AI agents were just the start. Parallel agents are next.

Hi Futurist,

We need a break. The past two weeks have been nearly impossible to keep up with. Two new frontier models. Four new image models. A new video model. And then a flood of updates that show AI is levelling up again. Six weeks ago, I wrote about the upcoming models from Anthropic, OpenAI and Meta. And here we are. Meta has launched the best open-source model to date, while Google has released the most powerful AI model yet. Therefore, I am sending you insights, inspiration, and innovation straight to your inbox. Let’s dive into the depths of the digital future together and discover the waves of change shaping our industry.

💡 In this post, we're dipping in:

• 📣 Byte-Sized Breakthroughs: We dip into the future and look at what comes after AI agents. We also take a look at Google Gemini’s newest AI model and discuss OpenAI’s 4o Image Generation model.

• 🎙️ MarTech Maestros: 71% of companies use Gen AI, yet only a few see real returns. Discover how top performers turn hype into value, and why leadership makes all the difference.

• 🧐 In Case You Missed It: From Zapier’s assistant commanding 7,000+ apps to Microsoft and OpenAI launching powerful copilots and agent tools across security, search, and productivity. Visual and media generation also leap forward with Midjourney V7, Ideogram 3.0, and Runway Gen-4 setting new creative standards.

Do you have tips, feedback or ideas? Or just want to give your opinion? Feel free to share it at the bottom of this dip. That's what I'm looking for.

No time to read? Listen to this episode of Digital Dips on Spotify and stay updated while you’re on the move. The link to the podcast is only available to subscribers. If you haven’t subscribe already, I recommend to do so.

Quick highlights of the latest technological developments.

Headstory: AI agents were just the start. Parallel agents are next.

In the last Dip, I wrote how AI Agent capabilities now double every 7 months. What used to take a developer an hour, AI now does in minutes. With last week’s releases, the next evolution is clear. What comes next after AI Agents? Parallel agents. Not just one AI doing one task. We’re talking teams of agents running tasks simultaneously, collaborating in real time. Parallel agents are set to evolve into AI heads of departments. The future of AI isn’t about more outputs. It’s about business outcomes.

And here’s what’s wild. Companies are already experimenting with AI heads of departments. Really. Each one managing 5 to 7 sub-agents. Picture a chief marketing agent overseeing brand, content, paid, SEO, and social, each run by their own sub-agent. Employees interact through a simple chat or voice command. It’s like giving your team a direct Slack line to an always-on, always-learning marketing crew that never sleeps.

This isn’t a far-off dream. AI is rewriting the rules of marketing. Planning? Agents now build the calendar, recommend content, pick channels. Testing? Hundreds of ad variants, optimized in real-time. Targeting? Forget segments. Agents find hidden patterns you’d never see. Content? Created, revised, published. Reporting? Autonomous dashboards that update and adjust. Budgeting? Spend flows to what works, minute by minute. Competition? Tracked and countered before your team even hits Monday morning.

And we’re just getting started. In 12 to 18 months, marketing as we know it fractures into something faster. Something sharper. Something relentless. You don’t execute. You orchestrate. You decide the story. The tone. The truth. AI handles the rest. A network of AI agents - interconnected, contextual, always-on - working together toward shared goals. Think “get 50,000 organic visitors a month,” and the system gets to work. A manager agent oversee the flow. Sub-agents execute. Each trained by human experts, fine-tuned to your brand. By then, what once needed 50 people will be handled by just five. What once needed a full hire is just… an agent away.

So, why is it crucial to prepare now? Well, these systems aren’t only getting smarter. They’re getting cheaper too. They do the work at a fraction of the cost of a human. They don’t log off. They don’t sleep. They don’t burn out. You can guess the outcome when others are moving forward and you’re still denying the shift. You need to invest in orchestration, in data, in agents. You don’t need to rebuild everything today. But you do need to rethink what work looks like in the next 6–12 months.

Here’s what you should ask yourself, right now:

• What are the key workflows in your company, and how can they be agentified?

• Where are your data silos? Can agents access and learn from them?

• Who in your team can think in systems, not just skills?

• How do you train leadership to delegate to AI?

• What outcomes do you want your marketing to drive, and how do you brief that to an agent system?

And here’s the bigger truth: this doesn’t stop at marketing. Any knowledge job - product, customer support, legal, HR, finance - will follow the same pattern. A lot of knowledge workers are going to be blindsided. Because the future of AI isn’t chatting with a bot to get a cool image. The future is fully autonomous systems. The ones that run, optimize, adapt… without waiting for you. Reimagine your organization as a system where humans bring strategy, story, and direction, and AI delivers the rest.

