ProductPricingDocsGet started
Teamest. 2024

A small team,
building bigger questions.

Sumidata is the AI-powered analytics platform that answers the questions before you finish asking them. We unify session replays, technical logs, product events and revenue context into one columnar store you can talk to in plain English.

1founder · operator
10+ yrssystems experience
Single storeClickHouse · SQL all the way down
Ownedevery line, every decision
01 · Leadership

One founder.
Decades of systems work.

The team that ships Sumidata is small on purpose — every line of the platform is owned by someone who can explain why it's there. No layers, no telephone game between research and engineering.

Ruslan Palahin
Founder · CEO · Lead Architect

Ruslan Palahin

LinkedIn

Systems architect with extensive experience building scalable software. At Sumidata, Ruslan leads the architectural vision — designing systems that can answer unpredictable questions on massive datasets in real time.

Core focus
System architecture & low-latency data pipelines
AI-powered analytics & text-to-SQL
Product vision & user experience
02 · Vision

"Analytics says 30% dropped off.
We show you why."

The observability stack you have today tells you what happened — never why. Sumidata closes that loop. Replays, technical logs, product events, and revenue context, in one query, with video evidence.

The problem
Four dashboards, three vendors, no shared schema. By the time you've stitched the story together, the user has already churned.
The fix
Ask in plain English. The assistant joins the right tables, cites the rows, and shows you the replay. One platform, one query language, one bill.
The future
Every signal a product team has, addressable as a SQL table — joined to revenue, replayable in a click, narratable by an AI that understands your data.
03 · Principles

How we work.

Four rules that quietly shape every decision — from the schema we picked to the price we charge.

01
Own the stack
No proprietary blobs we can't debug. Every byte is in ClickHouse, every query is plain SQL, every decision is traceable to a person on the team.
02
Cost is a feature
Charging per seat punishes the curious. We price for volume so the data scientist, the PM, and the on-call engineer can all open the same query without a budget meeting.
03
AI is a colleague
The assistant doesn't hide the SQL — it shows it. Every answer is an inspectable statement you can copy, paste, and ship. No black boxes, no hallucinated metrics.
04
Ship narrow
Better to do nine things well than ninety things poorly. Each surface earns its place by joining the rest — replays cite errors, errors cite revenue, revenue cites cohorts.
Curious about the architecture?

The technology
behind the answers.

One columnar engine. SQL all the way down. Read how we built Sumidata for sub-second analytics on billions of rows — or get on a call to talk shop.

Read the docsTalk to the founder