- Everyday AI
- Posts
- Ep 674: 2025 AI Roadmap Rewind. Part 1 of 2. From AI Agent Orchestrators to Deepfakes. The New Tech Order
Ep 674: 2025 AI Roadmap Rewind. Part 1 of 2. From AI Agent Orchestrators to Deepfakes. The New Tech Order
A look back at our 2025 AI predictions, OpenAI releases its impressive new AI image model, Google goes global with AI translation in your ears, BBVA rolls out ChatGPT to 120K staff and more
Everyday AI
December 16, 2025
👉 Subscribe Here | 🗣 Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI
Sup y’all 👋
Part 1 of our AI Roadmap Rewind is live. Make sure you check it out, as it kinda serves as a (partial) State of AI.
On today’s show, we talked that Part 2 would drop tomorrow. Buuuuuut, I’m feeling a bit under the weather again, so we might have to push Part 2 back to later this week.
✌️
Jordan
Outsmart The Future
Today in Everyday AI
8 minute read
🎙 Daily Podcast Episode: Back in January, we laid out 25 bold AI predictions. Now, we’re hitting rewind and checking back. Today is part 1 of 2. Give today’s show a watch/read/listen.
🕵️♂️ Fresh Finds: Google’s CC agent reads your emails, FrontierScience benchmark introduced, GPT-Image 1.5 takes LMArena crown and more. Read on for Fresh Finds.
🗞 Byte Sized Daily AI News: OpenAI releases its impressive new AI image model, Google goes global with AI translation in your ears, BBVA rolls out ChatGPT to 120K staff and more. Read on for Byte Sized News.
💪 Leverage AI: Did our predictions in January help or hurt? And how did we do? Keep reading for that!
↩️ Don’t miss out: Miss our last newsletter? We covered: Disney invests $1B in OpenAI while threatening Google, Trump blocks states from regulating AI, Meta may close its open-source doors, Google expands agentic Deep Research, and more. Check it here!
Ep 674: 2025 AI Roadmap Rewind: From AI Agent Orchestrators to Deepfakes. The New Tech Order
We don't do AI predictions lightly.
Everyday AI is trusted by millions each year to help guide them through the muddy AI waters.
So in the same way you'd want transparency out of your AI models, we're rolling back the clock on our January 2025 AI predictions we dished with our Roadmap Review.
We're busting out the receipts.
At the time, these AI predictions seemed nutty. So, did we lead you astray? Or, did we pave the road to paydirt?
Let's dive in y'all.
Also on the pod today:
• AI agents applying for jobs?! 🤖
• Consulting pricing crisis incoming 💸
• API prices dropping like crazy 📉
It’ll be worth your 43 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 – Readever lets you read books with your AI mentor, Okara is private AI chat with 30+ open models, Varchive shows off AI-assisted builds and projects
AI and Email — Google Labs is rolling out CC, a Gmail agent that briefs you daily.
AI Agents — Think more AI agents means better results? Not always. Google’s latest study reveals the surprising situations where one agent outperforms a whole team.
AI and Science — OpenAI just launched FrontierScience, a tough new benchmark built by experts to see how well AI models handle real science problems.
Google Deep Research — Google announced a nice upgrade to Deep Research, as it will now include visuals.
AI Models — The Allen Institute for AI just dropped Molmo 2, an open video model that can answer detailed questions and even track objects across clips. How does it stack up?
AI Images — GPT Image 1.5 has taken the top spot on LMArena
1. OpenAI Unveils Faster, Smarter Image Model in ChatGPT 📸
OpenAI has launched GPT-Image 1.5, bringing speedier and more precise image creation tools to ChatGPT users worldwide just as the AI image race heats up in late 2025. The new model features a dedicated creative studio for editing, better text generation in images, and richer style options, marking a major leap from the earlier chat-based system.
This release follows hot on the heels of Google's latest image model update and comes amid rising copyright battles and big-name deals, including a fresh partnership with Disney.
