Case Studies

Proof, not promises.

Three builds, written up with the stack, the timeline, and what actually shipped. Two of these are systems we built and still operate ourselves, which is why we can publish the detail.

  • Challenge
  • Approach
  • Stack
  • Timeline
  • Impact

Internal build, operated by Maestro

Multi-platform hospitality operations

An automated short-term rental operation that runs across two booking platforms without a human touching the calendar.

Stack

Node.jsiCal sync servicesSmart lock APIsAutomation workflowsCloud hosting

Timeline

8 weeks to first automated turnover, iterated across the following 6 months

Challenge

A short-term rental listed on both Airbnb and Booking.com is one double-booking away from a cancelled reservation and a damaged listing score. Manual calendar syncing does not survive contact with same-day bookings, and manual guest access means somebody has to be physically present or trust a lockbox code that never changes.

Approach

  1. 01Two-way iCal sync between Airbnb and Booking.com with a polling interval tight enough to close the double-booking window, plus conflict alerts when a clash slips through.
  2. 02Smart lock integration that generates a unique guest access code per reservation, activates at check-in time, and expires at checkout with no manual step.
  3. 03Guest communication SOPs: templated pre-arrival, arrival, mid-stay, and checkout messages, triggered by reservation state rather than sent by hand.
  4. 04An operations runbook the cleaning and maintenance crew works from, so turnover does not depend on one person remembering the sequence.

Impact

  • Calendars stay in sync across both platforms without manual intervention.
  • Guest access is issued and revoked automatically, per reservation.
  • Turnover runs from a documented SOP rather than one person's memory.
  • The operation runs hands-off from a different continent.

Internal build, operated by Maestro

Consumer beverage brand operations system

Operations infrastructure for a two-country beverage company: one factory, two retail locations, and a merchant payment integration.

Stack

ReactNode.jsPostgreSQLMerchant payment APIsCloud hosting

Timeline

10 weeks to first full month of clean data

Challenge

Production ran in Lagos, retail ran in two separate locations, and ownership sat in Houston. Inventory counts, production output, and daily sales lived in notebooks and WhatsApp messages. Nobody could answer how much stock existed, what a location sold last week, or whether a payment had actually landed, without calling three people.

Approach

  1. 01A production and inventory tracking system covering raw materials in, finished units out, and stock on hand per location.
  2. 02Sales capture at both retail locations, reconciled daily against inventory movement so shrinkage shows up as a number rather than a suspicion.
  3. 03Merchant payment integration so customer payments post against the correct sale rather than arriving as an unlabeled bank credit.
  4. 04Staff workflows written down and trained, so the data entry happens as part of the job rather than as a reporting chore at month end.

Impact

  • Full visibility across Lagos operations from Houston.
  • Daily reconciliation between sales, payments, and inventory movement.
  • Stock positions per location available on demand rather than on request.
  • Staff workflows documented, so a new hire is productive without shadowing for a month.

Internal tooling, used on client delivery

AI-assisted internal tooling

The delivery infrastructure we run our own business on, and the same systems we deploy inside client engagements.

Stack

GitHub ActionsPlaywrightCypressTypeScriptClaude and OpenAI APIsAWS

Timeline

Continuous, in production since 2024

Challenge

A delivery business that ships for other people has to hold itself to the standard it sells. Manual deploys, untested releases, and hand-written outreach do not scale past a few engagements, and they make quality depend on who happened to be paying attention that week.

Approach

  1. 01CI/CD pipelines with automated build, test, and deploy stages, so releases are routine rather than events.
  2. 02QA automation frameworks in Playwright and Cypress covering critical user paths, wired into the pipeline so a regression blocks the merge.
  3. 03AI-integrated outreach and delivery tools that draft, classify, and route work, with a human approving anything that leaves the building.
  4. 04Shared templates and runbooks so a new engagement starts from a known-good baseline instead of a blank repository.

Impact

  • Releases are routine, tested, and reversible.
  • Regressions are caught by the pipeline rather than by a client.
  • New engagements start from a baseline instead of from scratch.
  • These are the same systems we deploy for clients, not a separate showcase.

Client-confidential engagements available on request under NDA. Tell us your industry and we will send the closest match we are permitted to share.

Want this run for you?

Tell us the role or the project. Profiles in 48 hours, fixed quotes in 48.