Don’t define “net revenue” twice
If you’ve invested in Snowflake Semantic Views, you’ve already done the hard part: someone sat down and defined your tables, facts, and dimensions, and fought the battles over what “net revenue” actually means. The last thing you want from a new analytics agent is to do that work twice, or worse, watch two versions of the truth drift apart.
Zenlytic treats your Snowflake Semantic Views as a source of truth, not a parallel system. Zoë syncs with them directly and in both directions.
Reading from Snowflake
On a schedule you can control, a Zenlytic Proactive Agent connects to Snowflake using the same connection and role you’ve already configured, and reads your semantic view definitions via standard SQL. There’s nothing you need to deploy in your Snowflake account. Snowflake remains the source of truth for schema: the tables, facts, and dimensions you’ve already defined come across exactly as they’re governed today.
Zoë doesn’t just store a read-only copy; she translates those definitions into her own context layer, because she’s the window through which the business interrogates its data.
She sees every question people ask and learns from that usage. She sharpens her definitions, fills her knowledge gaps, and updates her business context as needed to best answer questions. That learning is why she maintains her own context layer.
Writing back (Optional)
Another Proactive Agent can close the loop. Like the sync agent, it runs entirely on Zenlytic’s side, there’s nothing to build, schedule, or maintain inside your Snowflake account, and it connects using the same role you’ve already configured (with write access granted only if you enable it).
Everything Zoë learns is committed to Git on every change, so the evolution of her context is always versioned, reviewable, and reversible. From there, the Proactive Agent automatically syncs those committed changes back to Snowflake, updating your semantic views in place. This means the definitions you govern in Snowflake always reflect what your business has actually learned, without anyone having to manually maintain them.
Why does it matter?
You get an AI analyst that’s useful on day one because she starts from definitions you already trust, and your Snowflake Semantic Views act as a source of truth that actually stays up to date with what the business is asking about. One set of definitions, with zero drift.
Start with Zoë reading from your Semantic Views, see Zoë answer governed questions immediately, and then turn on writeback when you’re ready to close the loop.