If your dashboards look clean but you’re still exporting to Excel… we’ve got thoughts (a few of them!).
First — good design can’t cover up bad data. But even when the data is solid, users often get stuck trying to close the gap between what happened and what to do next. Agentic AI is addressing this, and today we talk about how.
Meanwhile, we’re watching the terminology wars rage on (lakehouse, fabric, ecosystem — take your pick), and we have a few tech updates we think are worth keeping an eye on.
Conversational analytics isn’t just a trend — it’s changing how data teams support the business.
This blog breaks down how one platform is tackling common BI roadblocks using Agentic AI and compound intelligence to deliver faster, more intuitive insights.
💡Inside the blog:
Compound AI — how multiple agents work together to understand and explain data.
How “Genie” supports self-service by keeping context and surfacing clear, usable answers.
A five-step roadmap to launch conversational BI and scale it across your org.
“The value isn’t in the tool — it’s in how it removes friction between business users and trusted data.” – John Bemenderfer, Managing Consultant
📡 Data Signals: Hot Topic
1. 🛑 Dashboards aren’t the problem. The last mile is.
Most dashboards do a solid job tracking KPIs. But many stop short when you need to understand why something changed — or what to do next. You end up bouncing between tools, filing a support ticket, or exporting to Excel just to keep the analysis going.
This is the last-mile problem in analytics: the space between “what happened” and “what do I do about it?” And it’s where most self-service promises fall apart.
Why we’re talking about it: Because agentic AI is closing that gap. It gives business users a way to explore data conversationally, get deeper context without writing queries or needing to be trained on yet another reporting or BI tool, and actually follow their train of thought — all without waiting on a new report. If self-service hasn’t delivered, this might be the shift that changes things.
2. 🧠 Is your architecture a lakehouse, a fabric, or an ecosystem? Yes.
Let’s be honest — terminology is getting out of hand.
What started as a way to describe architectural patterns has become a buzzword buffet. Now, a “data fabric” might include everything from metadata to observability. Add in FinOps, DataOps, and Platform engineering, and suddenly we’re calling it an “ecosystem.”
The takeaway: Don’t get caught up in chasing the right label. Focus on the outcomes: Are your data systems integrated? Are governance and access improving? Can your teams move faster with confidence? Call it whatever you want. Just make sure it works.
Emerging Tech Insights
A few tech updates on our radar:
1. 🤖 Automated Data Products: Making Data Mesh More Realistic
Zhamak Dehghani (the mind behind data mesh) is back with a new concept: automated data products. And this time, there’s tech behind it. Her new platform, Nextdata OS, aims to take the manual lift out of implementing data mesh — something that’s held a lot of orgs back.
Why this matters:
Most teams love the idea of data mesh, but struggle with execution.
Nextdata OS bakes in things like semantic modeling, access control, lineage, and contract enforcement — core parts of the mesh playbook.
This could shift data mesh from theory to practice for orgs that don’t have the engineering horsepower to build it all from scratch.
👀 Worth keeping an eye on — especially if you’ve been waiting for mesh to become less… aspirational.
2. 🐍 dlt: A Lightweight Loader with Big Potential
We’re keeping an eye on dlt, a Python library that simplifies pulling data from REST APIs. It’s optimized for loading, integrates with Databricks, and could reduce the need for custom ingestion code.
Why it matters: If you’re spending too much time building wrappers for API sources, this might be your shortcut to faster pipelines. We haven’t tried it at scale yet — let us know if you have.
3. 🧱 Databricks Asset Bundles: IaC Comes to Analytics
Databricks is leaning into engineering best practices with Asset Bundles — an infrastructure-as-code approach that brings source control, CI/CD, and templating to your lakehouse projects.
Why it matters: You can now speed up initial project setup and cut down delivery time without sacrificing governance or consistency.
Busting Data Myths
🛑 Myth: "New tool? You can replicate old data models and processes into your modern tool for now and then fix it later."
Reality: Forcing modern platforms into legacy patterns wastes time, money, and opportunity.
Too often, we see teams retrofit their shiny new stack to match outdated assumptions — like splitting databases by reporting dimensions or over-engineering a fact table when a type-2 dimension would do.
💡 Takeaway: If you've invested in modern tools, let them be modern. Old-think limites your team. Let your architecture — and your people — evolve.
LOL Moment
When your dashboard is all vibe and no validation.
