GemVet: A Proposal to Build AI Trust Through Useful Pet-Care Support By Fiona van Iersel Minerva Empowerment Marketing Communications 19 April 2026 Awareness is not the same as trust. A technology brand can be familiar to almost everyone and still struggle to persuade people that a new AI service will be useful, understandable and appropriately limited. A campaign addressing that problem needs more than a large audience. It needs an experience that gives the promise something concrete to stand on. GemVet is my proposed pet-care support concept for Google Gemini. It is a hypothetical strategic campaign, developed in April 2026, not a launched Google product, a clinical service or an announced partnership. The proposal uses a six-month UK pilot to ask whether useful, carefully bounded support could improve confidence in the specific service and, potentially, in Google AI more broadly. The idea is to help owners organise observations, prepare questions and understand information around a veterinary appointment. Its value would depend on appropriate veterinary input, clear limits and reliable routes to professional care. It should not claim to diagnose a pet, replace a veterinarian or justify delaying treatment. The communications problem is confidence AI adoption depends on more than technical capability. The Technology Acceptance Model places perceived usefulness and ease of use at the centre of adoption; later work broadens the account of how people decide to use technology (Davis, 1989; Venkatesh et al., 2012). In a sensitive context, trust and the handling of information become especially consequential. A worried pet owner may already search online before deciding what to do next. That creates an opportunity to make information easier to organise, but it also creates a responsibility not to exploit uncertainty. A warm campaign cannot compensate for vague boundaries or an experience that sounds more authoritative than it is. The strategic proposition is therefore a collaborative role: GemVet would help owners prepare for better conversations and act more confidently within clear limits. Veterinary professionals are an essential audience as well as potential advisers. If they regard the service as careless or as an attempt to bypass their judgement, borrowed consumer attention will not repair the problem. I chose the UK as a pilot setting because the concept can be explored through a concentrated ecosystem of pet retailers, veterinary events, welfare organisations and digital channels. That is a strategic rationale for testing, not evidence that any named organisation has agreed to participate or that veterinary AI is an uncontested market. A halo must be earned by the experience The proposal is a halo strategy: confidence earned in a specific use case might influence perceptions of the wider brand. That effect is a hypothesis to measure, not a transfer of trust that can simply be declared. Someone can find one service useful while remaining sceptical of other AI products. The creative platform is “Love means understanding,” with “Smarter Care for Every Paw” as the campaign title. The emotional insight is recognisable: owners wish their pets could explain what they are feeling. The campaign can speak to that feeling while making the actual service clear. It should not imply that an AI system can read an animal's mind. The desired response has three parts. People should understand what the tool does, feel reassured by its boundaries and find a practical reason to try it. Repeat use should come from usefulness rather than anxiety. A campaign that increases concern in order to generate more symptom searches would work against the trust the proposal is meant to build. Creative-brief element | Proposed direction Primary audience | UK pet owners seeking calm, understandable support around veterinary care Secondary audience | Veterinary professionals and credible pet-care voices Core need | Organise observations, understand information and prepare useful questions Proposition | A supportive information layer that works with professional care Tone | Warm, clear, intelligent and non-alarmist Required proof | Demonstrable usefulness, appropriate expert involvement, clear limits and understandable data handling Desired behaviour | Informed trial, useful repeat engagement and appropriate professional contact Any “reason to believe” must be built before it is advertised. Veterinary review, expert participation, data quality or product performance cannot be used as decorative campaign claims. If the product has not established those qualities, the communications should not imply that it has. An integrated team needs distinct responsibilities The original proposal envisaged a lead creative and strategy agency, a media and analytics partner, and a reputation and issues-management specialist. Ogilvy Health UK, EssenceMediacom UK and Burson were suggested as candidates for those respective roles. These are prospective choices for a planning exercise; they are not appointments, quotations or endorsements. The important point is the division of work. Creative must make the proposition understandable. Media must reach people in relevant contexts without escalating fear. Analytics must distinguish exposure from incremental effect. Reputation management must prepare for criticism, confusion and genuine product concerns, with a route for those concerns to change the service. Veterinary and welfare expertise should inform the concept throughout. It is insufficient to invite an expert to appear in an advertisement after the product and claims have been fixed. Integration means that the experience, evidence, creative and measurement work toward the same objective. Build a network of useful encounters The collaboration