An Experiment - Top Data Trends for 2025 with Coalesce and Google NotebookLLM
AgileData Podcast #55
Join two LLM generated guests as they discuss the Top Data Trends of 2025 Whitepaper published by Coalesce.
This is a different episode. Instead of a human guest, we have two robot guests.
I decided to try and experiment. My experiment was, can I upload a white paper to LLM, have it generate a podcast listen to that podcast in my daily walk and see whether that summary removes the need for me to actually read the white paper.
So in this case, I have grabbed a white paper called Top Data Trends for 2025 from Coalesce, uploaded it to the Google Notebook LLM and got it to generate a podcast with two hosts chatting about the white paper.
Have a listen, let me know what you think.
I'm really keen to understand, do you think this approach is useful, or is it a load of bollocks?
Listen
Listen on all good podcast hosts or over at:
Read
Read the podcast transcript at:
https://agiledata.io/podcast/agiledata-podcast/an-experiment-top-data-trends-for-2025-with-coalesce-and-google-notebooklm/#read
Download
You can download the whitepaper from the Coalesce website here:
https://coalesce.io/reports/the-top-data-trends-for-2025/
Google NoteBookLLM Briefing
(How meta is that, a LLM commenting on a podcast created by the same LLM)
Briefing Document: Top Data Trends for 2025 - Based on Coalesce Report
Introduction:
This briefing document summarises the key data trends for 2025, as discussed in a podcast episode featuring AI-generated hosts ("ADI #1" and "ADI #2") dissecting the "Top Data Trends for 2025" report by Coalesce. The podcast format was used as an experiment to see if an LLM summary could remove the need to read the whitepaper.
Key Themes and Ideas:
The Rise of Knowledge Pipelines:
Traditional data pipelines are evolving into "knowledge pipelines." This isn't just about moving data; it's about enabling AI to reason with the data, understand context, and learn from it.
“It’s not just tables and columns, it’s about AI. Understanding all the relationships and the context. Hidden within the data itself. So think about it. Knowledge pipelines, they teach AI to reason about the data, not just process it."
This is crucial for generative AI, allowing it to learn and make smarter decisions.
Multi-Engine Compute:
The idea of having all data in one place is being challenged. Different compute engines are better suited for different tasks (real-time analytics vs. complex machine learning).
A unified storage layer is essential for a multi-engine approach, allowing for flexibility and the ability to "pick the right tool for the job."
Even big players like Snowflake are embracing more flexible, open source approaches such as Iceberg tables.
"The future is going to be about picking the right tool for the job."
Practical AI Takes Center Stage:
2025 will be about demonstrating real-world ROI and business value with AI, moving past the "hype" of 2024.
"2024 was the year of bold AI experiments, but 2025 is going to be all about ROI and tangible business value."
AI startups that cannot demonstrate real value may struggle.
Combating Industry Amnesia with AI:
There's a tendency in tech to reinvent the wheel, forgetting lessons from the past.
AI agents can help by capturing and analyzing vast amounts of historical data, filling in knowledge gaps.
"It’s like using AI to fight AI induced amnesia. These agents can help us learn from the past…"
The Importance of AI Governance:
AI needs to be used thoughtfully and responsibly, with clear guidelines for development, deployment, and use.
AI governance should align with company values and ensure AI is serving people, not the other way around.
“We need to be really thoughtful about why we’re using AI, and what the potential consequences might be."
Data Quality and Culture are Key:
Data quality isn't just a technical issue, it's a cultural one.
"If data literacy isn’t valued and prioritised throughout your entire organisation, AI initiatives are going to struggle."
If people don't trust the data or understand how to use it, even the most sophisticated AI will fail.
Scaling AI Deployments:
The focus shifts to real-world AI deployments at scale.
This brings challenges in terms of MLOps (managing machine learning models) and AIOps (using AI to automate IT operations).
Data teams become even more important in this environment, acting as a bridge between AI hype and business value.
Open Table Formats and Flexibility:
Open table formats like Iceberg become essential in data lakes, providing flexibility and avoiding vendor lock-in.
It's about having the "freedom to use different tools without having to constantly worry about those compatibility issues."
Unlocking the Potential of Unstructured Data:
AI is changing the game in terms of analyzing unstructured data (audio, video, text documents).
Companies are finding ways to structure this data for new AI-driven insights. For example, an insurance company used AI to transcript customer calls, and then rate them, converting that data into structured data.
"Companies are now finding ways to structure this previously untapped data. Opening up these incredible new possibilities for AI driven insights."
Platform Gravity:
Platforms like Snowflake are becoming "centers of gravity" not just for data, but for applications, AI, and decision-making.
This might create vendor lock-in issues, but also makes it simpler to have data and tools in one place for efficiency.
Data as a Product:
Companies are starting to treat data as a product, focusing on quality, reusability, and alignment with business needs.
This leads to companies building purpose-built data platforms and potentially moving away from large, expensive systems.
“People want trust. Building something reliable and actionable. That you have confidence in means providing that transparency for all users.”
Semi-Structured Data Revolution:
It's estimated that 90% of the world's data is semi-structured which highlights the huge potential to be unlocked.
Businesses will need to rethink their practices around storage, security and governance to handle this influx of data.
Combining structured data with semi-structured data can "unlock some seriously game changing insights."
For example, combining call center recordings with customer sales data to understand how agent empathy impacts retention, or combining satellite images with insurance claims.
The Human Element is Key:
Technology is just one piece of the puzzle.
Businesses need to combine technical advancements with a shift in mindset about data.
There's a need to close the gap between IT and the business side. It is not about the technology that holds a business back but rather the people, processes and culture.
Data literacy needs to be embraced organisation wide, so people understand how to interpret data and turn it into action.
"Technology is never the reason why a business can’t succeed or transform. It always comes back to people."
AI governance is essential, considering the ethical implications of AI from the very beginning, being transparent about how systems work, and being inclusive and diverse in the development process.
Data teams are evolving into product teams, requiring new skills and a customer-focused approach.
"Data teams need to become more collaborative, more customer focused, and more agile in their approach."
There's a need to increase awareness about the ethical implications of AI, with honest conversations, and diverse teams who can help prevent bias.
Strategic Business Implications:
Agility and adaptability are crucial for businesses to thrive in the rapidly changing landscape.
Cloud-based data platforms provide the flexibility and scalability needed to respond to changing conditions and capitalize on opportunities.
A customer-centric approach to data is essential, with a focus on creating better customer experiences.
Data privacy and transparency builds trust and can be a differentiator in a competitive landscape.
Companies need a holistic data strategy that encompasses all aspects of data from collection to action and should be aligned with the overall business strategy, not in a silo.
Investing in strong data teams with the right skills is crucial, and fostering a culture that values data and provides opportunities for growth and development is essential.
"Data as a strategic asset. Not just a technical afterthought."
Conclusion:
The podcast highlights that 2025 is set to be a pivotal year for data and AI. Companies need to move away from traditional data practices and embrace new concepts, such as knowledge pipelines, multi-engine compute, and treating data as a product. It is just as important to consider the human side, and build a strong culture of data literacy and governance. A holistic data strategy, aligned with overall business goals, will enable businesses to unlock value, drive innovation, and stay ahead of the game.