"Our Product Is Better" Is the Weakest Pitch in B2B

01 /The Concept

Lorikeet built the customer support AI that every other customer support AI should aspire to be.

The critical distinction: Lorikeet's AI agent doesn't just answer questions – it resolves complex issues end-to-end. That includes tracking down delayed shipments, replacing compromised payment cards, rerouting deliveries, and most other multi-step service tasks that normally require human judgment and system access.

To make it concrete: a customer contacts support claiming a credit card never arrived. Lorikeet checks the customer database, finds the address on file, and asks whether it's still current. The customer says they've moved. Lorikeet updates the address in the system – but recognizes the conversation can't end there, because the obvious next question is "when will I actually get the card?"

Lorikeet checks the customer's account tier, finds they qualify for express shipping, and offers to send a replacement card with a 2–3 day delivery window – a commitment it can make because it already queried the courier's documentation for delivery times to that city.

Four architectural decisions make this work.

Training requires no custom AI engineering: uploading the same internal support documentation used to onboard human agents is sufficient, and Lorikeet parses it to build its own branching decision logic. The agent can reach out to external data sources mid-conversation – including querying internal staff via corporate messaging when needed information isn't in its connected systems. Critically, it knows what it doesn't know: when it lacks the data or authority to resolve a case, it routes to a human rather than generating plausible-sounding nonsense. Every unresolved case is then collected, categorized by root cause, and surfaced to an administrator – closing documentation gaps and continuously improving the model.

Pricing ranges from $500/month to $24,000/year depending on request volume; enterprise clients with complex integration needs negotiate separately.

Lorikeet was founded in Australia in 2023, launched commercially in October last year, and raised $5M. After landing significant clients in healthcare, banking, and crypto, it has now closed a second round of $9M – less than four months after the first.

02 /Why It Matters

Customer support is the obvious sector for AI to transform through automation. The industry has been trying – which is why we now spend time cycling through support chatbots that can answer basic questions but can't resolve actual problems, hunting for a human who can.

Lorikeet's founders are direct about this: "few people are satisfied with AI bots that only serve up FAQ answers, or that are just wrappers passing customer requests to general-purpose AI models not designed for this class of problem."

That architectural critique is the point. Most first-generation support bots are built around conversation – output a reply to every input. Lorikeet is built around resolution – work backward from the customer's problem and determine what actions are required to actually solve it. Those are very different engineering problems.

The pricing structure reinforces this framing. Lorikeet charges differently for questions the AI resolves using documentation versus problems that require building a resolution plan. This creates a useful psychological effect: each month's invoice explicitly itemizes how many complex cases the AI handled – a running demonstration of value that makes the ROI comparison to human staff concrete and visible. The math typically comes out comfortably in Lorikeet's favor.

The company's positioning is also worth studying. Lorikeet never claims its agents are "better" than competitors' agents. "Better" keeps you in the same competitive category and invites customers to comparison-shop. Lorikeet says instead: "our clients use our AI agents to do things no other AI agent can do." That creates a separate category where there are no direct substitutes – you either want what Lorikeet offers or you don't.

The analogy is Apple's position in phones. Most buyers frame the decision as iPhone versus Android – and within Android, which one is better. The iPhone stands outside that comparison because it occupies its own mental category. That's the positioning Lorikeet is trying to build.

03 /Opportunities

For the past couple of years after ChatGPT's release, the AI market ran on euphoria. Builders assumed a smart wrapper and good prompts were sufficient to ship a great AI product. Buyers assumed AI could immediately replace human judgment. The result: first-generation AI bots that frustrate rather than help.

Now, as that initial wave recedes, the second generation is arriving – products that solve more real problems because they're architecturally different, not just because they run newer models. Lorikeet is a clear example.

The strategic window here is specific: being early enough to occupy a distinct mental category before the space crowds. Positioning doesn't live in the market – it lives in customers' heads. People are reluctant to revise strong impressions even when the underlying reality shifts.

That's the core of positioning theory: the goal is to own an association in the customer's mind between your product and a specific property. Competitors who later acquire the same capability will find it expensive and slow to dislodge that association.

The practical entry point: find a domain where first-generation AI products are already deployed and generating visible frustration – support bots that can't resolve, advisors that hallucinate, tools answering questions no one asked. Build the second-generation version around resolution rather than conversation. The positioning writes itself: not "better AI" but "AI that does what the current tools can't." In a market that has learned to be skeptical of AI claims, a specific, demonstrable capability beats every benchmark.