November 25, 2025
fotosedrik/Depositphotos.com
Would Recording In-Store Associates Help as a Training Tool?
Ethosphere, a Seattle-based startup that recently secured a new funding round, is launching a retail coaching tool that records the in-store conversations of store associates. The tool then leverages AI to analyze associate performance and offer feedback.
With the Ethosphere platform, store employees wear lightweight, watch-sized microphones during their shifts to record their interactions with customers. To avoid privacy issues, only the employee’s side of the conversation is recorded, not the customer’s.
Tapping LLMs [large language models] and voice AI technology, those conversations are transcribed and assessed on how well the associate follows the retailer’s selling approach — particularly around aspects such as greeting, understanding the customer’s needs, explaining the product, and upselling. The platform then offers praise, as well as suggestions for improvement to further develop selling skills.
Ethosphere said on its website that the platform allows retailers to “deliver unparalleled guidance, in your brand voice, tailored to each associate.”
Examples of such guidance for associates from Ethosphere’s website are shown below:
Evan Smith, Ethosphere’s CEO (and formerly VP, technology strategy and business transformation at Starbucks), told Geekwire that some employees enjoy the gamification element of platform. He also noted that retail’s high turnover rate comes partly due to subpar training and lack of feedback. He said, “Feeling seen, feeling heard, and feeling recognized is really a big piece of the puzzle that, in some respects, is missing in retail.”
For managers, the use of AI promises to “de-bias” how they view their teams, allowing them to better celebrate high performers and understand the weak spots of underperformers. Retailers also gain access to a dashboard, where they may evaluate selling areas in which a store’s team might collectively need more training.
The tool further enables retailers to “comprehend which managers are developing and moving their teams to customer experience greatness.”
Ethosphere told Geekwire the firm is currently working with quick-service restaurants, apparel stores, and luxury shops.
Sri Chandrasekar, managing partner at Point72 Ventures, which led the new funding round, said, “We see their focus on augmenting, not replacing, human interaction as a critical differentiator, making their solution a deeply compelling answer in the evolution of retail.”
Ethosphere: An Evolution of Past Retail Associate Training Tools?
Recording customer conversations for training purposes is already fairly common on customer service calls. On a 2023 Reddit post, a user claiming to be a service rep said they regularly hear feedback on their calls.
The rep wrote, “There is quality control guide that lays out what we should do, and how many points for each segment. (Greeting, 2 points; your name, 2 points; company name, department, offer to help…. confirming customer name, address, email, phone numbers, etc. ) Points can also be deducted for more subjective things like tone, too much silence, keeping caller on hold too long or failing to recognize opportunity to up-sell. If average quality falls below 90% you miss bonuses for the month.”
BrainTrust
Paula Rosenblum
Co-founder, RSR Research
Kevin Graff
President, Graff Retail
Bob Amster
Principal, Retail Technology Group
Recent Discussions
Discussion Questions
What do you think of a tool such as Ethosphere that records store associates’ conversations with customers and enables AI to critique the encounter?
Do you suspect store associates and managers would welcome, or revolt against, such a tool?
Poll
Loading …
I can see some merits to this. However, I can also see a lot of downsides, including invasion of privacy, making some employees nervous, and being too prescriptive about how employees should interact. The latter point is important, as some of the best associates for customer service are those who allow a bit of their own personality to shine through and do not come across as overly corporate. I am sure AI can analyze a lot of things and come up with suggestions, but it’s not terribly good at understanding the complexities of nuance and emotion that are critical to human interaction.
There was a company that did something similar a decade or so ago…that had cameras and microphones in the name tags. Just seems creepy to me.
But it’s a handsome companion piece for demanding employees smile if they’re within 10 feet of a customer
I can see an alarm going off every time an employee doesn’t smile. What wonderful uses we are finding for AI…
. One reason I decided to retire is my inability to find compelling use cases for AI in retail. This one is epic
Do I really believe that it is only observing one side of the conversation – wait a moment let me ask my eavesdropping TV about that. And then training and scoring based on (if you believe it) only one side of the conversation? Color me skeptical.
