Skruvat.se uses Real Talk AI for customer service calls and automation

Customer service in e-commerce faces one of its biggest changes since e-commerce itself took off. AI phone agents now take over routine calls. They handle order status, returns, and delivery questions. They answer around the clock without queues. At Skruvat.se, handling times of up to one hour during peak season dropped to a few minutes.

From 60 minutes to a few

Customers still call, despite chat, email, and self-service via customer accounts. "Where is my order?", "Does this part fit my car?", "How do I return this?". The questions are often simple, but the volume is large. This peaks during campaigns and high seasons. Staffing a phone line to match web traffic is expensive. It is also hard to plan. The result hits many e-commerce retailers: queues grow to an hour or more when things heat up.

An extra dimension applies for a Nordic e-commerce retailer. Customers call in Swedish, Norwegian, Danish, and Finnish. Staffing a customer service team with the right language skills around the clock on every market quickly becomes unmanageable.

With an AI phone agent connected to the right systems, you can:

• Answer all calls within seconds, in the right language, even at 10 PM on a Sunday
• Provide order status and tracking information directly from the order system
• Handle returns, complaints, and payment questions according to your routines
• Search for the right spare part by license plate or manually
• Escalate complex fitment questions to the right specialist
• Collect structured data from every call — by market, by case type

MCP integrations: the agent works where you work

Access to your systems makes the difference between a "smart answering machine" and a real AI phone agent. You can connect the agent to dedicated MCP integrations. This gives it the same tools an employee has in their customer service interface:

• Order information — retrieves order details, status, line items
• Package tracking — gives the customer tracking links and real-time delivery updates
• Car model lookups — manual car lookups when license plates are missing
• And more.

A customer calls and asks "does this brake disc fit my 2012 Volvo V70?". The agent does not need to guess or forward the call. It looks up the car, checks compatibility, and answers. The same applies to order status. The agent fetches the freshest data from the system, not a cache.

Knowledge base: routines in place from day one

A knowledge base runs parallel to the integrations. The agent builds on this when it meets questions that are not pure system lookups: return routines, payment terms, damaged delivery policies, how to change an address on an active order. This setup rolls out new routines or updated terms directly to the agent without retraining. Changes in the knowledge base apply immediately on the next call.

Automations: every call is quality assured

Automations trigger after a call ends. They handle manual post-call work. You can, among other things:

• Quality assure calls — reviews calls where the agent solved the issue, logs what worked or not.
• Change or cancel orders — triggers a multi-step flow that executes the action in the order system when the customer requests a change or cancellation during the call.

This is where the real operational gain lies. The agent takes the call. It also executes the action and quality assures itself. Employees avoid re-registering. They avoid moving tickets between systems. Instead, they get a continuous feed of flagged calls worth a closer look.

The result

"At the Nordic region's largest e-commerce retailer for car parts, Real Talk's AI phone agents handle large volumes of questions about everything from orders and returns to payment questions. The result has been shorter wait times and a better customer experience!"

— Tor Rauden, CEO, Real Talk AI

Queues and the handling of incoming calls previously took up to an hour during peak season. They now finish in minutes. Employees work on tasks that require a human: complex fitment, escalated complaints, VIP customers. The agent takes the first line in four languages, 24/7.

What happens next?

AI phone agents improve with every call. The integration surface grows. Most major e-commerce retailers in the Nordic region will have an AI agent as part of their customer service team within a year. It will not act as a replacement for humans. It will serve as the first line connected to order systems, PIM, shipping integrations, and payment providers.

Are you ready to take the step? Book a demo.