architecture has four layers: retail reach, rescue and adoption contexts, professional discussion, and explanation through creators and podcasts. Each has a different job. A long list of logos would not create trust if the customer could not explain what any partner contributes. Pet retailers could provide demonstration and discovery opportunities. In the proposal, Pets at Home and Jollyes were examples of prospective retail partners. The useful activation would be a clear introduction to the service and its limits, with a practical route to try it. Data sharing or use of retailer customer lists would require an agreed, appropriate basis rather than being assumed within the media plan. Rescue and welfare organisations such as Dogs Trust, Cats Protection, Blue Cross and Battersea were considered as possible collaborators because owners already seek guidance in those contexts. Their role would have to serve a real welfare purpose and respect their independence. A charitable association should not be presented as blanket clinical approval of an AI service. Professional events and awards could host demonstrations, discussion and questions. Sponsorship buys an opportunity to participate, not evidence of safety or efficacy. Consumer creators can make the proposition understandable, while profession-facing podcasts can examine its limitations in more depth. The plan should include cats and other relevant owner experiences rather than allowing a visually appealing dog campaign to define the entire audience. Collaboration layer | Proposed contribution | Trust requirement Pet retail | Discovery, demonstrations and practical trial | Clear explanation and appropriate handling of customer information Rescue and welfare | Useful owner education and context | Independent judgement and a real benefit to the audience Veterinary events and professional media | Scrutiny, discussion and qualified input | Sponsorship must not be confused with endorsement Podcasts and creators | Plain-language explanation and familiar stories | Transparent commercial relationships and accurate claims A six-month pilot with clear phases The original planning window was June to November 2026. It is retained here as the proposed schedule, not a statement that the pilot took place. Event references provide a possible planning framework; participation, dates, packages and permissions would need confirmation before any real campaign booking. Proposed phase | Main work | Decision supported Before launch | Test the concept, claims, product boundaries and creative; establish baseline measures | Is the proposition understandable and ready to test? June | Initial professional discussion and a limited launch burst | Do audiences understand the intended supporting role? July–August | Retail and owner education, creator explanations and controlled trial | Which encounters produce useful activation rather than curiosity alone? September–October | Refine creative, improve onboarding and evaluate repeat use | Is the experience supporting trust and practical value? November | Final pilot evaluation and professional feedback | Is there evidence to expand, revise or stop? After paid activity | Measure persistence and decay | Do effects remain when media pressure reduces? Geo-targeted activity around relevant events and participating retail catchments could connect media with a place to learn more. The original plan included £300,000 for selective digital out-of-home activity within the wider budget. That amount is a planning provision, not a supplier quote, and should be judged against what it adds beyond search, video and partner channels. The sequencing matters. Creative testing should happen before large-scale media commitments. Early feedback should have time to change the experience. A pilot that launches every channel at once can generate a large number of contacts while revealing little about why any of them worked. The budget is a scenario, not a disclosed Google allocation The full six-month pilot is modelled at £12.745 million. The proposal uses Alphabet's scale as context for asking whether a substantial test could be considered, but group financial reporting does not disclose a GemVet budget or a standalone Gemini marketing allocation. Apportioning a broad corporate expense line would remain an assumption, not evidence of a product's actual spending. For that reason, the campaign should be justified by its own incremental outcomes. A large company can afford an ineffective campaign; affordability is not proof of value. The model needs a clear relationship between the costs incurred, the users acquired, the trust gained and the margin those users could generate over time. Illustrative budget category | Amount Core creative and paid media | £8,220,000 PR, measurement and agency | £1,400,000 Retail collaborations | £835,000 Rescue and adoption collaborations | £385,000 Awards, events and geographically targeted media | £730,000 Podcasts, influencers and creators | £1,175,000 Total proposed pilot | £12,745,000 Planning measure | Illustrative value or treatment Total pilot cost | £12,745,000 Paid-user target range in the source model | 170,000–205,000 incremental paid users All-in cost per incremental paid user at 170,000 | Approximately £74.97 All-in cost per incremental paid user at 205,000 | Approximately £62.17 Selective digital out-of-home provision | £300,000 within the proposed total Status of all figures | Author's planning assumptions; not supplier quotes, approved spend or achieved results The label “paid CAC” needs care here. Dividing the entire pilot cost by incremental paid users gives an all-in campaign acquisition cost. A narrower paid-media CAC would use only the relevant paid acquisition spend. Both can be useful, but they