I turned off Siri on all my devices. No confidence that they really throw what they hear away
I have mixed feelings about a tool like Ethosphere that records conversations between store associates and customers, then uses AI to critique the interaction. On the positive side, the concept aligns with a long-standing retail need: better frontline coaching, more consistent service standards, and a quantifiable way to surface selling moments that work well or fall short. Ethosphere’s model of capturing only the associate-side audio and using LLMs to deliver feedback signals a thoughtful approach to operational insight.
That said, I suspect many in-store roles—both associates and managers—would have reservations, and rightfully so. The prospect of having every customer interaction recorded can feel invasive, especially if the purpose, scope, and privacy controls are unclear. If the tool is framed strictly as surveillance or compliance rather than development, it risks eroding trust and harming morale. My recommendation is clear: pilot with transparency, clear purpose, opt-in where possible, and ensure the tool supports coaching—not punishment.
In short: I believe this kind of technology can enhance training and unify service quality, but only when rolled out thoughtfully. The frontline team must feel respected, empowered, and supported—not monitored. As a former retail practitioner, I’d insist on a pilot, clear KPIs (store conversion lift, customer satisfaction scores, associate retention) before a chain-wide rollout, and a strong change-management plan that places people ahead of technology.
Let me get this straight: the app records employee conversations then uses AI to analyze what the employee did and or did not do? And the app creators think that’s fun? Nope. There is not a chance I’d put my team through this.
Given the high rate of employee turnover, real-time feedback like this can be very valuable. Reaction to this technology at the Store Operations Council’s last meeting was very positive.
Not to be Debbie downer, but I think it’s laziness or trying to shove 10 pounds of feathers into a 5 pound bag. Store managers are overworked. But you gotta be human
I agree with my colleagues. it’s a slippery slope. Something similar is done by many companies when they record a telephone interaction between a CSR and a customer “for training purposes” and those conversations are not natural, but rather measured and cold.
This technology wins based on the Hawthorne Effect, which basically says when you’re being watched, you always do a better job. Consider how much better most people wash their hands in a public washroom when someone else is there. And watch how well they wash their hands when it’s a unisex hand wash station!
I’ve seen this in action with a competitor to Ethosphere and it drives much better behaviour and results every time.
Worried about being recorded? Better throw out your phone, I guess.
When done right, the feedback given to staff is about 80% positive, so it feeds the need for positive feedback. In the end, the staff perform better, results improve, recognition and rewards improve.
And some of the A.I. “conversations” I’ve had are sometimes more human than those with actual humans.
The Braintrust correctly identifies privacy concerns and the surveillance creepiness factor. But is Ethosphere solving the wrong problem? Retailers already know what good service looks like. The real challenge is that knowledgeable, engaged associates leave for jobs that pay better and treat them with more dignity. Adding surveillance (even “supportive”) with gamification doesn’t address compensation, scheduling unpredictability, or meaningful career paths. The claim that AI will “de-bias” manager evaluations ignores that AI doesn’t eliminate bias; it encodes the biases present in its training data and design choices. Ethosphere should not be positioned as the solution to retail’s people challenges; however, it can work as one component of a comprehensive development program, with complete transparency and genuine coaching intent. Be careful what you wish for: you ask for a genuinely engaged workforce, and you get an algorithmically compliant one.
I think it’s useful at provoking reactions (I’ve never seen the commetariat reach an even dozen so fast!)
As for the product itself: is it really an improvement on how things are done now? Are we really to believe the biggest hindrance to customer service isn’t – oh,…maybe understaffing – but rather a lack of AI critiques?
I see absolutely no value in this because it simply does not fit how human beings learn — at least those working in retail. By analogy, young humans can run down stairways three steps at a time — but only if they aren’t thinking about what they are doing in any detail. It is sad that AI opens the opportunity for tech to invade human reality so deeply. It’s also no surprise as management by micro-metrics was enabled by computers, now management by nano-metrics is enabled by the algorithms collectively known as AI. Human beings need to relate to human beings and that cannot happen while surveilled nano-second by nano-second.
