Arbor needed AI that could solve hard problems — not just deflect
Arbor helps U.S. consumers in 13 deregulated states automatically lock in lower fixed electricity rates on a continuous basis. Their support team handles a steady stream of brand and energy questions — from decoding bills to explaining rate options — plus complex, state-specific inquiries tied to brokerage rules, utility breakdowns, and market conditions.
When Arbor started exploring AI, they weren’t looking to deflect — they wanted a partner with their shared support philosophy and the ability to understand and navigate the consumer energy landscape. “Most AI companies talk about deflection like it’s the goal,” said Jessica, Arbor’s Head of Customer Operations. But Jessica’s team wanted the opposite. They wanted to solve hard problems, especially when the customer stakes were high and emotionally-charged.
They considered Sierra and other well-known players, but most felt built for retail and e-commerce — not for navigating the often-complex world of retail energy supply and handling nuanced workflows. Lorikeet stood out by focusing on exactly what Arbor needed: AI agents that could handle nuance, de-escalate emotional moments, and follow SOPs without breaking a sweat.
Arbor was up and running by the end of the week
Adopting a new tool is the easy part — it’s integration and onboarding where things usually get stuck. With Lorikeet, Zendesk was connected in under 15 minutes, and custom API endpoints were deployed within days. But Arbor needed more. Their knowledge base wasn’t just state-specific — it spanned state, utility provider, and energy nuances. So the Lorikeet team built a bespoke Notion integration to sync that evolving, hyper-specific content into the system in real time.
Lorikeet agent treats Arbor customers as if they were their own
The real nailbiter? Putting your AI agent to the test. The Lorikeet team built out automated workflows like tagging tickets and logging CRM notes, alongside an FAQ system that blended Arbor’s dynamic knowledge base with real-time customer data. The agent was trained on dozens of utility-specific nuances across 13 states and rigorously tested to ensure it reflected Arbor’s voice, policies, and edge cases.
No endless loops. Only a clear handoff.
Lorikeet was configured with clear escalation logic for sensitive workflows like rate discrepancies and product bugs — where emotional tone and financial stakes make AI self-awareness and discernment critical. When confidence drops or policies require human judgment, the agent self-escalates cleanly, avoiding loops or vague replies — and routing tricky account issues to the right human at the right time.
“We believe deeply in the value we offer, and we don’t want customers to miss out just because they didn’t have the context. Sometimes, all it takes is the right conversation to help them stay."
Jessica Mishlove
Head of Customer Operations
Lorikeet performs at the same level as Arbor’s human agents
Customers praise “Olive” by name in Trustpilot reviews, often unaware they were chatting with an AI agent. What stands out isn’t just speed or tone — it’s nuance. Olive explains complex energy concepts, rate switching, and user-specific rate histories and savings in plain terms customers can understand. And when cancellation requests come in, Olive handles unique edge-case workflows with the same accuracy and empathy — turning support into a moment of trust and a reason to stick with Arbor.