2. Google Unveils Two Smart Glasses Projects with Samsung 😎
Google just pulled back the curtain on its AI smart glasses lineup, revealing two distinct products in partnership with Samsung, Gentle Monster, and Warby Parker. The first, "AI glasses," are audio-focused and let you chat with Gemini or snap photos, hitting the market next year.
The second, "display AI glasses," offer a private in-lens display for personal info like navigation and captions, but their launch date remains under wraps. Both devices run on the new Android XR system, marking Google’s latest push into wearables as the race for AI-driven gadgets heats up.
3. Mozilla’s New CEO Pledges “Modern AI Browser” Revolution 🦊
Mozilla has named Anthony Enzor-DeMeo as its new CEO, and he’s making a bold push to transform Firefox into a privacy-focused “AI browser” with clear user controls and transparency. This marks a significant shift just as AI features flood the browser market, promising users more power over their data amid growing privacy concerns.
Firefox's market share has slipped dramatically, but Mozilla is betting on AI-driven innovation and new products to win back trust and diversify its revenue. As Chrome and rivals race to pack in their own AI tools, Mozilla is fighting to stand out by prioritizing user choice and control.
4. Google Translate Gets Gemini AI Power-Up and Live Headphone Translations 🗣️
Google is rolling out a major AI upgrade to Translate, tapping its Gemini AI to finally crack idioms, slang, and nuanced phrases for more natural results across nearly 20 languages. Starting today in the US and India, users on Android, iOS, and the web will see smarter translations that actually get the meaning right instead of just word-for-word conversions.
Meanwhile, Translate for Android is debuting live, real-time headphone translations that preserve tone and cadence, making conversations and foreign content easier to follow. Google says this beta feature works with any headphones and over 70 languages, with global expansion and iOS support slated for 2026.
5. George Osborne Tapped to Lead OpenAI’s Global Government Push 🌏
OpenAI has just hired former UK chancellor George Osborne to spearhead its government partnerships worldwide, signaling a major push to get AI adopted at the national level.
Osborne will lead the new "OpenAI for Countries" division from London, forging big-ticket deals with governments as tech giants race to shape the future of public-sector AI. With heavyweights like Microsoft and Google already courting national leaders, this move shows OpenAI is betting that global alliances will be key in the AI arms race.
6. BBVA Rolls Out ChatGPT Enterprise to 120,000 Staff in Landmark AI Move 🏦
Today, BBVA announced it will deploy ChatGPT Enterprise to more than 120,000 employees worldwide, making it one of the largest corporate AI rollouts to date. The Spanish banking giant is moving beyond pilot programs and embedding generative AI directly into everyday workflows across the company, aiming to boost productivity and streamline core operations.
BBVA’s leadership says the shift positions the bank at the forefront of the industry’s AI adoption wave, with early results already showing significant time savings for staff.
A LOT can change in a year.
That’s because in AI time, a year is like a half lifetime.
Back in January, we released 25 kinda wild predictions for 2025. (You can go watch/listen to them here.)
At the time, these 25 predictions sounded aggressive, maybe even impossible. But looking back 11 months later, the scariest part isn't just that we were mostly right.
It is how fast the landscape shifted beneath our feet.
Wanna age ourselves? When we released these predictions, Gemini 2 had JUST been released.
So, we’re holding ourselves accountable today. We are rewinding the tape to see if we led you astray or handed you the AI roadmap before the road was built.
Let’s breakdown predictions 25 through 13 in Pt 1 of our 2 Part series.
(Part 2 may or may not be released tomorrow, depending on if Jordan feels less sick. Lol)
Prediction #25: Agent orchestrators as a growing position 🏗️
The Prediction: We said companies would create specific roles for humans to manage fleets of AI agents.
The Receipt: It is now happening at the biggest companies in the world. Walmart and KPMG are officially hiring for "AI Agent Architects," while Adobe is recruiting "AOs" to manage their digital workforce.
The Verdict: We absolutely nailed this one.
Prediction #24: Job postings for AI agents 🤖
The Prediction: We said companies would post jobs specifically for AI agents to fill, not humans.
The Receipt: AWS and Klarna have reportedly reclassified agent compute costs as "digital full-time employees." A startup called FireCrawl even went viral for posting a job application that only AI agents could apply for.
