Databricks Genie: What Natural Language Querying Is Good
How Databricks Genie turns plain English into governed SQL, where text-to-SQL pays off, and what to fix in your data before you switch it on.
Blog · Field notes
What I learn running search and AI builds for real companies: how AI search actually changes traffic, what earns a citation, and the process discipline behind it.
How Databricks Genie turns plain English into governed SQL, where text-to-SQL pays off, and what to fix in your data before you switch it on.
Databricks vs Snowflake for a mid-sized business: lakehouse vs warehouse, GenAI and RAG support, cost traps, and how to pick the right platform in 2026.
Five AI visibility tools compared on what they actually track, published pricing, and fit, plus eight ways AI search measurement quietly breaks down.
How logistics and fleet software vendors get named in AI answers: publish real specs, map the process, and win regulated vertical queries.
Industrial buyers ask AI first. See how manufacturers structure specs, schema, and entity data to win AI Overviews and LLM supplier shortlists.
Eight questions to ask an AI consultant before you hire, plus the red-flag answers that signal a failed build: data readiness, security, ROI, handoff.
AI summaries cut clicks to 8%. Compare solo consultant vs agency SEO on execution, GEO citation tracking and data risk, then pick your partner.
AEO consultant or AI SEO agency? Compare focus, engagement models and fit, then use three vendor questions to pick the right partner for AI search.
How to choose an answer engine optimization agency: entity depth, zero-click measurement, GEO capability and proof of revenue lift, not impression decks.
What drives Answer Engine Optimization cost in 2026: data structuring, schema, and provider comparison, plus how to scope AEO before you buy.
Most AEO advice assumes a specialist and a budget. This is the version for a team of one, ordered by what actually returns the most first.
Model tokens are the small number. Here is where the money actually goes in an AI content pipeline, and the guardrails that keep it predictable.
SEO fails more often on process than on tactics. DMAIC is the discipline that turns a list of activities into a programme you can actually defend.
Traffic loss usually arrives as a slow slide across many pages, not a cliff. Here is how to catch it in Search Console while it is still cheap to fix.
Most schema advice is a list of every available type. Only a few genuinely change how machines understand your business. Here are those, and why.
You cannot improve what you have never measured. Here is a free, repeatable way to find out whether assistants name your company, and how that changes.
Being quoted is mostly a structural problem. The same information, arranged differently, is the difference between a citation and a skipped page.
Position three still says position three, and the clicks are gone anyway. That pattern has a name, a diagnosis you can run today, and a response.
llms.txt is a proposed standard for telling AI systems which parts of your site matter. Here is what it does, what it does not, and whether to bother.
Assistants name a small set of sources per topic. Getting into that set is a repeatable process, not luck. Here is the sequence that works.
GEO is the practice of making your content the source a generative model reaches for. Here is the definition without the hype, and what it asks you to change.
Answer engine optimization is not a replacement for SEO. It is what happens to SEO when a machine reads your page before a human ever sees it.