Those numbers are useful company biography. They restore the small scale that tends to disappear once a startup acquires a list of recognizable customers.

Santa Barbara remained the base. The founders already had professional relationships there, including connections from Graphiq. Local investor ScOp’s Kevin O’Connor had founded that earlier company. In this story, geography has a business function: a place can hold the colleagues, trust and introductions needed to start again. An ocean view is pleasant; a person willing to join an unfinished company is rather more useful.

The first customers knew somebody

Millner approached more than 100 potential buyers while exploring Unwrap. He describes reaching back through his network, including former colleagues and people he had not spoken to for years. The first ten customers came through people the founders knew or introductions from those people. His own compact description of the outreach was: “I had no shame.”

There is a human detail in that sentence. Asking an old acquaintance for help requires accepting that the relationship may briefly become awkward. Founders can make this sound like a system once it works. At the beginning, it is still one person deciding to send a message to another.

Pricing required its own education. Millner credits a conversation with Amplitude’s Spenser Skates with encouraging him to think about charging more as the value justified it. Unwrap’s route toward larger contracts became part of his account of reaching its first million dollars in annual recurring revenue. The person responsible for the product was learning the buyer’s arithmetic, too.

The complaint outside the filing cabinet

Millner distinguishes between sorting complaints into subjects a team already recognizes and discovering a subject it has overlooked. The second task is central to how he talks about Unwrap. A product manager may know ten problems and want them ranked. An additional problem, absent from the original list, changes the conversation more fundamentally.

His explanation gives a useful test for any claimed insight. It should be specific enough to act on, with a sense of how many customers it affects. It should also be traceable: the team should be able to inspect the entries behind the conclusion. A confident paragraph can sound persuasive. Being able to open the underlying complaints gives colleagues something sturdier to debate.

In describing his engineering philosophy, Millner places the customer’s outcome ahead of allegiance to any particular technical approach. He says the company has used different models and NLP methods, with human checking as needed. That is an operational position with consequences: a team has to judge the quality of the result before celebrating how automatically it arrived.

There is also a social requirement. Several departments may care about the same feedback while asking different questions of it. A shared way to inspect that material makes it possible to discuss priorities without everyone arriving with a private selection of anecdotes. The useful meeting begins when people can disagree about the decision while looking at the same evidence.

Bring your own difficult data

Unwrap’s evaluation process gives that philosophy a commercial form. Millner describes a 30-day proof of concept using the prospective customer’s own data. The company and buyer agree on goals, and the buyer gets to test the working platform. This asks more of a product than a carefully prepared demonstration.

One example he gives concerns an outage revealed during a trial. The customer had not realized it was happening; it was still occurring when the finding was shown. The team went to fix it. An incident like that gives an abstract phrase such as customer intelligence an address: here is a problem, here are the people encountering it, and here is something to do.

Millner’s tests for a successful trial extend to the working relationship. Were the findings accurate? Was the platform usable? Did the team respond to questions? A buyer is evaluating people as well as software. For a founder who won early business through personal conversations, that is a recognizable continuation of the original sales method.

He uses similarly concrete examples when explaining value. Sunrun identified customers asking for additional energy storage and passed the information to sales. Rad Power Bikes made spare parts easier to buy after seeing related support demand. These are examples Millner offers of feedback changing a commercial decision. The interesting unit is the action that follows the analysis.

The founder gets some feedback

Listening is less comfortable when the subject is your own company. Millner has been candid about underinvesting in Unwrap’s early website. At an event in Seattle, he asked entrepreneur Gaurav Oberoi for advice on growing beyond a handful of customers. The response pushed him to consider how credible the company looked from outside.

His retrospective criticism is specific. The product was evolving faster than its public introduction. Visitors could judge only what they saw. It is an amusing predicament for a business devoted to understanding customers: prospective customers were forming an opinion before they ever got near the carefully built software. Even a listening company needs to make itself understood.

Fundraising produced another correction. Reflecting on two rounds, Millner wrote that he initially failed to explain the long-term vision clearly enough and designed the pitch deck himself. His verdict on that second choice was concise: “I’m a terrible designer”. He recommended getting feedback from friendly investors before approaching the most important prospective ones.

“I’m a terrible designer”

Ryan Millner, on an early fundraising deck

In January 2025, Unwrap announced a $12 million Series A led by Scale Venture Partners. Atlassian Ventures joined, alongside investors including former Stripe CTO David Singleton and Perplexity co-founder Johnny Ho. Millner’s public celebration thanked customers and earlier backers as well as the new arrivals. The announcement also brought a recruitment push across the company.

A reason to enjoy the work

The company’s public life contains ordinary pleasures alongside the product discussions. Millner celebrated a team gathering in Mexico City in 2024. Later, writing about a trip to Cabo to mark 2025 and discuss the year ahead, he described a message he had shared with colleagues: “Enjoy The Fight”.

He explained it as taking satisfaction in working through hard problems together. It is a useful sentiment to put beside the funding announcement. Raising money produces a date that people can celebrate. Building a company fills the days between those dates with decisions that rarely warrant a press release.

Unwrap’s stated ambition is to help fill the world with products people love. Millner’s route toward it has involved reading complaints, calling acquaintances, returning with demos and accepting criticism of his own work. The promise at the beginning remains a good description of the job: listen carefully, make something useful, and come back when you said you would.