Circles builds AI telco concierge

Telecom customer service is usually a mess of apps, scripts, queues, and handoffs. You ask a simple question, but the system treats billing, support, upgrades, and retention like separate worlds. Circles is trying to collapse that sprawl into one AI layer. On April 30, the company said it launched an AI concierge built on OpenAI’s API platform as the first big milestone in a broader push to build what it calls an AI-native telco stack. (prnewswire.com) ### What did Circles actually launch? Not just a nicer chatbot. The pitch is an AI concierge that sits across the customer journey — answering questions, routing requests, personalizing offers, and escalating to humans when needed. (prnewswire.com)ons and commercial decisions in one loop. (circles.co) ### Why is that different from normal telco AI? Most telecom AI still works like a bolt-on. One tool handles support deflection. Another does marketing. Another scores churn risk. Humans glue the pieces together. Circles is going after the opposite model — AI as the operating layer, where the same (circles.co)or sensitive. Basically, it is less “assistant on the side” and more “front door plus traffic controller.” (prnewswire.com) ### What numbers make this worth noticing? The company attached real performance claims, which is why this announcement matters more than generic AI branding. Circles says the multi-agent setup resolves 85% of customer queries and d(prnewswire.com)product is for: not just cutting support costs, but also selling more intelligently. (circles.co) ### Where does this go first? Circles has said the broader AI-native platform will launch in Singapore first through Circles.Life’s app, then expand to its operator partners globally. That matters because Circles is not only a consumer brand — it also sells software and infrastructure to other digi(circles.co)rs will buy. (circles.co) ### Why does OpenAI matter here? OpenAI is not being framed as a branding partner. Circles says the collaboration is multi-year and gives it access to OpenAI’s API platform, plus earlier-stage models and research programs. The point is speed. Telcos are slow-moving businesses with ugly legacy systems, and Circ(circles.co)lows, data structure, and operator integrations. (circles.co) ### What is the catch? The hard part is not answering questions in natural language. The hard part is trust, permissions, and edge cases. A telco concierge that can upsell, change plans, route complaints, and decide when to escalate is touching billing, identity, compliance, and retention all at once. If the m(circles.co)ailed handoff, or a customer who churns. That is why the escalation layer is load-bearing here. (circles.co) ### Why does this matter beyond telecom? Because service industries keep talking about AI assistants, but the bigger shift is AI as workflow infrastructure. Telcos are a good stress test — high volume, repetitive requests, lots of customer data, and expensive human support. If Circles can make one (circles.co)ybook. (mobileworldlive.com) ### Bottom line? This launch is really a bet about where enterprise AI is going next. Not standalone bots. Not search boxes with better prose. A layer that owns the interaction, decides the path, and tries to turn support from a cost center into a revenue engine. Circles is early, and the claims still need real-world proving. But the shape of the idea is clear now. (prnewswire.com)