answer different questions and should not share a label accidentally. Free-user activation should be measured separately. It may contribute to a future paid-user base or a wider trust objective, but free registrations do not automatically create equivalent economic value. The business would need evidence about activation quality, repeat use, conversion, service costs and retention before treating the funnel as commercially attractive. Nor is annual subscription revenue the same as customer value. A break-even model should account for contribution margin, retention, cancellations, discounts and the timing of revenue. The required incremental paid users equal the attributable campaign cost divided by the expected contribution per incremental user over a defined period. Without those inputs, a precise return-on-investment claim would be premature. Measure incrementality, not merely activity The strongest part of the proposal is the commitment to an exposed-versus-control design. The original plan suggests a 96% exposed population and 4% holdout, with assignment at user or postcode-cluster level depending on what can be implemented. That allocation is a candidate design. It should be supported by power calculations and practical checks, not assumed to be sufficient because the percentages look tidy. The control group should be comparable and measured in the same way. Where feasible, it would be excluded from the specific paid and nurture interventions being evaluated. Organic exposure, national publicity and sharing between households can still create contamination. Those limitations need to be considered when interpreting any difference. The core calculation is straightforward: compare the relevant conversion rates in the exposed and control groups, estimate the incremental effect for the defined eligible population and apply the agreed economic value. The difficult work is ensuring that the comparison is credible and that attribution does not count users who would have converted anyway. Outcome family | Examples | What it helps establish Understanding and trust | Perceived role, usefulness, ease of use and confidence in data handling | Whether the campaign improves comprehension and justified confidence Activation | Completed onboarding and meaningful first use | Whether interest becomes a usable experience Repeat use | Return visits tied to appropriate tasks | Whether the service remains useful after the first encounter Commercial behaviour | Incremental free-to-paid conversion and retained contribution | Whether the campaign creates economic value Partner and media response | Relevant QR interactions, enquiries and qualified traffic | Which encounters contribute to the journey Persistence and quality | Post-campaign effects, complaints and understanding of limits | Whether gains endure without undermining the promise Creative and brand-testing services such as Zappi, Kantar or YouGov were considered as possible evaluation partners in the original proposal. Their use would depend on a specific scope and commercial agreement. Testing should cover claims and comprehension as well as recall: an advertisement remembered for implying that AI can replace a veterinarian would fail this strategy even if its awareness score were impressive. Scale only after the trust proposition survives the test Rollout should depend on pre-agreed criteria for usefulness, understanding, trust and economics. A campaign can meet one objective and miss another. High sign-up volume with poor comprehension would require a different response from strong comprehension with weak conversion. A pilot should preserve the option to improve the concept or stop it. International expansion would then need local work. Pet-care habits, access to veterinary services, trusted institutions, media costs and the handling of information differ across markets. The transferable asset is the proposition of useful, bounded support. The partners, language and customer journey should be adapted to the setting. GemVet's strategic value lies in making an abstract claim about AI trust concrete enough to evaluate. The halo would be earned if people understood the role, received useful support and found the limits credible. Communications can introduce that experience and help people navigate it. The experience itself has to deserve the confidence the campaign asks for. References Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13(3), 319–340. Venkatesh, V., Thong, J. Y. L., & Xu, X. (2012). Consumer acceptance and use of information technology: Extending the unified theory of acceptance and use of technology. MIS Quarterly, 36(1), 157–178. Kliatchko, J. (2008). Revisiting the IMC construct. International Journal of Advertising, 27(1), 133–160. Holliman, G., & Rowley, J. (2014). Business to business digital content marketing. Journal of Research in Interactive Marketing, 8(4), 269–293. Schmitt, B. (1999). Experiential marketing. Journal of Marketing Management, 15(1–3), 53–67. Alphabet Inc. (2025). Annual report 2024 (Form 10-K). https://s206.q4cdn.com/479360582/files/doc_financials/2024/q4/goog-10-k-2024.pdf Alphabet Inc. (2026). Annual report 2025 (Form 10-K). https://s206.q4cdn.com/479360582/files/doc_financials/2025/q4/GOOG-10-K-2025.pdf EssenceMediacom. (2023). Google & EssenceMediacom: An enduring relationship. https://www.essencemediacom.com/news/google-and-essencemediacom-an-enduring-relationship Ogilvy UK. Health capabilities. https://www.ogilvy.com/uk/capabilities/health Pets at Home. Pets Club. https://www.petsathome.com/support/pets-club Jollyes. https://www.jollyes.co.uk/ Dogs Trust. https://www.dogstrust.org.uk/ Cats Protection. https://www.cats.org.uk/ Blue Cross. https://www.bluecross.org.uk/ Battersea. https://www.battersea.org.uk/