Gemini 2.5 Pro claims the AI crown

TL;DR

Google’s new Gemini 2.5 Pro model is a beast. With a 1 million token context window, real-time stream support, and full multimodal capabilities, it’s pushing boundaries. Knowledge is now fresh up to January 2025, and it crushed the Arena leaderboard with the biggest jump ever, taking the top spot across all categories, especially in math, code, and creative tasks.

Read it yourself?

Sentiment

Online, people are raving. Especially about its coding strength. Nobody’s even blinking at the fact it just posted the best score ever in the Arena leaderboard. That 1 million context window is also getting love, though some developers are already calling it too small for really heavy-duty workflows. Of course they are.

My thoughts

Gemini 2.5 Pro is now the smartest model on paper—an offline IQ test gave it nearly 120. High-average human level. And the pace? Unreal. Just last week Claude Sonnet 3.7 was the best at coding, now it’s Gemini 2.5. Grok 3 was topping the charts, now it’s Gemini 2.5. Meanwhile, Altman’s already teasing new OpenAI models. I’d bet o3-pro will hit 130 IQ, and o4 or GPT-5 might crack genius level. We’re in a full-blown AI arms race. And DeepSeek’s new model? Coming anytime soon.

OpenAI introduces GPT-4o's advanced image generation capabilities

TL;DR

GPT-4o now features integrated image generation, enabling users to create precise, photorealistic images directly within ChatGPT. Whether you need a logo, infographic, comic strip or photorealistic scene, GPT-4o handles it all with precision. This enhancement allows for seamless interaction between text and visuals, expanding creative possibilities.

Read it yourself?

Sentiment

The online response was unmistakable. Social media feeds were flooded with AI-generated images produced by the new 4o Image model. From puppets to high-quality advertisements, users explored the model's potential to the fullest. New workflows emerged as individuals shared tips and techniques, such as creating an image with GPT-4o, animating it via Sora, resulting in dynamic videos or advertisements. Simply brilliant.

My thoughts

The advent of GPT-4o's image generation marks a pivotal shift in the creative landscape. With the ability to produce high-quality visuals becoming more accessible, the emphasis now shifts from execution to ideation. In an era where execution is democratized, the value lies in original concepts, strategic insights, and unique perspectives. Companies are reallocating resources from production to ideation, exploring a broader range of creative directions due to reduced costs of experimentation. While AI can generate aesthetically pleasing designs, the human touch remains irreplaceable in fostering authentic relationships, developing innovative products, and offering perspectives that resonate on an emotional level. The true advantage now lies in discerning which ideas are worth pursuing, as execution becomes increasingly accessible to all.

More byte-sized breakthroughs:

• Meta drops Llama 4: Behemoth, Maverick & Scout
  Maverick and Scout are currently available for download, while Behemoth remains in preview. Scout stands out with its ability to operate on a single GPU while offering an unprecedented 10-million-token context window, surpassing previous models like DeepSeek. These models are designed to excel in multimodal applications, supporting text, images, video, and audio inputs. Serving costs range from $0.19 to $0.49 per million tokens.

• ​Microsoft's Copilot evolves into your personal AI companion
  Microsoft expands Copilot into a full AI workspace with powerful new tools. Copilot Pages gives you an AI-powered canvas to write, brainstorm, and collaborate using text, images, and video. Deep Research acts like your personal analyst, delivering rich reports from one question. Copilot Actions completes tasks in your browser. And with Copilot Podcasts, you can generate custom podcast episodes with two virtual hosts in minutes.

• Lindy AI unleashes swarms of parallel-working AI agents
  Lindy’s new Agent Swarms let your AI duplicate itself to handle hundreds of tasks at once. Think research, outreach, and content creation in parallel.
  Lindy can run through lists like leads in Sheets and generate personalized emails simultaneously. It’s like having an army of AI researchers working side-by-side, at lightning speed. Thanks to 5,000+ integrations and 4,000+ scrapers, Lindy connects to nearly any app or website.

A must-see webinar, podcast, or article that’s too good to miss.

Gen AI is in. But where’s the ROI?

71% of companies use Gen AI, but only a few see profit in return. Here’s how the top performers turn use into value. Leading firms are redesigning how they work, embedding Gen AI deep into their operations, and putting the C-suite in charge. But most organizations are still far from seeing real returns. This McKinsey report shows what early winners are doing differently, and why executives can’t afford to sit back. If you’re a C-level executive, manager, or professional leading AI efforts within your organization, this is a must-read to understand what to do next.

A roundup of updates that are too cheesy to ignore.

• Zapier MCP lets your AI assistant understand and execute natural language commands across 7,000+ apps and 30,000+ actions.

• Proxy AI's new Parallel Agents are live, dividing tasks into subtasks and completing them in seconds as part of DeepWork.

• Perplexity plans to rebuild a transparent, AI-enhanced TikTok algorithm, enriching user experience and content discovery.

• Perplexity introduces answer modes for enhanced search experiences in verticals like travel, shopping, places, images, videos, jobs.

