From Signed to Live: Building VanCityVet with AI-Assisted Workflows

The Brief: A Brand New Practice, Starting from Zero

Van City Veterinary Hospital signed with us on March 12, 2026. Dr. Sran was opening a brand new practice at a new location — not a rebrand, not a refresh, but a clinic that did not exist yet. No website, no social media accounts, no content library, no digital presence of any kind. The goal was to build all of it, and to move fast enough that the digital presence would be ready when the doors opened.

The scope was the full Swift AI Package: website design and build, social media setup and management, a content and blogging program, monthly email newsletters, short-form video, and ongoing Google environment management across Business Profile, Search Console, and Analytics. That is a lot of deliverables to run in parallel. Without AI-assisted workflows, it would have been a three-to-six month engagement to get everything established. We needed to move significantly faster than that.

What the Swift AI Package Actually Includes

The website is a static HTML, CSS, and JavaScript build hosted on Cloudflare Pages — fast, secure, and with zero server overhead to manage. The repository lives on GitHub, which means deployments are automatic on push and the entire history is version-controlled.

The content program runs at 18 articles per year, all written to a brand voice that was established before the first word was drafted. Social media accounts on Instagram and Facebook were created from scratch with profiles, bios, and an initial content calendar built in parallel with the website. Email newsletters go through Mailchimp on a monthly cadence. Short-form video is scripted and planned using the same AI workflows that produce written content.

What the Swift Package actually delivers is not a set of disconnected services. It is one unified brand being expressed across every channel simultaneously, from the same source of truth. That is what makes AI-assisted delivery viable at this scope and speed.

How Claude Code Changed the Build Process

The website was built using Claude Code as the primary development agent. Account leads could iterate on copy, structure, and layout without waiting for separate developer cycles. A change to the services page, a new section, a refinement to the appointment form flow — these happened in the same session as the conversation that prompted them.

The part of this that surprised us was how much the brand parameters mattered up front. When the voice documentation, the colour palette, the typography, the clinic’s positioning, and the target audience were all defined and fed into the workflow before a single page was drafted, the outputs needed dramatically less revision. When that groundwork was missing, outputs drifted toward generic. The AI is only as specific as what you give it.

The build for VanCityVet — yellow, teal, and navy; Poppins for headings, Inter for body; warm, community-focused, neighbourhood-first — came together quickly because the inputs were tight. The multi-step appointment form, the services architecture, the content hierarchy: all of it built through conversational iteration with Claude Code, not through traditional back-and-forth between a brief and a developer.

Brand Voice as Infrastructure

The biggest risk in AI-generated content is drift. Outputs that are technically competent but do not sound like the client — slightly too formal, or slightly too casual, or missing the specific local references that make a neighbourhood business feel real. That risk is not mitigated by better prompting alone. It is mitigated by treating brand voice as infrastructure.

For VanCityVet, the brand voice documentation lives in the project as a first-class input, not as a style guide someone reads once. Every content output — articles, social posts, email subject lines, video scripts — is generated against the same parameters. Warm. Honest. Community-focused. Specific to East Vancouver. Clear about what the clinic does and does not do (no emergency care; referrals go to the sister brand Trout Lake Animal Hospital down the road).

When brand voice is an input rather than a review criterion, the feedback loop compresses. There is less revision because there is less drift. Clients see their voice in the content from the first round, which builds trust in the process faster than any amount of explaining would.

What a Client Experience Like This Feels Like

The misconception about AI-assisted work is that it feels impersonal. That the outputs are generic, the process is transactional, and something is lost compared to a slower, more traditionally crafted engagement. That has not been our experience, and it was not the experience with VanCityVet.

When AI handles execution at speed, human attention can stay on the decisions that actually matter: the positioning, the voice, the structure of the patient experience the clinic wants to create, the story Dr. Sran wants to tell. The AI does not replace that thinking. It removes the gap between thinking it and having it built.

VanCityVet now has a website, a content calendar, social accounts, and a newsletter infrastructure. It was built in weeks, not months. It sounds like the clinic it represents. And the team that built it spent most of their time on strategy, not production.

Looking Ahead

The VanCityVet build is a proof point for the model, not a one-off. Every new Swift Package client benefits from the same workflow: brand voice defined up front, AI-assisted builds executed against tight parameters, parallel delivery across channels, human review at the checkpoints that matter.

The underlying infrastructure — the plugin conventions, the skills library, the content approval workflow in GD Command — gets stronger with every engagement. The speed advantage compounds. That is the difference between using AI tools and building AI-native operations.