Kanopi

Energy retailers and solar installation platforms are sitting on a protection opportunity that most have not identified yet. A customer installs a $15,000 solar system. The installer provides a product warranty. The inverter manufacturer provides a separate warranty. Neither covers storm damage, electrical surge, or the full replacement cost if the system is destroyed by hail. The customer has made a significant financial commitment to a piece of infrastructure attached to their roof. The installation platform has the system specifications, the installation date, the property address, and the customer's contact details. Every data point needed to generate a protection offer already exists inside the platform's records. And the trigger point, the moment the installation is confirmed and the system goes live, is the highest-intent moment to present it.

Most insurance product disclosure statements were written for humans to read. Dense. Legal. PDF format. Dozens of pages of terms, conditions, exclusions, and definitions. AI agents do not read PDFs. They parse structured data. When an agent evaluates insurance products on behalf of a consumer, it needs machine-readable product information: coverage parameters, eligibility criteria, pricing logic, and exclusion rules in structured formats that can be compared programmatically. The carriers whose product data exists only in PDF form are invisible to this channel. Not because the product is wrong, but because the format makes it inaccessible to the systems that are increasingly doing the shopping. Restructuring product data for machine readability is a distribution decision. The carriers that make their products easy for AI agents to read, evaluate, and transact with are the ones those agents will recommend.

First-mover advantage in embedded protection is structural, not temporary. And the reason is renewals. When a customer attaches protection through a platform, the policy renews annually. The switching cost is not financial. It is behavioural. The protection is already embedded in the platform the customer uses. Renewing requires no action. Switching requires finding an alternative, re-entering data, and completing a new checkout elsewhere. A competitor entering the same vertical six months later is not competing against technology. They are competing against embedded behaviour. The customer already has protection. It already renews automatically. There is no trigger point to capture because the incumbent platform already captured it. This is the same dynamic that plays out in payments and lending on platforms. The first platform to embed a financial layer sets the customer expectation. Followers compete for the diminishing share that remains. The compounding effect is worth modelling. More policies generate better data. Better data enables better risk segmentation. Better segmentation earns better pricing from carriers. Better pricing lifts conversion. Higher conversion increases volume. The flywheel favours the platform that started first. For platform operators evaluating timing, the relevant calculation is not the revenue generated by moving now. It is the revenue forfeited by waiting while a peer platform captures the position.

Cyber risk is evolving faster than most insurance products can keep up with, and agentic AI is about to accelerate that gap significantly. When AI agents begin transacting on behalf of consumers across multiple platforms simultaneously, the attack surface for identity fraud, social engineering, and data interception expands in ways current cyber products were not designed to cover. An agent making purchasing decisions across insurance, banking, and retail creates new vectors that sit between traditional personal cyber policies and commercial cyber coverage. The carriers and underwriters thinking about this now are asking a genuinely difficult question: how do you underwrite a risk that changes shape every time the AI agent's capabilities are updated? The answer is probably not a static annual policy. It is more likely a dynamic product that adjusts coverage based on the agent's activity profile, updated continuously through the same data infrastructure that powers the distribution. The insurers with the architecture to support that kind of product flexibility will be first to market. The ones that need 6 months to change a field in their PAS will be watching.

80% of insurers are now experimenting with generative AI, according to the latest Earnix research, and that number reflects genuine progress across the industry. The challenge most carriers are hitting is not whether AI works in insurance. It clearly does, particularly in claims processing, underwriting support, and pricing. The challenge is moving from a successful pilot in one function to governed production deployment across distribution, where the regulatory and audit requirements are fundamentally different. Efficient agentic systems create auditable records of the entire decision chain: which product was configured, what pricing logic applied, what the customer saw, and what alternatives existed. That is a different infrastructure requirement than most pilot programmes were designed to test. The carriers making this transition well are the ones treating it as a structural change to their business models, not a feature rollout. The AI is the easy part. The governance infrastructure around it is where the real work lives.

