Skip to content
Let’s talk

Success story · Housing and utilities · Kazakhstan

AulaThree releases a day instead of one in four months, 3 people doing the work of 16, and a cloud bill down 70%

Aula is one app for apartment residents and the 109 companies that manage their buildings, with 247,000 accounts. As fractional CTO I rebuilt how the product is made and run: development and testing went from 16 people to 3, the servers from 18 to 3, the cloud bill fell by 70% and errors in the mobile app by 97%. A task now reaches production in a day instead of 2 to 3 weeks, and the product ships three times a day at 99.99% uptime, a pace most in-house teams never reach.

Get the full case as a PDFAll stories
Client
Aula
Industry
Housing and utilities
Country
Kazakhstan
Users
247,000 accounts, 38,000 active a month
Customers
109 property management companies
My role
Fractional CTO
Platforms
iOS, Android, web
Website
aula.kz

In numbers

releases, up from one every four months, a pace few teams reach
3 a day
people now do the development and testing that took 16
16 → 3
cloud bill after I cut the servers from 18 to 3
−70%
from task to production, down from 2 to 3 weeks
1 day
uptime, with a switch to the reserve data center in minutes
99.99%
call center cost, now that a voice agent takes the calls
−62%

The resident app

Requests to the management company, building news, votes and neighbours' listings. The screens come from the App Store listing.

App StoreGoogle Play
  • Aula app home screen with the management company's announcement
  • List of maintenance requests with their status
  • Notice about a hot water shutdown
  • A residents' vote on bicycle racks in the parking
  • Neighbours' listings
01

The starting point

Aula connects residents of apartment buildings with the companies that manage them. Residents file requests, read announcements, vote and sign documents with a digital signature, and management companies run their dispatchers and repair crews in the same system.

When I joined, a release went out once every four months. Sixteen people developed and tested the product, it ran on 18 servers, more than six years of data sat in MongoDB, and human operators answered the call center. The product was slow to change and expensive to run.

02

What we did

  1. AI agents do the work of a 16-person team

    I moved coding, review and maintenance to Claude Code and Codex agents and built the team around them. Development and testing went from 16 people to 3, and development got at least five times faster. AI tools and tokens cost 15% of the payroll budget, a small price for the output of a whole department.

  2. Three releases a day, and a task in production within a day

    Unit and integration tests cover about 90% of the code, and our own end-to-end suite runs on top of them. The full set runs on every change, from the pre-commit hook to a pre-production check with automatic rollback. That is how a release goes out three times a day instead of once every four months, and a task reaches production in a day instead of 2 to 3 weeks, without the manual QA most teams still pay for.

  3. 18 servers down to 3, and a cloud bill 70% smaller

    I cut the infrastructure from 18 servers to 3, and the cloud bill fell by 70%. The attack surface shrank with it: distroless containers and a single entry point behind the Cloudflare WAF. Uptime is 99.99%, backups trail production by about a minute, and switching to the reserve data center takes minutes, the kind of recovery that usually needs a dedicated ops team.

  4. Full observability from day one, with nothing to build

    Aula runs on my engineering platform: my GitLab instance, my CI runners and my Sentry. The team saw everything from the first day and never spent time or money building its own tooling.

  5. 97% fewer errors in the mobile app

    Sentry and session replay catch every error together with the steps that led to it. Agents pick the errors up and fix them in cycles every six hours, so a bug is gone the same day instead of waiting in a backlog.

  6. Six years of data moved to PostgreSQL in six working days, and 3 times faster

    I moved six years of production data from MongoDB to PostgreSQL with less than ten minutes of downtime, a migration that usually keeps a team busy for months. The database now runs 3 times faster than the old MongoDB. The core stayed a single Laravel monolith with few moving parts, which keeps it fast and cheap to run.

  7. A voice agent cut the call center's cost by 62%

    A voice agent that speaks Russian and Kazakh takes the call, identifies the resident and files the request. The call center already costs 62% less, and as more of the operators' work moves over to it, more calls no longer mean more operators on the payroll.

