In our industry, tools capture the headlines, but the work behind them (the work you do) is what usually makes the difference.
Technology vendors seem to be recognizing that, too. The latest AI announcements aren't just smarter models. They're about giving AI more business context, stronger governance, and better guardrails so people remain in control of the outcomes.
Below are several recent updates that reflect that trend, helping you to build data and AI solutions that solve the right problems, not just generate faster answers.
And since this lands just before the Fourth of July 🇺🇸, happy early 250th to America. Hope you get a long weekend with good food, good company, and only one person asking you to explain what you do for a living at the barbecue.
Below you'll find:
📖 Our Recommended Read: Databricks Data + AI Summit takeaways
🔥 Hot Topic: The work behind AI-enabled delivery
🛠️ Other Tech Insights: Updates from Snowflake, Power BI, Sigma, ThoughtSpot, and Coginiti
😂 LOL Moment: A little too familiar for anyone in data
📖 Databricks Data + AI Summit takeaways you need to know
Databricks had an influx of product announcements at their annual Summit, so our experts distilled the most relevant updates into a consumable list to help you stay up-to-date on what impacts you most as a data leader:
Genie is giving AI agents and business users more context for AI-driven decisions. Genie Ontology gives agents a layer of business definitions, relationships, and trusted metrics to reason over. Genie One brings that context to business users through natural-language interaction inside tools like Slack and Teams, instead of sending them to another dashboard.
Lakebase brings applications closer to governed data. It gives teams a way to build operational apps directly on the Lakehouse, reducing the need to copy data into separate application stacks and govern the same information twice.
Unity AI Gateway takes on model sprawl. As teams use more models across more use cases, they need a central way to manage access, permissions, usage, and cost. AI governance needs to happen where the work is running, not after the fact.
ZeroOps is a new Databricks capability for AI-assisted platform operations. Itmonitors jobs, machine learning workflows, and platform health – then recommends improvements.
The full article connects these updates to the challenge most all facing: moving from AI experimentation to trusted, production-ready AI solutions.
Accelr8 is helping clients modernize complex data environments faster, with less risk and more confidence.
At EMC Insurance Companies, it reduced a modernization timeline by 90%. For another client, Accelr8 compressed what is typically 4–8 weeks of manual analysis and conversion of legacy QlikView scripts into days.
Those results reflect why we built Accelr8: to combine AI, reusable accelerators, governance, and proven delivery practices developed across hundreds of engagements. The result is faster execution, greater reliability, and solutions built on proven experience instead of starting from scratch.
Read more about how Accelr8 is removing bottlenecks for common data and AI initiatives and helping companies get to the point quicker: getting more value from their data ➡️
Emerging Tech Insights
Top updates for Snowflake, Power BI, Sigma, ThoughtSpot, Coginiti:
Snowflake adds more AI capabilities across engineering, analytics, migration, and enterprise workflows
The common thread: helping teams move from AI features that look useful in a demo to AI tools that can work inside real data environments.
CoCo and CoWork: CoCo supports engineers with code development, migration work, system profiling, and modernization. CoWork gives business users an AI-assisted way to generate insights, create dashboards, and work with persona-specific agents.
Context-aware agents: Cortex Sense, Horizon Context, skills, and agent-ready metadata all point to the same need: AI has to understand enterprise data context, not just query tables.
Enterprise workflow support: Natoma strengthens Snowflake’s MCP capabilities and connections into tools like Slack, Outlook, and Zoom. Datastream brings native streaming into Snowflake, AIM supports AI-assisted migration, and Open Sharing expands cross-platform data access.
The practical implication: AI tools are only as useful as the foundation underneath them. If your metrics are inconsistent, definitions are unclear, or business logic lives in dashboards, spreadsheets, and people’s heads, an AI interface will surface the mess faster than it solves it.
Power BI introduces protection policies during report runtime
The practical implication: This reduces the need to manage sensitive reporting one workplace at a time. The bigger lift is making sure semantic models are labeled correctly, since the policy is only as good as the classification behind it.
Sigma AI columns bring prompt-based enrichment into workbooks
Sigma has introduced AI columns in beta, allowing users to enrich table data with natural language prompts inside a workbook. Users can reference existing columns in a prompt, preview the output on a subset of rows, and apply the result as text or structured JSON. AI columns summarize text, classify records, translate content, and extract details from unstructured fields. The feature works with Snowflake and Databricks connections to generate results (Snowflake’s AI_COMPLETE or Databricks’ ai_query).
The practical implication: This could make common enrichment tasks easier to test inside the analytics workflow, especially for text-heavy data like transcripts, feedback, tickets, and notes. Teams will still need rules for who can create AI columns, how prompts get reviewed, and when test logic is ready to be trusted.
ThoughtSpot's Spotter 3 moves into general availability
ThoughtSpot has made Spotter 3 generally available, adding more context-aware AI analysis to its cloud platform. The release includes data-aware responses, chat history, narrative insights, automatic model selection, and analyst-style reasoning that can plan, check assumptions, and self-correct. ThoughtSpot also added Spotter instructions, which let administrators set organization-level guidance for how Spotter responds. Teams can define defaults for formatting, currency, comparisons, ambiguous questions, and sensitive data guardrails.
The practical implication: AI-powered analytics is becoming less about asking a one-off question and more about shaping how the system should behave across real business scenarios. That means teams must define business terms, model readiness, and usage rules before broader rollout.
Semantic intelligence platform Coginiti puts AI inside the data pipeline
Coginiti, a semantic layer intelligence platform, released LLM Blocks in CoginitiScript, allowing teams to run AI steps directly inside transformation pipelines alongside SQL. Teams can use prompts and structured outputs as part of repeatable pipeline logic, including generating synthetic data, classifying records, analyzing sentiment, and extracting structured details from unstructured text. This builds on Coginiti’s broader AI direction, including Coginiti Guide for AI-assisted BI, report summaries, and visualizations, plus AI agents that help generate semantic layers and transformation code.
The practical implication: AI can move closer to where data work already happens. Instead of wiring up a separate tool after transformation, teams can define the prompt, set the output schema, test the result, and schedule it as part of the pipeline.
.gif?width=562&upscale=true&name=Accelr8%20Subway%20Poster%20(AI%20Video%20Effect).gif)