60.1% of consumers would add insurance to an online purchase if it were offered at the point of checkout. That number comes from a retail product insurance study and it challenges a fundamental assumption most platform operators hold: that their customers are not interested in protection. Customers are not rejecting protection. They have never been offered it in a context where the relevance is obvious. When the offer appears at the moment of transaction, alongside the product they are buying, with data pre-filled from the purchase, conversion rates are 10x higher than standalone insurance models achieve. The gap is not demand. It is distribution. Traditional insurance is sold through dedicated channels: comparison sites, broker networks, direct campaigns. The customer has to actively seek it out, re-enter data the platform already holds, and complete a separate transaction with a different provider. Embedded distribution removes every one of those friction points. The offer is contextual. The data is pre-populated. The checkout is native. The customer does not leave the platform. For platform operators running conversion rate optimisation on their core product, embedded protection is the equivalent of discovering a high-intent segment they were not serving. The demand was always there. The distribution infrastructure was not.

Configure once. Distribute everywhere. The carriers that understood this principle three years ago are now operating at a scale their competitors cannot replicate by building faster. That is the commercial reality of orchestration infrastructure playing out across Australian insurance right now. The carriers who connected to a platform early are adding distribution partners in weeks at near-zero marginal cost, while the carriers who started custom builds more recently are still working through their first or second integration with resource allocation conversations ahead for the third. The product is identical. The pricing is identical. The only variable is the architecture underneath, and it turns out that variable compounds over time in ways that most insurer finance teams have not modelled yet. PwC projects embedded insurance in Australia will grow at 34% annually through 2033. The carriers positioned to capture that growth are the ones whose distribution economics improve with every connection, not the ones whose costs scale with every partnership.

A global device protection platform scoped what it would take to build insurance distribution infrastructure in-house. The estimate: 3-4 dedicated people, 4-6 months of development, and no reusable IP at the end. That was for one market. One carrier. One product line. Scaling to a second geography would mean a second build. A different carrier with a different API. Different regulatory requirements. Different payment rails. Different document standards. The same engineering effort repeated, with no leverage from the first. They chose orchestration instead. One integration. Carrier complexity managed behind the API. Device and cyber protection live in Australia with one insurer and in the UK with another, both managed from a single platform connection. Every new geography and product line added revenue without adding integration effort. The insight is not that building is hard. It is that building does not compound. Every additional market requires the same marginal effort as the first. Orchestration compounds because each new carrier, product, and geography is additive to the existing connection. For platforms operating across multiple markets, or planning to, the economics of orchestration versus direct integration diverge more with every market added.

A question all Tier 1 insurers should be asking right now - What adjacent product categories are we leaving on the table because the coordination infrastructure is not there yet? Major insurers like Zurich and Bupa have already started exploring adjacent product lines through partnerships with providers, coordinated through flexible orchestration layers. The pattern is worth watching because it shows up increasingly across health, travel, device, and cyber as well. Carriers are choosing coordination over construction, and the ones doing it well are reaching new markets without adding headcount or rebuilding systems.

For a property management platform, the deposit sits at the centre of a web of manual work: collecting it, holding it, verifying certificates of insurance against it, arbitrating disputes over it, and refunding it. None of it shows up as a line item, so it reads as the cost of doing business. It is not the cost of doing business. It is a process that was never built to scale. Here's the tell: a resourcing problem gets solved by hiring. A structural problem gets worse as you grow, because the work is generated by volume, not effort. Certificate chasing never ends because every new tenancy adds one and every lapse creates an exception a human has to catch. The fix is not a faster chase. It's removing the artefact entirely. When cover is arranged inside the tenancy, there is no separate certificate to collect, verify, or reconcile against a deposit. The verification loop does not get faster. It stops existing. And it doesn't mean becoming an insurer. The platform owns the tenancy moment, the insurer owns the risk, and an orchestration layer connects them. The workload disappears without a migration project attached. 👇