Building Your Own Customer Service Superstar

Klarna launched an AI chatbot and it's crushing their customer service workload. Two-thirds of chats handled automatically in the first month? That's awesome, because you can get similar results for your business too!

According to various sources cited below, the customer service agents handled work equivalent to that of 700 full-time human agents, or two-thirds of all customer support chats. Resolving tasks in fewer than 2 minutes instead of the 11 minutes necessary for human agents, the AI agents achieved customer satisfaction scores “on par” with those of human agents.

According to their estimates, Klarna’s 700 human agents supported 2.3 million conversations at a cost of $3.6 million per month, or $1.58 per conversation. Addressing the same number at 2 minutes per chat, an OpenAI-powered agent would cost no more than ~$9,000 per month, or $0.004 per conversation, based on the current pricing structure of GPT-4. In other words, by switching from human agents to AI agents, Klarna could save nearly 90 to 100% in labor costs.

For all that we know, Klarna and other consumer-facing companies will probably retain human-in-the-loop systems to resolve customer-support cases that AI cannot, AI-related efficiency gains are likely to reduce the number of humans in customer support positions significantly.

All language-centric knowledge work associated with enterprise sales, technical support, and customer success will face significant?disruption.

https://www.klarna.com/international/press/klarna-ai-assistant-handles-two-thirds-of-customer-service-chats-in-its-first-month/

https://www.warpnews.org/artificial-intelligence/klarnas-new-ai-assistant-does-the-same-job-as-700-people-after-one-month/

What Makes Klarna's AI Assistant So Effective?

• Self-Service Power: Customers get answers right away to basic questions like tracking orders or payment details. This frees up human agents for the trickier stuff.
• Multilingual Magician: AI breaks down language barriers, meaning you can serve customers worldwide without a massive support team.
• Smarter Over Time: They are willing to bet that those 2.3 million chats it handled helped train the AI to get even better, reducing the need for customers to repeat themselves.

Let's Build Something Similar (Simplified Version)

1. The Data Foundation: You'll need a LOT of past customer conversations. Anonymize it to protect privacy, but this is what you'll use to train your AI model.
2. Choose Your Model: There are pre-built chatbot models out there that can give you a head start, especially if you're new to AI.
3. Teach Your Bot: This is where the magic happens. Feed your customer data into your model. Tools often have visual interfaces for this, so you don't need to be a hardcore coder.
4. Test, Tweak, Repeat: Launch a basic version, monitor how it performs, and see where it stumbles. Use those mistakes to improve your chatbot!

Pro Tips

• Start Small: Don't aim to replace all customer service at once. Pick common, easy questions your bot can confidently answer.
• Transparency Rocks: Let customers know they're chatting with AI. Option to reach a human agent is crucial!

https://www.inc.com/ben-sherry/klarna-says-its-new-ai-assistant-is-outperforming-customer-service-representatives.html

Disclaimer: All opinions are my own and not those of my employer.

Follow me for more insights on LinkedIn: https://lnkd.in/eJ5gubCg ??

#AI #ArtificialIntelligence #chatbots #customerservice #customerexperience