• Everyday AI
  • Posts
  • How a Transformative Data Strategy Powers AI Success

How a Transformative Data Strategy Powers AI Success

Apple speeds up chip design with GenAI, Microsoft’s major job cuts for AI, IBM unveils AI security and governance software and more!

Author

Everyday AI
June 19, 2025

In Partnership With

Meet Gemini, Your Personal AI Assistant

Check out Veo 3, Google’s state of the art AI video generation model in the Gemini app, which lets you create high quality, 8-second videos with native audio generation.

Try it with the Google AI Pro plan, or get the highest access with the Ultra Plan. Sign up at Gemini.Google to get started and show us what you create.

Outsmart The Future

Today in Everyday AI
6 minute read

🎙 Daily Podcast Episode: An expert from Deloitte explains why data is the ultimate differentiator in digital AI transformation. We dive deep into data marketplaces, agentic AI, and the power of structured and unstructured data. Give it a listen.

🕵️‍♂️ Fresh Finds: Humain AI forming new ads and gaming divisions, Meta hires new AI investors and AI avatars that made $7M in 7 hours. Read on for Fresh Finds.

🗞 Byte Sized Daily AI News: Apple speeds up chip design with GenAI, Microsoft’s major job cuts for AI and IBM unveils unified AI security and governance software. For that and more, read on for Byte Sized News.

🧠 Learn & Leveraging AI: We break down how you can fuel your company’s AI success with the right data strategy. Keep reading for that!

↩️ Don’t miss out: Did you miss our last newsletter? We talked about Google Search Live joining AI Mode, OpenAI cutting ties with Scale AI, Amazon’s CEO speaking on AI shrinking the workforce and more. Check it here!

 How a Transformative Data Strategy Powers AI Success 📊

You think using AI is your moat? Nope.

Just using LLMs isn't enough to power your company's AI success. But do you know the real fuel?

Having your data right is the ACTUAL key. So how do you do it? And how does your company's data strategy change with agentic AI?

Also on the pod today:

• Multi-Agent Orchestration Challenges 👥
• Structured vs. Unstructured Data in AI 🤔
• Synthetic Data and AI Transformation 🛠

It’ll be worth your 28 minutes:

Listen on our site:

Click to listen

Subscribe and listen on your favorite podcast platform

Listen on:

Here’s our favorite AI finds from across the web:

New AI Tool Spotlight – Firebase is Google’s megaviral AI coder, Crono is AI-powered outreach at scale, Voicenotes uses AI to turn your voice into content and publish it directly.

AI Startups – Saudi Arabia’s AI firm Humain, is forming a division on ads and gaming.

Scale AI’s CEO is reassuring employees and customers that the company isn’t “winding down.”

Meta – Meta is hiring AI investors Nat Friedman and Daniel Gross and partially buy out their venture capital fund, NFDG.

Trending in AI – These AI influencer avatars raked in $7 million in 7 hours.

AI Robotics – This four-legged robot plays Badminton with you using AI.

AI Chips - If you haven’t been following what’s been happening with AI chips in 2025 here’s a timeline of the semiconductor market so far.

AI Video - HeyGen has released a new product placement feature.

1. Apple Eyes Generative AI to Speed Up Chip Design ⚡️

Apple’s senior hardware exec Johny Srouji revealed the tech giant is exploring generative AI to accelerate the design of its custom chips, a move that could transform how hardware innovation happens. Speaking in Belgium, Srouji emphasized how Apple’s all-in bet on its own silicon, starting with the A4 chip in 2010, paid off by giving it full control over performance and efficiency.

Leveraging AI-enhanced electronic design automation tools from leaders like Cadence and Synopsys promises to boost productivity and shorten development cycles.

2. Microsoft to Slash Thousands of Sales Jobs Amid AI Push 😲️

Microsoft is set to announce thousands of job cuts, predominantly in its sales division, as it reallocates resources toward expanding its artificial intelligence capabilities, Bloomberg News reports. This follows a previous round of layoffs in May affecting 6,000 employees, signaling a clear shift in priorities toward AI-driven growth.

With an $80 billion capital expenditure planned mainly for data center expansion, Microsoft aims to overcome AI service capacity bottlenecks and maintain its competitive edge.

3. IBM Unveils First Unified AI Security and Governance Software 🔐

IBM just announced a breakthrough in managing AI risks by combining its Guardium AI Security and watsonx.governance into the first unified platform to secure and govern AI agents at scale. This new integration not only helps businesses comply with global AI regulations like the EU AI Act but also automates red teaming and detects hidden AI threats across cloud and code environments.

With enhanced lifecycle monitoring and compliance accelerators, companies can now better control and audit their AI systems while reducing vulnerabilities.

4. OpenAI’s AI in Biology: Accelerating Science While Preventing Biothreats 🔬️

According to OpenAI’s latest update, their advanced AI models are transforming biology by speeding up drug discovery, vaccine design, and enzyme development, while simultaneously addressing serious biosecurity risks. Their multi-layered safety measures involve expert collaboration, continuous monitoring, and intensive red teaming to prevent misuse, especially in bioweapon creation.

This forward-thinking approach aims to unlock medical breakthroughs without waiting for a crisis, marking a new chapter in responsible AI use in life sciences.

5. YouTube Shorts to Get Smarter AI Video Generation This Summer 🤳

YouTube announced at Cannes Lions that its latest AI video model, Veo 3, will roll out to Shorts creators later this summer, promising better video quality and audio integration. Currently, Shorts users can access Veo 2 for background generation, but Veo 3’s upgrade could significantly enhance creative options for short-form content.

It’s still unclear if the new AI features will remain free or require a paid Google AI subscription, which may impact accessibility for casual creators.