I think a tool like Ethosphere can work well if retailers keep the transparency piece front and center. In other parts of the business, especially sales and marketing, teams already use call-analysis tools like Clari to coach toward a better process. This feels like a natural extension of that idea into the store.
The friction will mostly be around the device itself and the sense of being monitored. If associates clearly understand when the recording is active, who can see the feedback, and how it will (and won’t) be used, adoption should improve. Most store teams genuinely want to deliver good service, and structured coaching based on real interactions can help them get there.
There will always be a little discomfort early on, but with the right guardrails and communication, this kind of AI-supported coaching becomes less about surveillance and more about giving associates a useful tool to grow.
Ever worked in a call center? When I did back in the 1990s, they monitored your calls for coaching and that was without your knowledge. It could happen on any given call at any time. This is the modern-day version of the same thing using current tech. I’m not sure how this works, though, if only one side of the conversation is recorded since it is the interaction that needs to be assessed.
This is an excellent training tool. While AI can record and make suggestions, a manager providing human-to-human coaching is the recommended follow-up. When used correctly, this is a great tool. Used the wrong way – to spy on associates – you will get pushback, and even a “revolt” as mentioned in the discussion question.
Without training what success looks like, this is the dumbest idea I’ve heard of. !/2 conversations graded by AI to coach performance is like grading a conversation while pretending the other person doesn’t exist. Terrible.
More Discussions
Should US Consumers Be as Upbeat Over Their Finances as Data Suggests?
A new study suggests U.S. consumers are rosy over their finances, but should they be?Has Storytelling Become a Rediscovered Retail Art?
Are retailers finally leaning (again) into narrative to excite?Will the Gulf Between Walmart and Target Deepen in 2026?
Will Walmart widen its lead over Target in 2026?
I can see some merits to this. However, I can also see a lot of downsides, including invasion of privacy, making some employees nervous, and being too prescriptive about how employees should interact. The latter point is important, as some of the best associates for customer service are those who allow a bit of their own personality to shine through and do not come across as overly corporate. I am sure AI can analyze a lot of things and come up with suggestions, but it’s not terribly good at understanding the complexities of nuance and emotion that are critical to human interaction.
There was a company that did something similar a decade or so ago…that had cameras and microphones in the name tags. Just seems creepy to me.
But it’s a handsome companion piece for demanding employees smile if they’re within 10 feet of a customer
I can see an alarm going off every time an employee doesn’t smile. What wonderful uses we are finding for AI…
. One reason I decided to retire is my inability to find compelling use cases for AI in retail. This one is epic
Do I really believe that it is only observing one side of the conversation – wait a moment let me ask my eavesdropping TV about that. And then training and scoring based on (if you believe it) only one side of the conversation? Color me skeptical.
I turned off Siri on all my devices. No confidence that they really throw what they hear away
I have mixed feelings about a tool like Ethosphere that records conversations between store associates and customers, then uses AI to critique the interaction. On the positive side, the concept aligns with a long-standing retail need: better frontline coaching, more consistent service standards, and a quantifiable way to surface selling moments that work well or fall short. Ethosphere’s model of capturing only the associate-side audio and using LLMs to deliver feedback signals a thoughtful approach to operational insight.
That said, I suspect many in-store roles—both associates and managers—would have reservations, and rightfully so. The prospect of having every customer interaction recorded can feel invasive, especially if the purpose, scope, and privacy controls are unclear. If the tool is framed strictly as surveillance or compliance rather than development, it risks eroding trust and harming morale. My recommendation is clear: pilot with transparency, clear purpose, opt-in where possible, and ensure the tool supports coaching—not punishment.
In short: I believe this kind of technology can enhance training and unify service quality, but only when rolled out thoughtfully. The frontline team must feel respected, empowered, and supported—not monitored. As a former retail practitioner, I’d insist on a pilot, clear KPIs (store conversion lift, customer satisfaction scores, associate retention) before a chain-wide rollout, and a strong change-management plan that places people ahead of technology.
Let me get this straight: the app records employee conversations then uses AI to analyze what the employee did and or did not do? And the app creators think that’s fun? Nope. There is not a chance I’d put my team through this.
