The problem
Operations data lives in twelve different SaaS tools. Every team needs the same five numbers from those tools, and every team builds the same five spreadsheets to get them. The work isn’t analytics. It’s just gathering.
UniDeck started from a simple observation: most teams don’t need a dashboard tool, they need a way to stop building dashboards.
What we built
A no-code dashboard layer that sits on top of the SaaS your team already uses. Configure once, share with the team, stop maintaining spreadsheets.
The product is built around three primitives:
- Sources — first-party integrations to Google Workspace, GitHub, Slack, Stripe, and a long tail of custom APIs.
- Widgets — composable views that read from sources and render in the dashboard. Designed for a non-engineer to wire up in minutes.
- Dashboards — shareable, embeddable, snapshottable.
The engineering shape
The hard part is not the visualisation. It’s the integrations: rate limits, refresh tokens, schema drift, vendor outages. UniDeck’s integration layer is the load-bearing piece, with strict isolation so a vendor outage at one source doesn’t cascade across the dashboard.
The frontend is opinionated about composition: widgets compose like Lego, not like a query builder. That choice keeps the surface usable for the people who actually need to read these dashboards every Monday morning.
What it looks like in production
unideck.app is the running product. UniDeck is a separate B.V. inside the Besharat Enterprises group; the lab continues to do engineering on it.
What we learned
Integration design is product design. Most of the user-visible quality of UniDeck comes from how the integrations behave when the source is misbehaving — partial failure, rate-limit, stale data. Getting those right was the bulk of the year.