Skip to contentClosing Brackets
Start a project

What we build, and how a project runs

Two of these builds are live: Evergreen Softwash and Doge Buddy are in production at their own domains, and each links out so you can use it. The rest are concept builds — configurations we use to show how a property and a delegated agent look in a real operation — and every one says so. Every project, bot or property, runs through the same five stages below.

Evergreen Softwash — launch film still
Similar builds from $699
LivePropertyExterior cleaning & detailing · Puget Sound

Evergreen Softwash

That floor buys a fast static site that scores 100 on Lighthouse. The other end is a full SaaS app. Same team, and the scope document says which you need.

Luxury soft washing, pressure washing and concierge detailing for fine homes, vehicles, aircraft and watercraft across greater Puget Sound. The business needed its front of house built end to end: a site that explains five services, prices them in tiers, takes a quote request instead of a phone call, and gets found across twelve service areas.

What was built

  • Static Astro site on the brand kit: five service pages, each with tiered pricing, a process and its own FAQs
  • Quote form with spam defences, relayed straight to the sales inbox
  • Service-area coverage for twelve Puget Sound towns, with structured data for local search
  • A journal with four launch articles and an RSS feed
  • One 19 KB script, WebP-only images and inlined critical CSS: the engineering bar, on a phone
  • A five-second hero spot, cut as a concept piece ahead of launch

In production

Fully static, so there is nothing to patch and nothing to scale. Every service, price, town and opening hour is a data file rather than a page edit, and the quote form lands in the owner's inbox the moment it is sent.

Visit evergreensoftwash.com
  • Astro
  • TypeScript
  • Web3Forms
  • Render

Lines of work involved

How the film was cut
LiveBuild-a-Bot + propertyDog supplies · DTC e-commerce

Doge Buddy

A dog-supply store run by one owner and the agents they delegate to. The storefront is a Shopify Hydrogen build; behind it an ops service sources products, places supplier orders, answers support email and scores the catalogue — and every consequential action waits on an owner approval.

What was built

  • Hydrogen storefront on Shopify, with four collections built from one category source of truth
  • Sourcing agent: weekly trend harvest, supplier lookup and priced proposals the owner approves into live listings
  • Fulfilment that splits a mixed-origin order into one supplier order per warehouse, with per-leg delivery windows
  • Support agent on the shared inbox: triage and drafted replies that only go out behind an owner approval
  • Nightly product scoring and a weekly deprecation digest; underperformers retire only when the owner says so

In production

The job letter in practice. The agents may propose, draft and score; the owner approves. Refunds are locked to manual approval and the store's all-sales-final policy is enforced in code, so no agent can promise what the business does not offer.

Visit dogebuddy.com
  • Hydrogen
  • Shopify
  • Claude Agent SDK
  • Postgres
  • pg-boss
  • Railway

Lines of work involved

Concept buildBuild-a-Bot on existing systemsRegional logistics · 40 trucks

Halcyon Freight

Rate confirmations, bills of lading and proof-of-delivery photos arrived as email attachments and got typed into a TMS by hand. Roughly nine hundred documents a week, three people doing it, and a two-day lag before anyone could invoice.

What was built

  • Document extraction agent reading rate cons, BOLs and POD photos
  • Confidence scoring, with anything under threshold routed to a human queue
  • Two-way sync into the existing TMS — no migration, no replacement
  • Exception dashboard showing what the agent could not read, and why

What it shows

Build-a-Bot hired onto systems the client already runs. The delegated job is document intake and extraction; anything under the confidence threshold goes to a named person, and the TMS stays the system of record.

  • Python
  • Claude
  • Postgres
  • TMS API

Lines of work involved

Concept buildBuild-a-Bot + propertyHealthcare · six locations

Meridian Dental Group

Six practices, six front desks, and a recall list nobody had time to work. Patients who lapsed at eighteen months were never called, because calling them was always less urgent than the person standing at the counter.

What was built

  • Recall agent working the lapsed list by SMS, per location and per clinician
  • Intake forms that pre-fill from the practice management system
  • Escalation to a named human the moment a message reads as clinical
  • Per-location reporting the regional manager actually opens

What it shows

A job letter in a regulated setting. The agent is delegated recall and scheduling language only; every clinical thread goes to a person by design, not by exception.

  • Next.js
  • Claude
  • Twilio
  • PMS integration

Lines of work involved

Concept buildDelegated monitoringB2B SaaS · subscription revenue

Cartwright & Vale

Churn was visible in the numbers about a month after it was decided in the product. Support tickets, usage decay, invoice disputes and quiet champions leaving all lived in different systems, and nobody could see them as one shape.

What was built

  • Account graph joining product usage, support, billing and CRM into one model
  • Signal agents watching each edge for the patterns that precede a cancellation
  • Weekly risk digest to customer success, with the evidence trail attached
  • A written record of every signal that fired and what happened next

What it shows

Delegated monitoring, still a concept: the agent is allowed to watch and report, never to contact an account. Model the relationships first and the risk digest becomes an ordinary query.

  • Postgres
  • dbt
  • Claude agents
  • Metabase

Lines of work involved

The process

How a project runs

The same five stages for a property build and for Build-a-Bot, in the same order, so you can always say which one you are in and what comes next. Thirty days of stabilising is included in both.

  1. Scope

    Tasks in, tasks out, the tools involved, and the systems it touches. You get one document: what gets built, a timeline with real dates, and one fixed price.

  2. Build

    Work lands in stages, and every stage comes off the floor as software you can use. You see progress the week it happens rather than at a reveal.

  3. Integrate

    Your systems, your records, your business rules. A pilot against real data first, then production, with exceptions routed to the person the scope names.

  4. Hand over

    You own the spec and the code and configuration. Nothing is locked to us, and nothing about the handover depends on keeping us on retainer.

  5. Stabilise

    30 days included. We watch it run in production, fix what real traffic finds, and tune the stop rules before we step back.

Stage one

Start with a fixed-scope plan

The first stage is the scope: a job letter for a bot, a scope document for a property. Tell us what you want built and you will have what gets made, a timeline with real dates, and one fixed price — before any work begins.

Start with a fixed-scope planSee Build-a-Bot