Given the high rate of employee turnover, real-time feedback like this can be very valuable. Reaction to this technology at the Store Operations Council’s last meeting was very positive.
Not to be Debbie downer, but I think it’s laziness or trying to shove 10 pounds of feathers into a 5 pound bag. Store managers are overworked. But you gotta be human
I agree with my colleagues. it’s a slippery slope. Something similar is done by many companies when they record a telephone interaction between a CSR and a customer “for training purposes” and those conversations are not natural, but rather measured and cold.
This technology wins based on the Hawthorne Effect, which basically says when you’re being watched, you always do a better job. Consider how much better most people wash their hands in a public washroom when someone else is there. And watch how well they wash their hands when it’s a unisex hand wash station!
I’ve seen this in action with a competitor to Ethosphere and it drives much better behaviour and results every time.
Worried about being recorded? Better throw out your phone, I guess.
When done right, the feedback given to staff is about 80% positive, so it feeds the need for positive feedback. In the end, the staff perform better, results improve, recognition and rewards improve.
And some of the A.I. “conversations” I’ve had are sometimes more human than those with actual humans.
The Braintrust correctly identifies privacy concerns and the surveillance creepiness factor. But is Ethosphere solving the wrong problem? Retailers already know what good service looks like. The real challenge is that knowledgeable, engaged associates leave for jobs that pay better and treat them with more dignity. Adding surveillance (even “supportive”) with gamification doesn’t address compensation, scheduling unpredictability, or meaningful career paths. The claim that AI will “de-bias” manager evaluations ignores that AI doesn’t eliminate bias; it encodes the biases present in its training data and design choices. Ethosphere should not be positioned as the solution to retail’s people challenges; however, it can work as one component of a comprehensive development program, with complete transparency and genuine coaching intent. Be careful what you wish for: you ask for a genuinely engaged workforce, and you get an algorithmically compliant one.
I think it’s useful at provoking reactions (I’ve never seen the commetariat reach an even dozen so fast!)
As for the product itself: is it really an improvement on how things are done now? Are we really to believe the biggest hindrance to customer service isn’t – oh,…maybe understaffing – but rather a lack of AI critiques?
I see absolutely no value in this because it simply does not fit how human beings learn — at least those working in retail. By analogy, young humans can run down stairways three steps at a time — but only if they aren’t thinking about what they are doing in any detail. It is sad that AI opens the opportunity for tech to invade human reality so deeply. It’s also no surprise as management by micro-metrics was enabled by computers, now management by nano-metrics is enabled by the algorithms collectively known as AI. Human beings need to relate to human beings and that cannot happen while surveilled nano-second by nano-second.
I think a tool like Ethosphere can work well if retailers keep the transparency piece front and center. In other parts of the business, especially sales and marketing, teams already use call-analysis tools like Clari to coach toward a better process. This feels like a natural extension of that idea into the store.
The friction will mostly be around the device itself and the sense of being monitored. If associates clearly understand when the recording is active, who can see the feedback, and how it will (and won’t) be used, adoption should improve. Most store teams genuinely want to deliver good service, and structured coaching based on real interactions can help them get there.
There will always be a little discomfort early on, but with the right guardrails and communication, this kind of AI-supported coaching becomes less about surveillance and more about giving associates a useful tool to grow.
Ever worked in a call center? When I did back in the 1990s, they monitored your calls for coaching and that was without your knowledge. It could happen on any given call at any time. This is the modern-day version of the same thing using current tech. I’m not sure how this works, though, if only one side of the conversation is recorded since it is the interaction that needs to be assessed.
This is an excellent training tool. While AI can record and make suggestions, a manager providing human-to-human coaching is the recommended follow-up. When used correctly, this is a great tool. Used the wrong way – to spy on associates – you will get pushback, and even a “revolt” as mentioned in the discussion question.
Without training what success looks like, this is the dumbest idea I’ve heard of. !/2 conversations graded by AI to coach performance is like grading a conversation while pretending the other person doesn’t exist. Terrible.