--- title: "How Starshipit Is Putting One AI Agent on the Whole Customer Journey" publishedAt: "2026-08-10" summary: "Starshipit is replacing Intercom's Fin with Aimdoc inside its app. Within five weeks of rollout, nearly a quarter of new users got onboarding help from the AI — most within 15 minutes of signing up — and engaged users were 2.6× more likely to stay active the next month." author: "Ben Skoog" authorImage: "/ben_skoog.jpeg" --- - **Industry**: Shipping & fulfilment (eCommerce) - **Use Case**: Website engagement + in-app onboarding - **Customer**: [Starshipit](https://starshipit.com/) - **Spokesperson**: George Plummer, Founder Starshipit is replacing Intercom's Fin with Aimdoc inside its app, and within five weeks of rollout, nearly a quarter of new users were getting onboarding help from the AI — most within 15 minutes of signing up. ## About Starshipit Starshipit is the highest-rated shipping and fulfilment automation platform for retailers in Australia and New Zealand, serving 30,000+ retailers. It connects e-commerce stores to 60+ couriers — Australia Post, DHL, FedEx, Aramex, and more — automating everything from label printing and address correction to customs documentation and live checkout rates. That power comes with an inherently technical setup. Before a new retailer prints their first label, they need courier accounts connected, printers configured, and their store integrated. For Starshipit, activation has a precise definition: the first label printed. Every customer's path there is different. ## The Challenge Starshipit's support and onboarding teams serve tens of thousands of retailers with real, technical questions: How do I set up my printing? How do I configure checkout rates? Why is my webhook returning a 422? Their existing AI support tool, Intercom's Fin, answered some of this — at a per-resolution price that grew with every conversation. Meanwhile, their website had an untapped problem of its own. ## Chapter 1: The Website Agent Starshipit deployed [Aimdoc Website](/agent/website) on starshipit.com in January 2026. Over six months, the agent: - Held **3,000+ conversations** with ~3,000 unique visitors across Australia, New Zealand, the US, the UK, and beyond — answering 6,400+ questions about couriers, integrations, pricing, and setup - Handled **43% of those conversations outside business hours**, when no team was at a desk - Qualified prospects automatically — classifying visitor type (prospect, customer, trial) and shipment volume in real time — and synced every conversation to HubSpot with full attribution While lead capture was running, the agent produced ~**23 sales-qualified leads per month**, every one arriving with a name, email, company, job title, and shipment-volume qualification: no forms, no SDR chasing. But the conversation data surfaced something more interesting: roughly one in five website conversations wasn't a prospect at all. It was an existing customer or trial user asking product questions. The website was a support channel in disguise — and asking those people for contact information Starshipit already had made no sense. So instead of treating customers like leads, Starshipit took the agent to where its customers actually were: inside the product. ## Chapter 2: The In-App Agent > "The biggest thing we're after is onboarding. A customer signs up, and in their first few visits we want to prompt them through the tailored steps that get them started." > > — George Plummer, Founder, Starshipit In June 2026, Starshipit began replacing Intercom's Fin with [Aimdoc Application](/agent/in-app) inside its new app UI — the same agent from the website, now running in app mode. Integration took days, not months: Starshipit's app calls `aimdoc.identify()` when a user signs in, so the agent knows exactly who it's talking to (name, company, account) with no forms and no "What's your email?" dance. > "From what I've seen, it's already a better experience." > > — George Plummer, at rollout ### What happened in the first five weeks The rollout covered a pilot slice of Starshipit's base: new users landing on the new UI. Among those users: - **23%** engaged the AI assistant, unprompted, with no proactive nudges yet - The median user had their first AI conversation **15 minutes after signing up** — help arrives during the first session, exactly when onboarding friction hits - Conversations run nearly twice as deep as website chat (**3.6 messages per conversation** vs 2.0), and **41%** are multi-turn working sessions: printer setup, courier configuration, webhook debugging, carrier integrations - Usage compounds: weekly conversation volume grew roughly **60%** across the pilot while the flow of new users stayed flat — adoption deepening, not just more traffic - **1 in 6** engaged users came back for two or more separate conversations; one user had twelve ### The number that matters most Starshipit's activation metric is the first label printed, and retention follows activation. Comparing the June cohort of new users: **Users who engaged the AI during their first month were active the following month at 67%, versus 26% for those who didn't — a 2.6× difference.** Engaged users are, of course, a motivated group. But this is precisely the moment Starshipit wants to own: a new customer, in their first session, working through the setup that determines whether they ever ship a label. ## The Full-Agent Advantage Most teams buy a marketing chatbot for the website and a support bot for the app, and the two never speak. Starshipit runs one agent with one memory across the entire journey: 1. On the website, it qualifies a prospect by shipment volume and syncs them to HubSpot 2. Fifteen minutes after signup, the same agent greets them inside the app — knowing who they are via a cryptographically verified identity, with no re-introduction 3. It reads the context Starshipit chooses to share (account, plan, CRM properties) and helps with the actual work: printing, couriers, integrations ## The ROI Picture Across six months, the agent absorbed roughly **500–600 hours** of sales- and support-equivalent workload: website conversations at a conservative 10-minute handle time, in-app technical sessions at 15. Much of that total comes from the website, where the agent has run since January. But the run-rates tell the real story: - The website agent handles a steady **~15–19 hours per week** of combined sales and support conversations - The in-app agent reached **~15 hours per week** within its first five weeks, from a pilot slice covering only new users on the new UI That pilot cohort is a small fraction of Starshipit's 30,000+ retailer base. Holding today's engagement rate constant (before proactive onboarding prompts ship), a full rollout projects to more than **10×** the current in-app workload: **150+ hours per week** — the equivalent of 3–4 full-time support and onboarding roles — with around-the-clock coverage. And unlike per-resolution pricing, that capacity doesn't get more expensive as it gets more useful. ## What's Next Starshipit is migrating its full customer base to the new UI — and the agent — by the end of August. Together, Starshipit and Aimdoc are building the next layer of measurement: trackable onboarding objectives the AI works through with each user (for Starshipit, the path to that first printed label), and a cohort analysis pairing agent engagement with conversion data to measure the effect on trial-to-paid rates directly. ## The Takeaway Starshipit didn't buy a website chatbot and a support bot. It's putting one AI agent on the whole customer journey — from first website visit to first shipping label — and measuring what matters: time-to-help, activation-adjacent engagement, and next-month activity. > "From what I've seen, it's already a better experience." > > — George Plummer, Founder, Starshipit --- Want to put one AI agent on your whole customer journey? Explore [Aimdoc Website](/agent/website), [Aimdoc Application](/agent/in-app), or book a demo.