What we build

Seven systems. Built custom, every time.

We do not have a product to sell you. Each build is architected for the operation it serves, then wired into the software that operation already runs.

The seven

Automation first. Intelligence second. In that order.

We started in automations, wiring together systems that were never designed to talk. That is still the foundation, because AI bolted onto a broken process only fails faster. Most engagements begin with one of the seven below.

The foundation

Workflow & operations automation

Where we started, and still the foundation everything else sits on. Airtable-backed operations, integration pipelines between systems that were never meant to talk, triggers and approvals, invoicing, procurement, scheduling, and reporting. The unglamorous plumbing that decides whether the clever AI on top actually works on a Tuesday.

Airtable-backed operationsIntegration pipelinesTriggers & approvalsInvoicing & procurementScheduling & reporting
And on top of it, the six

AI websites & applications

Conversion-focused sites and web apps with the AI inside the funnel rather than bolted on beside it. Booking and ordering engines, dynamic personalisation, lead capture that qualifies before it routes, and mobile apps where the job needs one. This site is one of ours.

Booking & ordering enginesPersonalisationQualifying lead captureMobile apps

Conversational agents

Web chat, voice, and messaging assistants handling enquiries around the clock, in your customer's language, trained on your business rather than on the open internet.

Web chatVoiceMessenger & WhatsAppMultilingual

Autonomous outreach

Prospect discovery, contact enrichment from public sources, and personalised campaigns that run unattended, with open tracking, retry logic, and unsubscribe handling built in.

Prospect discoveryEnrichmentPersonalised campaignsTracking & unsubscribe

Content & video automation

AI-written posts, generated imagery, and presenter video, published to Facebook, Instagram, LinkedIn, and TikTok on a schedule you set once.

AI-written postsGenerated imageryPresenter videoScheduled publishing

Computer vision

In-situ product visualisation, so a customer sees the thing in their own room from a phone photo, at true-to-scale perspective, inside the buying journey rather than beside it.

In-situ visualisationTrue-to-scale renderingImage understandingInside the buying journey

Unified analytics

One view across the platforms you already pay for, reconciled, so the numbers finally agree with each other.

GA4 + MetaReconciled numbersOne dashboardScheduled reports
The stack

The stack, in full.

No black boxes. Google Cloud and Firebase underneath, Claude and Gemini on top, secrets in Secret Manager and never in a config file. Everything we deploy, you can open up and read.

40+
Integrations in production
4
AI models orchestrated
43
Cloud functions live
0
Secrets in config files
AI & models
Anthropic ClaudeGoogle GeminiVeo 3.1ElevenLabsD-IDAnthropic SDKGoogle GenAI
Cloud & runtime
Google CloudFirebaseCloud FunctionsFirestoreCloud StorageCloud SchedulerSecret ManagerNode.js 22
Automation & ops
AirtableZapierMake.comWebhooksCron pipelinesApproval workflowsPipedriveClickUp
Data & analytics
GA4Meta PixelGoogle Places APIGoogle Sheets APIFirestore queriesOpen & click trackingSMTP / Nodemailer
Front end & apps
JavaScriptTypeScriptNext.jsReactViteTailwindCapacitorStatic + CDN
How an engagement runs

One slice at a time, priced before you commit.

No twelve-month transformation programme, no hourly billing, no surprise invoice. Every build is scoped to one problem, quoted as a fixed price, and in production before we talk about the next one.

1. The call

Thirty minutes, no deck. You describe the operation, what breaks, what it costs, who it affects. We say honestly whether AI is the answer. Free, and it usually ends with a clear first slice or a clear no.

2. The scope

Within a few days you get a one-page brief: the exact problem, what we will wire together, what "working" means, the fixed price, and the date. You approve it or we change it. Nothing is built until you do.

3. The build

The first slice ships to production in weeks, not quarters, inside the software you already run. You see it working on real data, with admin controls and logs, before the next slice is discussed.

Single build

One system, one fixed price, handed over with the source, the documentation, and a walkthrough. You own it outright.

Best for a first slice

Build and run

The build plus a monthly care plan: monitoring, model updates, small changes, and a person who answers when something looks off.

Best for systems that face customers

Fractional AI operations

A standing engagement for teams that want a run board of automations delivered slice by slice, with the priorities set together each month.

Best for a roadmap, not a one-off

Every price is quoted after the call and fixed before the work starts. Payment in stages tied to the slices, never to hours.

Questions we get asked

Straight answers, before the call.

The things people ask in the first ten minutes, answered the way we would answer them on the phone.

What does RyV-ai actually build?

Custom AI systems, wired into the software you already run. Workflow and operations automation is the foundation. On top of it we build conversational agents for chat, voice, and messaging, autonomous outreach, content and video automation, computer vision, AI websites and applications, and unified analytics. They are production systems, not prototypes or strategy decks.

Do we have to replace the software we already use?

No. Your PMS, POS, CRM, and channels stay. The AI layer goes on top and talks to all of them. Most builds start by wiring together what you already have, because intelligence bolted onto a broken process only fails faster.

Will this replace our staff?

No. AI is the tool and your team is the point. We automate the repetitive work: data entry, report compilation, routine enquiries, scheduling, and the spreadsheet everybody pretends is temporary. Your people keep the parts that need a human, and every system ships with admin controls and real logs, so they stay in charge of it.

How does a project start?

With a thirty-minute call and no deck. Describe the operation, what breaks, what it costs, and who it affects. We will say honestly whether AI is the answer and what it would take to build. Then we cut one slice: one painful, expensive problem, shipped into production and earning its keep before we discuss the next one.

How does pricing work?

Each build is scoped to one problem at a time, and you know what it would take before you commit. We do not carry an agency's overhead, and the automation foundation means less bespoke code per build, so engagements come in well under typical agency rates for comparable scope.

Where do you work?

We build for businesses in the USA, New Zealand, Australia, Indonesia, and Europe, and we work remotely across those time zones. The run board looks the same wherever the operation is.

Which AI models and platforms do you use?

Anthropic Claude for reasoning and copy, Google Gemini and Veo for imagery and video, and specialist models for voice and avatar, all running on Google Cloud and Firebase with secrets in Secret Manager rather than in config files. We choose the model per job, and you own the system we hand over.

Can we see it working before we commit?

Yes. The interactive demo asks four questions about your operation, assembles the run board we would build for you, then writes the brief live. This website is one of ours too: the chat, the lead capture, and the outreach behind it run on the same stack we sell.

Tell us what's eating your week.

Thirty minutes, no deck. Describe the operation and we'll tell you honestly whether AI is the answer, and what it would take to build.