

Two years ago, Snowflake decided to locate its major European R&D center in Warsaw. Coming from two different backgrounds (a design studio & a big tech giant): we share our retrospective on two years working at Snowflake. We'll talk about how an extremely collaborative culture, "slow haste," and a strong focus on execution, contribute to an environment where design is the key differentiator.
Watch this WaysConf session, then continue with related talks or explore the current programme.
A case study of design inside a rapidly growing cloud data platform, from technically simple beginnings to a product serving broader and more varied roles. The speakers explain how data can be combined securely, how role-based access creates a demanding interaction problem, and how independent product trios use staged releases to learn without placing every customer at risk. The central lesson is that scale requires both speed and restraint: teams should reach a small real audience quickly, learn from it, and delay broad availability until the experience is trustworthy.
Technical simplicity for specialists is not enough when a platform reaches new roles; navigation, terminology, and workflows must evolve with the audience.
Watch from 8:49External and internal datasets become valuable when the interface helps people combine them in service of a specific operational decision.
Watch from 15:16Role-based access must communicate both what a person can do and why, without exposing the full complexity of the security model.
Watch from 21:14Private preview, public preview, and general availability create explicit levels of exposure, evidence, and product responsibility.
Watch from 26:59Independent product trios work more coherently when their ownership follows an outcome or journey rather than a list of disconnected features.
Watch from 39:15The speakers introduce the platform, its growth, and the role of design in making complex infrastructure usable.
A product originally aimed at data engineers must adapt as more roles and use cases arrive.
Worksheets and shared datasets illustrate how the platform supports questions that cross data sources.
A security model becomes a concrete design challenge involving roles, abilities, and organizational scale.
Release stages limit exposure while giving teams progressively stronger evidence about product quality.
Fast access to early customers is separated from the decision to expose an unfinished feature broadly.
The closing discussion describes autonomous teams responsible for coherent customer outcomes.
These concise notes were edited from automatic captions and checked against the talk structure. They are not a verbatim transcript.
A platform may begin with an elegant technical model for a specialist audience and still become difficult as its reach expands. New roles bring different vocabulary, goals, and tolerance for complexity. Design has to preserve the power of the underlying system while presenting paths that make sense to analysts, administrators, and operational users.
Watch from 8:49A company dataset rarely contains every variable needed for a decision. Bringing marketplace data into the same working environment lets teams examine relationships such as weather and sales without rebuilding an integration for each question. The interface should keep the connection understandable while allowing specialists to inspect and manipulate the underlying structures.
Watch from 15:16Role-based access control protects large organizations, but its concepts can be hard to predict from the user’s perspective. Good permission design makes the current role, available actions, and restrictions legible at the moment they matter. It should also help administrators understand consequences before changing access across a complex organization.
Watch from 21:14A private preview allows a small invited group to encounter the feature while the cost of change remains low. A public preview broadens the sample and raises expectations. General availability means the feature can no longer depend on special participation. Naming these stages gives teams a shared language for learning, quality, and customer risk.
Watch from 26:59Product trios can make decisions quickly when they own a recognizable user journey and its outcome. The team is not merely handed a feature specification; it investigates a problem, interprets evidence, and chooses how to respond. That autonomy remains accountable because customers, metrics, and the release model expose whether the journey actually improves.
Watch from 39:15