Emma proves trusted AI powers customer experience at scale
Alexandra Pashkovska of Emma Sleep shares how AI now handles 88% of presales conversations — while enabling more human, empathetic customer engagement at scale.
By Alexandra Pashkovska, CX Digital Transformation Manager, Emma
For most customers today, the ideal support experience is simple: they never have to contact support at all.
As a customer myself, I feel friction the moment I have to send an email, wait for a response or search for answers across channels. By that point, something in the journey is already broken. And I know I’m not alone. Many customers would rather abandon a purchase entirely than navigate a complicated support process.
That belief has shaped how I think about customer experience at Emma. Digital-first CX isn’t about replacing humans with technology. It’s about designing journeys that remove friction, respect the customer’s time and solve problems instantly, without sacrificing trust or empathy.
Our Sprinklr AI agent now resolves 88% of inbound presales interactions end-to-end, across 11 markets. That translates to roughly 3,200 service hours per month handled by AI agents — freeing human agents to focus on complex, high-value cases."
Why digital-first CX starts with self-service
At Emma, we operate a direct-to-consumer sleep brand across multiple global markets. Choosing a mattress online isn’t easy, so customers naturally have questions. They want to know about differences between products, return policies, delivery timelines and financing options.
Historically, many of these questions flowed straight to human agents. They were repetitive, predictable, and time-consuming, yet still critical moments in the customer journey.
Instead of digitizing those manual processes, we made a deliberate decision to design digital-first self-service — not as a deflection tactic, but as a primary CX channel. We had a simple goal: give customers immediate, accurate answers, 24/7, while ensuring they could still reach a human when it truly mattered.
That mindset changed everything.
AI can be more empathetic than we think
One of the more controversial beliefs I hold is that AI agents — when designed correctly — can sometimes deliver more empathetic experiences than humans.
That doesn’t diminish the role of live agents. Human judgment and emotional intelligence are irreplaceable. But empathy in customer service is deeply tied to context, attentiveness, and consistency, and that’s where fatigue becomes a real factor.
Customer service agents manage high volumes of cases every day. After dozens of conversations, maintaining the same level of attention is difficult. But the AI agent doesn’t experience fatigue. It doesn’t lose patience. And it can instantly access context — order history, product information, policy — without placing the burden on the customer.
In this model, empathy is designed.
Tone, conversational structure, escalation logic and guardrails all need to be defined intentionally. Once they are, the AI agent consistently follows those principles — every time. That consistency is often what customers perceive as “being heard.”
Conversations handled by the Sprinklr AI agent consistently deliver higher CSAT and customer effort scores than human agent-led interactions. Customers find the experience faster, clearer and easier to navigate."
From support to consultation
Our data revealed something we didn't initially anticipate: customers aren't just using AI for support. They're using it for consultation.
At Emma, customers send more messages to our AI agent than they do to human agents. With faster response times and 24/7 availability, customers feel more comfortable having longer, more exploratory conversations. They're not just resolving issues — they're learning.
This happens partly because the AI agent operates without the structural pressures that shape human interactions. A human agent works within SLAs. There's an implicit expectation to reach resolution quickly, to close the conversation and move to the next. That pressure, however well-intentioned, can cut conversations short before the customer is truly satisfied.
The AI agent has no such constraint. It can stay in the conversation for as long as the customer needs — answering follow-up questions, going deeper on a topic, exploring alternatives. Resolution isn't the finish line. Understanding is.
This also unlocks a knowledge depth that's simply unrealistic to expect from human agents at scale. We can't reasonably expect support agents to cite sleep science research or explain the biomechanics of spinal alignment during a high-volume shift. But an AI agent — built on verified, structured knowledge — can. At Emma, that means customers can have genuinely informed conversations about sleep health, not just product specifications.
Governance isn’t a constraint. It’s the foundation.
One of the biggest misconceptions about AI agents is that it should “figure things out on its own.” In reality, the opposite is true.
Successful AI agent implementations require structure, clarity and discipline, especially in customer experience. In CX, trust is non-negotiable. At Emma, governance was our starting point.
When we implemented our Sprinklr AI agent, we defined strict boundaries from day one:
- The AI agent cannot provide medical advice
- It cannot confirm pricing, discounts, or promotions
- It must rely exclusively on approved knowledge sources
- Product information must always align with verified data
These guardrails weren’t there to slow innovation. They allowed us to scale safely.
Early on, we learned how important that control was. In one example, an overly helpful response led customers to believe they would receive free products that weren’t actually included in an offer. The business impact was manageable, but the lesson was critical: an AI agent without governance erodes trust quickly.
Strong governance made it possible to iterate confidently, knowing that speed wouldn’t come at the expense of accuracy or brand integrity.
Designing for impact: Efficiency and conversion
The success of a digital CX strategy ultimately needs to show up in outcomes. In our case, the impact was clear.
Our Sprinklr AI agent now resolves 88% of inbound presales interactions end-to-end, across 11 markets. That translates to roughly 3,200 service hours per month handled by AI agents — freeing human agents to focus on complex, high-value cases.
What surprised many stakeholders was that customer satisfaction didn’t decline; it improved.
Conversations handled by the Sprinklr AI agent consistently deliver higher CSAT and customer effort scores than human agent-led interactions. Customers find the experience faster, clearer and easier to navigate.
Beyond efficiency, we also saw signals of commercial impact. Customers who engage with our AI agent demonstrate 12–16% conversion rates in certain markets — significantly higher than site averages. While conversion wasn’t our primary goal, it reinforced an important point: removing friction builds confidence, and confidence drives decisions.
Leading CX change from the middle
My role at Emma sits at the intersection of CX, technology and change management.
One of the most important parts of this work is alignment. New tools naturally create skepticism, especially when AI is involved. What changed perceptions most wasn’t presentations or documentation, but exposure.
Letting stakeholders test the system themselves. Involving teams in validating intents and content. Being transparent about limitations as well as possibilities.
I also bring a unique perspective from working on both sides of the table — as a CX practitioner and as someone who understands SaaS implementation realities. That dual view shapes how I design solutions, ask questions and measure success.
AI isn’t about replacing humans. It’s about creating a smarter service model that respects customers, supports agents and scales responsibly.
The future of CX is intentional
The future of customer experience isn’t just faster support. It’s thoughtful design at scale.
AI will continue to evolve rapidly. But the organizations that succeed won’t be the ones that adopt it fastest. They’ll be the ones that implement it most responsibly.
Digital-first CX starts with trust. Everything else builds from there.