03

The result

Three people on three servers now ship Aula three times a day at 99.99% uptime, on a cloud bill 70% smaller. It used to take 16 people and 18 servers to ship once every four months, and 2 to 3 weeks for a task to reach production; now a task gets there in a day.

The resultBeforeAfter
ReleasesOnce every 4 months3 a day
From task to production2 to 3 weeks1 day
Development and testing16 people3 people
Servers183
Cloud billPaid for 18 servers70% lower
TestingManual QAAI agents, ~90% coverage
DatabaseMongoDB, six yearsPostgreSQL, 3 times faster, moved in 6 days
Call centerHuman operatorsVoice agent, ru and kk, 62% cheaper
04

What comes next

Over the next few months I am moving the backend from PHP to Go. We estimate about twice the performance at half the server cost, with fewer services to maintain.

Stack

  • Laravel
  • PostgreSQL
  • Redis
  • Vue 3
  • Flutter
  • Docker
  • Cloudflare
  • GitLab CI
  • Grafana
  • Sentry
  • Claude Code
  • Codex
oleg.isSent on request

Full case · PDF

Aula

Three releases a day instead of one in four months, 3 people doing the work of 16, and a cloud bill down 70%

  1. 01Architecture before and after, with diagrams
  2. 02How 18 servers became 3, and what it did to the cloud bill
  3. 03How 3 people do the work of 16: roles, agents, review
  4. 04The pipeline behind three releases a day with no manual QA
  5. 05The six-day migration plan, step by step
  6. 06The voice call center: stack and cost per minute
Prepared by Oleg SotnikovPDF

The full Aula case

The public story stops here. The PDF has the rest: the architecture before and after, the migration plan, how the team works with AI agents, what it all costs to run and where the savings came from.

I send every case myself and use your email for nothing else.

More stories

All stories
02MetizPromServiceMetalworking and parts manufacturing · KazakhstanFractional CTOA factory saves $300,000+ a year and halved its IT costs after two engineers I hired replaced its contractorsMetizPromService machines parts to customers' drawings in Kazakhstan. When I joined, contractors ran its IT and its accounting, production and lathe-control systems, and the company paid for outside software and licences. I built its own infrastructure, hired and trained two engineers, and we replaced every paid production system with an ERP and CRM of our own, with the ERP live in 3 months. The company now saves more than $300,000 a year on software, spends over 50% less on IT than the contractors cost, salaries included, and its CNC machines stand idle 23% less, with half as many unplanned repairs.Read the story03CV RocketJob search and AI CVs · USA and worldwideFractional CTOOne engineer and a fractional CTO took CV Rocket from zero to sales in five weeks and to 1,000+ customersCV Rocket writes a CV for one job posting in 15 to 50 minutes and brings the US and world job market into one place: more than 1.5 million postings from 89,000+ job sites. I designed the architecture, the cloud services and the payment integration, and one engineer and I built the product in four weeks, a launch that usually takes a full team several quarters. It sold from week five and now has more than 1,000 customers.Read the story04CallbrainAI call centers and telephony · WorldwideFractional CTOAn AI operator closes 92% of calls without a human, in 57 languages, and costs 3 to 5 times less a minuteCallbrain's AI operator answers and makes calls on a company's own number, works in its CRM or 1C during the call and closes 92% of calls without a human, handing the rest to people. It replies in under a second and lets the caller cut in mid-sentence, and a minute of it costs 3 to 5 times less than a minute of a human operator, which brings call center costs down by 50 to 80%. I designed the service end to end, from the PBX to the realtime voice models, and built it with one engineer, where a product like this usually takes a telecom company and a large team.Read the story

Want the next story to be about your company?

In a 30-minute call we pick the first task to hand to AI and estimate what it will save you.

Book a callThe call is free, and you talk to me directly.