6. Midjourney Launches V1 AI Video Model 🎥

Midjourney has unveiled V1, its image-to-video AI model that creates four five-second videos from a single image, accessible exclusively via Discord at launch. This move places Midjourney in direct competition with established AI video generators like OpenAI’s Sora and Adobe’s Firefly, signaling a push beyond static images into dynamic content.

Despite the excitement, the company faces ongoing legal challenges from Disney and Universal over copyright concerns, underscoring the tension between AI innovation and intellectual property.

🦾How You Can Leverage:

Wanna up your chance at AI success? Get serious about data. 

Case in point? Consulting juggernaut Deloitte. 

Their US Chief Data and Analytics Officer, Ashish Verma, told us Deloitte runs a marketplace with 520 different data feeds. 

Most companies can't even catalog their internal spreadsheets properly, yet they’re still scrambling to implement multi-agentic AI swarms and whatever LLM-enfused buzzword of 2026 that’s about to pop off. 

Nahhh, shorties. That’s backwards. 

Here's what's really happening: while executives are busy buying the latest AI tools, their competitors are quietly building data infrastructures that make those same tools 10x more powerful.

The brutal truth?

The ceiling for your AI success is predetermined by your foundational data strategy.

Sorry…  "big data" ain’t a Google Sheet file with 87,000 rows.

The enterprises doing it right? 

Companies like Deloitte are spending HUNDREDS of millions on clean data sources because they figured AI is kinda worthless without it. 

So on today's show, we got the Data Strategy 101 from Ashish, a seasoned AI vet who’s helping propel the world’s largest company’s AI strategies forward. 

Listen up. 

Don’t miss this. 👇

1 – Your Internal Data Ain’t Cutting It 📊

That cozy little data warehouse you've been babying for years?

Cute.

Also completely useless for serious AI deployment.

Ashish dropped some knowledge about modern AI requirements: you need your internal data PLUS second-party data PLUS third-party data PLUS synthetic data to make anything meaningful happen.

The companies actually winning at AI?

They've built data marketplaces that work like Amazon for information.

Picture this: one central hub where 178,000 employees can discover, access, and combine multiple data sources without some poor IT person playing middleman and slowing everything down.

Ashish brought up an important point.

Their data marketplace is "contextual" - meaning it learns from user behavior to suggest relevant datasets before you even know you need them.

No more playing guessing games where you gotta know exactly what you're hunting for.

Welcome to AI that actually helps instead of frustrates.

Try This: 

Time for some brutal honesty about your data situation.

Map every single data source your company uses right now.

Yes, including that random Access database your accounting team refuses to abandon because "it just works."

Create four columns: Internal, Partner, Third-Party, and Synthetic.

Be real about what percentage falls into each bucket.

If 80% of your data is internal, your AI dreams are probably way too small for your actual business problems.

Next step?

Identify the three external data sources that could 10x your AI capabilities.

This might be industry datasets, government databases, or partner information you're currently ignoring.

Calculate the cost of acquiring these datasets versus the potential business impact.

Most executives get their minds blown when they discover a $50K annual data subscription could unlock MILLIONS in AI-driven insights.

Do the math.

2 – Unstructured Data is Where The Magic Happens 🗂️

Those PowerPoints and Word docs everyone's been ignoring?

Yeah, those are actually your secret competitive weapon.

Most organizations are literally sitting on goldmines of unstructured data hiding in documents, emails, and presentations that could transform their AI systems if they just processed and indexed them properly.

Ashish shared how Deloitte completely transformed their resource management game.

Get this: they contextualized and indexed their ENTIRE resume database.

We're talking 455,000 unstructured documents.

Previously?

Searchable only through basic keyword matching that sucked.

Now their system matches role descriptions to candidates in near real-time by actually understanding context, not just hunting for keywords.

The resource manager went from manually digging through resumes one by painful one to getting ranked, relevant results instantly.

That's the difference between AI that works and AI that makes everyone wanna quit.

Try This: 

Right now, identify your three largest repositories of unstructured data.

This is probably your shared drives, email archives, customer service tickets, or meeting transcripts.

Pick the repository that your team accesses most and causes the biggest headaches when searching.

Start small with a pilot of 100-500 files from this repository.

Use a tool like Azure Cognitive Search, Elasticsearch, or OpenAI's embedding API to create semantic indexes of this content.

Here's the fun part: run side-by-side tests.

Search for the same business-relevant query using your current keyword search versus the new contextual search.

Document the time savings and accuracy improvements like your promotion depends on it.

When you can show leadership that finding critical information went from 20 minutes of frustrated searching to 30 seconds of precise results, you'll have their attention AND their budget approval for scaling this approach.

Make the case impossible to ignore.

3 – Build Data Strategy For Tomorrow’s Problems 🔮

Excellent question.

Why are most data strategies failing spectacularly?

Because they're designed for yesterday's reporting dashboards, not tomorrow's autonomous agents.

Ashish emphasized the critical insight about walking with the end in mind - your data strategy should anticipate agentic AI capabilities, not react to immediate use cases.

Companies that wait until they need multi-agent orchestration to fix their data labeling and annotation processes will find themselves completely outmaneuvered by competitors who prepared their data infrastructure for probabilistic reasoning months earlier.

The annotation and labeling requirements for agent-driven systems are exponentially more complex than traditional analytics, especially when agents start working with external data sources beyond organizational boundaries.

Try This: 

Map out your AI roadmap for the next 18 months, then reverse-engineer what data labeling, annotation, and cataloging capabilities you'll need.

Start building those processes now, before you need them for agent deployment.

Reply

or to participate.