• Microsoft introduces AI-powered Security Copilot agents to tackle phishing, data security, and identity management.

• Microsoft unveils Researcher and Analyst, they’re like having a highly skilled expert on call for you 24/7 across your work data and the web.

• Bing's Copilot AI Search is now accessible to all, challenging Google's AI mode with its summarization and answer capabilities.

• GitHub's new open-source MCP Server is now in public preview, offering enhanced tool customization and seamless VS Code integration.

• Ideogram 3.0 now brings stunning realism and consistency to all users for free, setting new standards in generative media.

• Midjourney releases V7 with Draft Mode, offering rapid, low-cost image iterations and seamless personalization features.

• Reve AI launches Reve Image 1.0, a pioneering text-to-image generation model excelling in text rendering and user intent comprehension.

• ChatGPT’s Advanced Voice Mode receives an update today, delivering enhanced personality, smoother conversations, and fewer interruptions.

• ChatGPT Team rolls out internal knowledge integration in beta, connecting to your organizations Google Drive for personalized responses.

• OpenAI just added support for MCP in our Agent SDK, enhancing its capabilities for developers worldwide.

• OpenAI announces the imminent release of O3 and O4-mini, with plans for a significantly improved GPT-5 in the coming months.

• Otter's AI Meeting Agent now answers questions and completes tasks in meetings using company data.

• DeepSeek releases DeepSeek-V3-0324, featuring enhanced reasoning, improved front-end skills, and smarter tool-use capabilities.

• FlutterFlow launches AI Agent Builder to create and deploy conversational AI Agents using Gemini, Firebase, OpenAI, and Anthropic.

• FlutterFlow introduces DreamFlow. It offers instant text-to-Flutter app creation and will soon enable direct app customization in FlutterFlow.

• Bolt introduces Discussion Mode, enabling brainstorming and planning without code and cutting token usage by up to 90%.

• Bolt announced Supabase Edge Functions support. It connects your project to external API’s, runs background jobs, and custom webhooks.

• Bolt introduces User-Provided Error Context. You can now describe what happened and what you expected, making Bolt fixes more effective.

• Bland AI introduces its new API feature: AI-powered emotional analysis for real-time insights into callers' emotions.

• Devin 2.0 introduces a new agent-native IDE experience, enabling concurrent multitasking and dynamic collaboration.

• Devin launches Search, for codebase understanding by scanning all repos and responding in seconds, complete with cited files and code snippets.

• Kling AI introduces AI Sounds, offering an emotive twist to elevate your storytelling with soulful audio enhancements.

• Gumloop unveils guMCP, the largest open-source aggregation of MCP servers, available to host for free.

• Luma AI debuts Camera Motion Concepts, offering 20+ precision-tuned motions for cinematic control and creating hundreds of new possibilities.

• Udio launches Styles, a feature that lets you craft new songs using existing tracks as stylistic blueprints, eliminating the hassle of text prompts.

• Runway introduces Gen-4, pushing the boundaries of media generation with unprecedented fidelity, motion control, and narrative capacity.

• Krea introduced a 3D tool, natural language image editing powered by Gemini, and Video Re-style to change the style of any video.

• Letta introduces Agent File (.af), an open format that imports/export agents effortlessly, preserving behavior and memories.

• Google Workspace allows users to create forms with AI assistance from various Drive files, currently available on desktop to select users.

• NotebookLM introduces the "Discover Sources" feature, effortlessly gathering and curating relevant web sources for your projects.

• Apple Intelligence now offers support for new languages and is accessible to iPhone and iPad users in the EU.

• Claude for Education launches, enabling universities to integrate AI into teaching and learning with specialized tools and programs.

• Amazon introduces Nova Act, an AI agent that can manage web browsers autonomously.

• Amazon reportedly submits last-minute bid to acquire TikTok as the app faces a looming U.S. ban deadline.

• Amazon tests "Buy for Me," an AI shopping agent that purchases from third-party sites without leaving the app.

• GenSpark Super Agent is an AI Agent capable of autonomously managing complex tasks by learning, planning, acting, and utilizing tools.

• Pikaframes now offers up to five frames for your photos, creating magical 25-second videos.

• Bytedance introduces DreamActor-M1, elevating full-body, expressive animation far beyond Act-One and Animate Anyone.

• Supabase MCP Server lets you manage your database, create projects, and run migrations directly from AI tools like Cursor and Windsurf.

This was it. Our thirty-third digital dip together. It might seem like a lot, but remember; this wasn't even everything that happened in the past few weeks. This was just a fraction.

It’s incredibly important to plan ahead now, so you won’t be left behind later. If you need help understanding all the developments and what the impact might be on your work or organization, send me a message and I’ll drop by.

Looking forward to what tomorrow brings! ▽

-Wesley