Meta recently commissioned Forrester Consulting to study how consumers experience generative AI.
One key finding stood out:
👉 “Consumers’ top challenges with genAI are receiving incomplete answers and an inability to answer their questions. Nearly half of the consumer survey respondents who have interacted with a genAI chatbot or automated chat complained that it was unable to complete their request fully, and one-third noted that it couldn’t answer their question and they had to be rerouted to a human.”
This is exactly the challenge we identified two years ago. That’s why we built a FEEDBACK LOOP into our AI agent framework. Every response is:
✅ Checked manually
✅ Refined if needed
✅ Added back into the knowledge base
The result? Accuracy improves rapidly — from an initial 40–50% correct rate to 95% within a couple of months. The remaining 5% are escalated to a human expert — much like a call centre agent escalating to a manager.
I’m proud that our software makes this process seamless, ensuring the AI learns and improves continuously. Customers get accurate answers, and businesses build trust.
Full Forrester report here: https://business.whatsapp.com/resources/resource-library/forrester-generative-ai-survey
Subscribe to our newsletter and keep up to date with useful AI news.