The Verdict: This is becoming standard practice.
Prediction #23: Reasoning data collection focus 🧠
The Prediction: We said companies would rush to collect data specifically to support and train reasoning models.
The Receipt: OpenAI's latest report reveals that enterprise reasoning token consumption increased 320-fold in twelve months. Red Hat is now explicitly helping enterprises generate synthetic data to feed private reasoning engines.
The Verdict: We were mostly right, yet most companies are still behind.
Prediction #22: Professional services pricing crisis 📉
The Prediction: We warned that firms like law and consulting would face a pricing crisis as AI destroyed the billable hour.
The Receipt: S4 Capital and Thomson Reuters confirmed a shift to "output-based pricing." McKinsey reports call this a "fundamental pricing model disruption" that offers forty percent cost reductions.
The Verdict: This is happening right now.
Prediction #21: UBI becomes a mainstream topic 💵
The Prediction: We said Universal Basic Income would become a kitchen table topic due to AI job displacement.
The Receipt: A UBI bill was actually introduced in the US House citing AI displacement. Locally, Cook County made their guaranteed income program permanent with a $7.5 million budget.
The Verdict: It is officially on the political table, but we were probably off on this one. (Maybe next year?)
Prediction #20: Open source models surge 🌊
The Prediction: We said open-source models would temporarily overtake proprietary giants.
The Receipt: DeepSeek's R1 model shocked the world in January by beating GPT-4o on major benchmarks. Today, open-source models consistently hold three to four spots in the top ten rankings.
The Verdict: Pretty spot on.
Prediction #19: Chinese AI dominance and confusion 🇨🇳
The Prediction: We said Chinese AI would dominate usage and cause global confusion.
The Receipt: Reports show that 30% of global AI usage is now Chinese open-source models. The DeepSeek market cap shock in January proved that Chinese efficiency is forcing the US market to react.
The Verdict: This is a massive geopolitical reality and the prediction came to fruition.
Prediction #18: Perplexity's pivot or acquisition 🔄
The Prediction: We said Perplexity would either pivot, get acquired, or get crushed.
The Receipt: They pivoted hard. Perplexity launched a shopping platform, a finance product, and an agentic browser to diversify away from just being an "answer engine."
The Verdict: They survived by changing the game. Pivot it was.
Prediction #17: Dramatic API price drops 💸
The Prediction: We said API prices for top models would drop below 50 cents per million tokens.
The Receipt: DeepSeek v3 is now clocking in at two cents per million tokens when cached. That is twenty-five times cheaper than our "bold" prediction. Uncached, we still have multiple Top 10 models coming in below the 50 cent per million token mark.
The Verdict: We crushed this one.
Prediction #16: Embodied AI sector explosion 🦾
The Prediction: We said robotics, drones, and physical AI would see massive capital injections.
The Receipt: VC funding for robotics and drones soared over 40% this year. Figure AI raised a billion dollars alone to put humanoid robots into the workforce.
The Verdict: We were spot on.
Prediction #15: One-shot 5-minute HD video 🎥
The Prediction: We said you would be able to generate a five-minute HD video with a single prompt.
The Receipt: Kling AI now allows up to three minutes of video generation via extension features. Disney also invested a billion dollars in OpenAI to bring AI video to streaming.
The Verdict: We just barely missed the five-minute mark, but we wouldn’t be surprised if this happened in a few months.
Prediction #14: Traditional Internet in question ☠️
The Prediction: We said the traditional ad-supported internet model would come into question as AI abstracts content.
The Receipt: NPR declared AI an "extinction-level event" for publishers. With 74% of new web content now estimated to be AI-generated, the "human" internet is shrinking fast.
The Verdict: The old internet is officially on life support.
The Prediction: We said social media would accelerate misinformation and impersonation problems.
The Receipt: CEO fraud now targets 400 companies daily. An AI imposter even successfully mimicked Senator Marco Rubio to fool foreign officials.
The Verdict: This was the unfortunate truth.
Make sure to join us for Part 2, either tomorrow or maybe Thursday!
Reply