Product analytics you own, typed end to end

Define your events once in Protobuf. Trulytix carries them from web and mobile apps through Vector into ClickHouse, where Grafana turns them into dashboards. Every piece runs on your own infrastructure.

Email us See how it works

Open source under Apache 2.0. Agent-ready, built with GDPR and the EU AI Act in mind.

client_event.proto

message ClientEvent { Platform platform = 1; EventType event_type = 2; string user_id = 3; string device_id = 4; int64 event_timestamp_ms = 5; string screen_name = 6; }

analytics.client_events in ClickHouse

event_time	platform	event_type	screen_name	user_id
12:04:31.208	IOS	APP_OPENED		7f3c…a1
12:04:32.940	IOS	SCREEN_VIEWED	catalog	7f3c…a1
12:04:33.517	WEB	SCREEN_VIEWED	pricing	b20e…5d

One schema, from the tap to the dashboard

The event schema lives in Git next to your code. Developers, analysts and product managers review changes to it like any other pull request, so web, iOS and Android always send the same fields with the same types.

Describe events

Protobuf in Git

A typed .proto file is the event map: one place that says what each event means and which fields it carries.
Send batches

Client SDKs

Apps serialize events to compact Protobuf and send them in batches, each with its own ID so retries don't double-count.
Ingest and validate

Vector

Vector receives batches over HTTP, drops duplicates, unnests and validates each event, enriches it server-side and writes it on.
Query and chart

ClickHouse + Grafana

Raw events land in ClickHouse for fast SQL over billions of rows. Grafana reads the same tables for dashboards and alerts.

Why build it this way

Events stop drifting between platforms

When iOS sends screenName and the web sends screen, every chart needs a footnote. A shared typed schema rules that out before the code ships.

Your data stays yours

Hosted tools like Mixpanel or Amplitude get expensive as volume grows and keep raw data on their side. Here every event sits in a database you run and can query directly.

Proven parts, little custom code

Vector, ClickHouse and Grafana are mature open-source tools. Trulytix wires them together rather than replacing them with a bespoke service.

Starts small, scales out

A single docker-compose runs the whole pipeline for a first project. The same pieces scale horizontally when traffic grows.

Ready for AI agents, built for regulated data

Analytics is increasingly read by agents as well as people, and event data is personal data. Trulytix treats both as first-class concerns from day one.

Agent-friendly interfaces out of the box

The typed event schema, the tables and the queries on top of them are exposed through documented, machine-readable interfaces, so AI agents can discover what you track and answer questions about it without screen-scraping dashboards.

Designed with GDPR and the EU AI Act in mind

Data stays in infrastructure you control. Personal fields are declared in the schema, so access, retention and deletion can be enforced where the data lives, and what agents can see is scoped and auditable.

Tell us what you track

Replacing Mixpanel or Amplitude, untangling events across web and mobile, or giving agents safe access to product data: write to us and we'll get back to you.

Email hello@trulytix.com