Hoping this email finds you wrapping Q3 with all good things data — clean reconciled reports, dashboards delivering timely metrics, and clear insights for Q4 🤞
In this issue of The Insider we go a little philosophical (how to think about AI when “perfect” isn’t realistic) — and lot practical: guaranteed ways to overcome trust issues with your reporting, plus a look at two new helpful tools for Databricks users.
Plus:
Our Recommended Read: Building AI products in the 'Probabilistic Era'
Hot Topic: Tools to make Databricks less complicated
Emerging Tech Insights: Databricks partnered with OpenAI??
From the Field: Two ways to gain trust in your dashboards
LOL Moment: The response you want to give vs the one you send
AI makes many of us uncomfortable because it’s not binary.
You can ask the same question twice and get two different answers. That can feel broken or wrong. But that’s the point. It can handle nuance in a way your old data tools never could.
So, how should we shift our mentality with this new reality?
Stop chasing 100% with your model. You’ll never get perfect answers every time. The trick is figuring out how much unpredictability your business can live with without breaking trust.
Your buyer funnels are toast. Users won’t move in straight lines anymore. They’ll zig, zag, and stumble into new use cases you didn’t see coming. You’ve got to track the whole messy journey.
All context matters. Every weird query, every off-the-wall answer, it’s all downstream data you can mine to see what’s actually working.
Adopt the lab coat. You can’t plan this stuff in a conference room. You experiment, you measure, you adapt. And you need to do it fast.
🔎 This article by Gian Segato — “Building AI Products In The Probabilistic Era” — comes highly recommended by our consultants. Read it to go deep on the shift from determinism to empiricism that data engineers and scientists face today.
Hot Topic
Skip the plumbing in Databricks
All the choices that make Databricks powerful can mean weeks of setup and multiple SMEs.
From project-tested best practices, we built 2 tools streamline setup:
Platform Bootstrapper – Spins up dev, test, and prod environments with best practices baked in — done in hours, not weeks.
Platform Profiler – A live dashboard to track progress, spot inefficiencies, and keep things healthy.
Here’s how it played out for one architectural engineering client:
Before, every new environment meant pulling senior architects off billable work to hand-configure jobs — risking breakage when they pushed changes downstream. With the Bootstrapper, those environments stood up in days, fully wired with ingestion jobs and CI/CD. The Profiler gave them a single view of what was running smoothly and what wasn’t. Instead of spending weeks untangling broken pipelines, their data team is now focused on delivering insights that solve business problems.
If this caught your attention, we’re featured in a deeper dive on how these tools work and what they unlock.
Translation: your prompts and data stay in the lakehouse, Unity Catalog remains the single control plane, and you stop writing glue code to bolt external models onto your stack.
What changes for you:
No duplicate governance or data egress to “some API.”
State-of-the-art models without custom wrappers and one-off logic.
New model capabilities arrive natively; you adopt, not re-platform.
“The real win isn’t model access — it’s deleting the scaffolding. If your best people are wiring evals, secrets, and policy shims, you’re losing. Native GPT-5 lets them ship agents while audit trails live in one place.” -John Bemenderfer, Consultant
From the Field
Two ways to gain trust in your dashboards
Ah, it feels good to know exactly how to tackle a problem.
Make sure everyone is defining metrics the same. When systems define metrics inconsistently, analysts spend more time reconciling reports than delivering insights. It’s expensive, frustrating, and keeps teams stuck in the weeds.
The fix? Stand up a consistent, queryable layer for standardized metrics so everyone’s speaking the same language, and analysts get their hours back.
Make data transformations transparent. Dashboards don’t matter if no one believes them. Without transparency in how data is transformed, stakeholders second-guess results and adoption stalls.
The fix? Make data transformations auditable and reliable. How to do it depends on your tech stack, but every modern tool makes this very achievable!
LOL Moment
The ask: Next week? Maybe this week? What’s the hold up?
The response you want to give: “What’s the hold up?? How about 20 years of tech debt and a data warehouse that cries when it rains!”
The response you’ll give: “I’ll get back to you soon with a plan…”
Have any good memes? Send it to our content manager and get featured on our social channels.
