Decagon

Meet Duet Apprentice: AI that learns your business the same way your best people do. Your team's knowledge is spread across onboarding docs, Slack threads, and the judgment of reps who've done the job for years. Apprentice gives Duet access to all of it: 🔹 Docs: Reads your onboarding material and process wikis in Notion, Confluence, SharePoint, or Google Drive, with nothing to migrate or reformat 🔹 Escalations: Learns how your most experienced reps handle edge cases no AOP anticipated 🔹 Slack and Teams: Keeps up with decisions as they're made, and asks in-thread when it isn't sure Frontier models keep getting better at reasoning, but none of that teaches an agent how your team actually operates. That knowledge already exists inside your company, and Apprentice is how Duet reaches it. Full details are in the comments.

Today, we’re open-sourcing the Personal Agent Consent & Trust Protocol (PACT), co-developed and supported by Instinct. We’re also excited to join the Personal Agent Protocol working group announced by the team behind Meta’s Muse to help shape an industry standard. As personal agents book travel, make purchases, and resolve issues for their users, businesses need a way to verify whom those agents represent and what they've been authorized to do. PACT gives these interactions explicit, verifiable customer permissions and is built on four principles: 1. Build on existing standards for agent communication and customer authorization 2. Verify which agent is calling separately from what the customer has authorized 3. Let each business define its own permissions 4. Keep customer login with the business Excited to share what we’ve been building. Read the announcement and explore the spec in the comments.

Duet Autopilot can't ship a change on its own. Every update is tested against real conversations and lands in a versioned workspace with the exact diff. A human approves it or it never reaches production.

The agent you launch with is just the baseline. Duet Autopilot improves it with every conversation, turning production experience into logic you can ship. Each change is validated against hundreds of real conversations first.

Stop making your customers repeat themselves. The biggest frustration in customer service is having to explain your problem three different times to three different agents. With the launch of User Memory, we are ending the cycle of customers having to repeat themselves. AI agents can now maintain context across separate conversations so they never have to start over. Whether it's a change request made last month or a signal from yesterday, Decagon ensures every interaction picks up exactly where it left off. For brands like Away, this provides the continuity to deliver a great customer experience and build long-lasting loyalty.

We’re proud to announce a strategic partnership with Kaluza. Energy retailers worldwide are looking to deliver best-in-class customer experiences, and Kaluza and Decagon are committed to making that a reality. Behind every smart meter, EV charger and flexible energy tariff is intelligent software making it all work. Kaluza has spent years building this infrastructure, helping utilities across the UK, Australia, Europe and Japan deliver smarter, cleaner energy services to millions of customers. Kaluza selected Decagon to power AI voice and chat within Kaluza Customer Care, its next-generation customer service offering. We’re proud to help bring an AI concierge experience to the utilities building the energy system of the future.