It's not often you can say you were a startup's very first customer, but that's the reality for Eucalyptus and Lorikeet.
We've been flying with Lorikeet since Day 1, and I’ve had the privilege to learn from this incredible team as they’ve scaled.
As we've integrated Lorikeet into our support processes, I’ve also been learning a lot about what it takes to make generative AI work in service delivery. Here are some of the key (and always evolving) lessons:
1. Build systems for your AI agents, just like your human team
The Bot isn’t just a tool—it’s part of your team. At Euc, we’ve treated Lorikeet just like a human team member. We set up feedback loops, quality assurance systems, and even performance ratings for the bot. From my experience, AI learns best if it gets the same structured guidance that helps humans succeed.
2. Hold AI to the same standards—and then aim higher
The benchmarks should be set based on human performance metrics, and then the goal should be to exceed those. Whether it's response time, accuracy, or customer satisfaction, your bot should be able to match human team members out of the gate and, over time, surpass them.
3. Contain the risk by starting small
Starting with low-risk workflows is key. By limiting initial automation to straightforward tasks, you build confidence in the system while reducing the risk of failure. As we’ve expanded Lorikeet’s scope, this approach allowed us to scale without compromising patient trust.
4. Create new roles to drive automation forward
Invest in your people early. I realised quickly that some of our top customer support team members were invaluable in managing the automated workflows. We created dedicated roles for them to build workflows, prompt the bot, audit its responses, and provide feedback to improve its logic. It’s a short-term capacity trade-off for long-term automation gains, but it's one of the most impactful decisions we made.
We're excited to announce our public launch, exiting stealth mode and announcing $5m USD in seed funding from Square Peg Capital and a wonderful cast of angel investors including Claire Hughes Johnson, Cristina Cordova, Bob van Winden, Cosmin Nicolaescu, Tara Seshan, Kalyani Iyer, Kristin Thayer, Jamie Cuffe, Snir Kodesh, Kenneth Auchenberg 🛠, Jeff Weinstein, Keith Walewski, CFA and many others.
To drive real value over time, customer experience AI agents need to be able to solve the most complex cases. We pride ourselves in being a team that goes after the hardest problems first, and we've been privileged to work with great teams at companies like Eucalyptus, Step, Remote and StashAway to push the boundaries of what AI agents can do to make life better for customers.
We're all in the first innings of one of the most exciting moments in software, and we're raring to go after what comes next. If you're interested in reading more, Fortune covered our story (link in comments) today.
סוכן ביטוח ומתכנן פנסיוני דואג ללקוחות שלי לטוב ביותר עוסק בכל סוגי הביטוחים פנסיות ומומחה לפיננסים
1moIf you have a good team you can succeed in anything and Decart team are the best one by one