All things dAIta with Jerry

18/05/2026

4 levels of AI and Data adoption

Level 0 - You use AI through chat (ChatGPT, Claude, Gemini, Grok) to ask something, basically using it like google. You ask some info and you use that info to do a task. For example, a real estate agent from Cebu will ask Claude to suggest content to attract buyers, then the real estate agent will create content based on Claude's suggestion.

Level 1 - You use AI to make stuff for you, but there is still a lot of human in the loop. For example, a Civil Engineer can ask Grok to create a weekly progress report in Powerpoint, based on the pictures, reported progress from subcons, minutes of meeting etc. The CE will still need to refine its output through iteration

Level 2 - You give AI an end to end task, from research, planning, and creating. For example, an accountant can ask Codex to create invoices for this week. Codex AI then will pull data from a CRM (Salesforce, Hubspot), apply the necessary filters. Structure the data using a data pipeline (python or SQL, no more excel files) to create different API payloads, then upload those data to an accounting software (Quickbooks, Xero, Netsuite) and call an API to create invoices based on uploaded data. After all that, the accountant will still review everything, once done, let AI send all the invoices.

Level 3 - You use multiple AI agents in parallel. For example, a software engineer will create a mobile app using Spec Driven Development principle, then let claude code create the app using the written spec. One agent will start building the infrastructure, one agent will research competitors with similar concept, one agent will start conceptualizing the UI.

19/04/2026

Client: what is the status of the project?

Me: Me and the team are working on it

The team: ...

18/04/2026

Two years ago, I wrote an article called Generative AI in Data Engineering - https://medium.com//generative-ai-in-data-engineering-f915f01fb22b

I mapped how AI could be used across the data engineering lifecycle. Synthetic data generation, AI-assisted schema detection, AI-generated SQL, schema optimization, natural language querying. I wasn't using any of these tools yet. I was just exploring how they fit.

Now, two years later as a data and analytics engineer, a lot of what I wrote about has become part of my actual workflow.

I use Cursor and Claude Code daily, connected to dbt MCP, SnowSQL, and dbt Cloud for orchestration.

I curate AI context inside our BI semantic layer so AI gives consistent answers. Adding context to fields, writing sample queries, defining question patterns.

I deal with governance questions like who owns the model, who verifies the output, how do you label AI-assisted vs manually built reports.

BI and semantic layer changes are moving into GitHub-based workflows so AI agents can participate in the delivery pipeline.

Documentation is being written to serve both humans and AI agents.

Two years ago, I thought AI in data engineering was mostly about the transformation layer. Generate SQL faster. Automate code.

Now the bigger impact for me is in serving and semantics, trust and governance, workflow automation, and knowledge capture.

I wrote a full breakdown on Medium: https://medium.com//ai-in-data-engineering-what-i-wrote-then-vs-what-i-actually-do-now-3ef2a34012c1

I'll probably write another follow-up in two years. Curious what I'll learn next.

Two years ago, I wrote an article called Generative AI in Data Engineering.

14/04/2026

Around 2-3 years ago, - Cebu Python Users Group had a Python book club.

For the first 3 months, we read and discussed 3 books about data engineering and analytics.

The first book was about Modern Data Architecture with Python. It covered data architecture history, then mainly Databricks and PySpark for modern architecture.

The second book was about Analytics Engineering with dbt. It discussed the evolution of the data engineer and analytics engineer, and covered tools like Polars, DuckDB, and mainly dbt.

The third book was about Graph Data Modelling in Python. It discussed graph databases in Neo4j.

At the time, I hadn't used any of these tools in production. They were just interesting topics we were reading about as a community.

Then a few years later, I got hired as a Senior Data Engineer where I used Azure Databricks. Book 1.

Then I got hired as an Analytics Engineer, where I use mainly dbt and Snowflake. Book 2.

And just recently, I implemented a custom data knowledge graph using Neo4j at work. I initially planned to use GitNexus, but found out it was mainly built for code functions and repos, and wasn't really applicable to dbt/SQL-based repositories. So I decided to build my own. Book 3.

All three books. All three became part of my actual career.

I didn't plan it that way. I just showed up to a book club with a community of people who were curious about the same things.

13/04/2026

Everyone is a data analyst.

Before I went into the data/analytics engineering career, I worked as a field engineer at a highrise construction company. I analyzed labor productivity and made decisions based on it. I analyzed actual site conditions and figured out how to sequence construction.

When I was a structural engineer, I analyzed the data needed to build a structural model, applied design loads, then analyzed the output to design structural members and foundations.

When I worked as a real estate agent, I analyzed market data so I could give clients informed advice. I also analyzed my Facebook ad campaigns to know which ones to pour more money into.

When I worked in SEO, I analyzed which keywords, which articles, what to do for the client's website to get good rankings.

When I worked as a technical sales engineer, I analyzed data from PhilGEPS, CRM, and prepared leads to call. Analyzed client requirements to recommend the right structural steel profile.

Then I transitioned to tech.

My first job in tech was a backend developer and web scraper. I scraped data and made it available for the frontend.

Then I became a data engineer, analyzing financial and accounting data for a banking client.

Then a senior data engineer, dealing with finance data and data output from LLMs.

Now as an analytics engineer, I deal more across domains. RevOps, Finance, CRM, GTM, Product Analytics.

The thread is the same every time. I analyzed data and made decisions from it. The tools just changed. Excel became SQL. Site reports became dashboards. Gut feel became dbt models.

I remember asking ChatGPT around a year ago, "What is a data analyst?"

It said something like: "A data analyst collects, processes, and interprets data to help organizations make informed business decisions, typically using tools like Excel, SQL, and Power BI."

So I asked, "Why does it seem like data analyst is more synonymous with business roles? The ones dealing with Excel and Power BI?"

And honestly, fair answer from ChatGPT. The title just got formalized in business first. That's where it stuck.

But the skill was never just a business thing.

Everyone is a data analyst. Some of us just ended up with the title.

10/04/2026

This page is where I share everything about data and AI. How they work together, how businesses can use data to get real results from AI, and the practical side of making it all happen.

Let's discuss anything and everything through the lens of data.