We build infrastructure, data pipelines and AI features, then release
or keep them running. A small international team, working for clients
across the globe with attention to details and your ideas.
AI & HumanWe combine AI speed with human reliability
Complex products6 big data and real-time products we built and run
GitOpsEvery change reaches production through Git automatically
Global coverageSupport in English, in your time zone
What we built
🎥
Crackleboom — random video chat
WebRTC · LiveKit · coturn · Go
Get randomly matched with strangers, AI agents, video content
and interactive boards. If it clicks, add them to your contacts.
Orchestration of AI and code that puts multiple types of
interactive content in one app
Scalable, secure, cost-effective and fun
On the web and on Google Play
🤖
An AI that joins the call
LLM · STT · TTS · GPU
An AI actor with a face and a voice that holds several parallel
real-time conversations on a single gaming GPU. Many characters
and voices, interruptible conversation, a face worth looking at
and a story worth hearing.
No vendor API in the loop, open-source code and bleeding edge
solutions
Scale the agents and every model behind them from one machine
to hundreds of nodes with Kubernetes
Build a new character in minutes, from the face to the story
it tells
☸️
Kubernetes for startups
k3s · Flux · Terraform · SOPS · Grafana
A Kubernetes cluster that scales to zero. You pay for the hours
it actually runs and it stops when nothing needs it. Full
control and privacy, continuous deployment, observability, and
GitOps all the way down to secrets and AI integration.
You do not need to know Kubernetes to ship on it, just push
the code
Secrets and environment variables live safely alongside your
code
Your database runs inside the cluster, managed for you
No vendor lock-in, move to your own Kubernetes in hours
🗄️
Postgres that looks after itself
CloudNativePG · Barman · GCS
Backups stream to object storage continuously, so we can restore
the database to any minute you name. Certificates rotate without
anyone noticing. Schema changes travel inside the application
binary and apply themselves on start.
Restore to any point in time
Encrypted connections, no manual renewals
Schema and code can never drift apart
💬
Clerk Chat — messaging and video calls
Go · HTMX · Postgres
Accounts, conversations, messages that arrive instantly, and
one-to-one video. The server sends plain HTML instead of a
single-page app, which is why it stays quick on a bad train
connection.
Very little JavaScript to download
Real-time app that scales to thousands of users
One binary and one database to run
The same structure backs every app we build
📈
Observability Platform
Cost · Observability · SRE
Logs, metrics and dashboards for every service from the first
day, not after the first outage. The same view shows what each
part costs to run, which is how web apps, video infrastructure,
GPU jobs and databases stay on a bill you would not notice.
Every service ships with dashboards and alerts
Logs and metrics in one place, searchable
Machines sized for real load, and we read the invoice every
month
swipe →
How working with us goes
1 · Call30 minutes, free. You describe the problem. We tell you honestly
whether we are the right people for it.
2 · ScopeA short written plan: what we deliver, in what order, and what it
costs. Fixed price where we can, a rate band where we
cannot.
3 · BuildA working demo every week. Delivered in your cloud account, on
our cluster, or both. You own everything from the first
commit.
4 · LaunchWe ship to production and hand over the docs, runbooks and keys.
Take it from here, or keep us on.
5 · Run optionalWe stay on as your ops team: deployments, monitoring, alerts and
on-call. You get a monthly report, the bill stays in check.
Talk to us for half an hour
Choose a time slot that suits you — it's free. Bring an architecture
diagram, a cloud bill that annoys you, or an idea you have not written
down yet.
