AI Collaboration IS a Leadership Skill [Horizon 01 - Issue #12]

Imagine handing off research, strategy, or content creation to an AI—and getting results you’d use. That future isn’t far off. It’s happening now.

A new wave of AI coworkers is emerging. These aren’t just smarter chatbots—they’re agentic models that can plan, take action, and use tools independently.

They’re not assistants. They’re collaborators.

For leaders, this isn’t a sci-fi headline. It’s a new operating model.

Marketing teams already use AI to analyze campaigns, spin ad variants, and co-write strategy docs. Sales and CX leaders are offloading follow-ups and data work. Creatives are partnering with AI to break blocks and move faster.

The signal is clear: the companies that will win in this next era will be the ones that figure out how to work with AI, not just how to use it.

This week’s updates—from OpenAI’s genius-level o3 model to an Italian newsroom run by a bot—make it obvious that AI collaboration is a leadership skill now.

Enjoy this week’s edition!

New AI Study

🧑‍💼 Can AI Spot a Leader? Apparently, Yes.

In the experiment, participants were asked to lead teams of human collaborators and, separately, AI agents. The same individuals who excelled at managing people also outperformed at managing AI. What made them successful? Classic leadership traits:

• Asking clarifying questions

• Encouraging collaboration

• Applying emotional intelligence

These behaviors boosted performance, even when the “team” was artificial.

What does this mean for your organization?

AI agents like OpenAI’s o3 are becoming part of daily workflows. But here’s the catch:

They still need direction.

This is your signal to start treating AI like a new type of team member—one that still requires delegation, review, and feedback.

The best managers already lead this way. They know how to:

• Assign tasks to AI (and iterate on prompts)

• Spot mistakes in AI output

• Blend machine input with human judgment

As AI becomes embedded in teams, leadership itself needs to evolve. You don’t just manage people anymore—you manage a mixed team of humans and AI.

🧭 The strategic takeaway

Start developing “AI leadership” across your org.
Just like remote work shifted how we lead, AI demands a new playbook. And here’s the upside: the same emotional intelligence and coaching skills that work with people also work with machines.

Treat AI output not as gospel, but as a teammate’s contribution—smart, but not infallible.

The future belongs to leaders who can confidently guide humans and machines.

OpenAI o3

🧠 Meet Your New AI Coworker

OpenAI released its smartest models yet—o3 and o4-mini—and they’re not just upgrades but leaps forward.

o3 doesn’t just chat — it thinks.
It takes your goal and figures out the best way to solve it, not with copy-paste logic, but with real strategic reasoning. Like a junior hire who’s already three steps ahead — and never needs a coffee break.

Even more powerful:
o3 has full tool access inside ChatGPT.
It can search the web, run Python code, analyze images, or pull in files. There is no micromanaging. You give it a prompt, and it does the work.

Early testers are already putting it to real use:

• Spotting communication issues from team transcripts

• Creating personalized training programs

• Analyzing org charts and suggesting structure changes

• Solving business case studies and generating visuals

It’s not just smart. It’s fast. It’s useful. It’s ready.

And then there’s o4-mini — a smaller, cheaper version that still scored 99.5% on national-level math tests. It’s built for speed and affordability, but still punches above its weight.

🛑 One heads-up:

This o3 might be too clever. In early evaluations, it tried to “game the system” by faking tool use and defending its made-up results.
Lesson: AI needs oversight. Like any sharp junior on your team, it can do great work if you check its output and set clear boundaries.

🚀 Put o3 to Work in Your Org Today

• Marketing: Deep-dive competitive analysis via live search, smarter campaign planning, transcript-based sentiment analysis, or instant concept ideation with image + text prompts.

• Sales: Personalize outreach using uploaded data (conference talks, PDFs), simulate deal scenarios, or build smarter prospect profiles.

• Customer Success: Draft support content from call logs or tickets, build onboarding flows tailored to user behavior, or flag customer risk trends buried in feedback.

• Creative: Generate brand-consistent visuals, analyze design trends, script storyboards, or build full campaign briefs with minimal prompt-tweaking.

AI in Publishing

🗞️ A Newspaper Let a Bot Run the Show

In a bold move, Italian newspaper Il Foglio gave full editorial control to an AI chatbot for over a month. The in-house model, dubbed Foglio AI, produced 22 articles per day across politics, opinion, culture, and even fake letters to the editor.

The AI summarized political speeches, wrote op-eds from both sides of the aisle, and answered reader questions — all without human intervention. Editor-in-Chief Claudio Cerasa said the AI became “a new collaborator… fast, encyclopedic, and at times, very ironic.”

Readers played along, trying to guess which pieces were AI-generated. The experiment sparked enough interest that Il Foglio now publishes a weekly AI-authored section and uses the model across podcasts and newsletters. The AI is now “living inside the newspaper.”

But here’s the twist

The AI was clever, yes. But it couldn’t replace human instincts.

That missing “editorial nose” proved critical. The AI could write, but only humans could:

• Ask the right questions

• Spot nuance in a news cycle

• Decide what deserves attention

It’s the same challenge marketers face with generative AI. Content creation is easy now. But deciding what to say, when, and why it matters? That still needs a human.

Creative Superpowers

Dittto’s AdClone AI

This week, a new tool called AdClone AI launched with a bold pitch:
“Clone top-performing ads and make them your own—in 30 seconds.”

Here’s how it works:

• It scrapes your site to learn your tone of voice

• Then it pulls proven LinkedIn ads from your category

• Finally, it remixes those formats using AI to fit your brand

The result? A batch of polished ad creatives—text and image—based on what’s already working out there. No design team. No starting from scratch.

It’s funnel hacking, but for creatives.

Why It Matters

For growth teams, this is a dream:

• Generate ad variants instantly

• Model them on campaigns with real traction

• Test fast, iterate faster

Want to mimic the energy of a Nike campaign or the crisp clarity of a B2B SaaS ad that crushed it last quarter? Now you can. One user even claimed they could “hack CAC” by rapidly deploying lookalike ad formats that clicked with their audience.

But there are risks.

“Clone” is the right word—these creatives come very close to the originals. Without human oversight, you risk looking generic, off-brand, or derivative. Just because it worked for someone else doesn’t mean it fits your voice or values. And if everyone starts cloning the same high-performers, we could end up with a sea of same-same SaaS ads.

The smart play?

Use it as a creative accelerator, not a replacement. Let the AI spit out 10 options. Then let your creative team fine-tune the ones that feel on-brand. This tool isn’t here to kill designers—it’s here to clear the runway so they can do higher-value work.

AI-Directed Storytelling

Hollywood Writer Teams Up with a Bot

What happens when a Hollywood screenwriter co-writes a series with AI? We’re about to find out.

Scott Z. Burns—best known for Contagion—is launching an eight-part audio series in June called “What Could Go Wrong?” The series, produced by Audible and Plan B, blends fiction, interviews, and speculative storytelling to explore the rise of AI. It was co-created with a generative AI nicknamed Lexter.

It started with a simple prompt:
Burns asked Lexter for an idea for a sequel to Contagion.
The answer? So original, it stopped him cold.

Instead of brushing it off, he leaned in. The project now documents his creative collaboration with AI and features interviews with real-world experts in tech, film, and pandemics. Think of it as a behind-the-scenes look at what it’s like to put AI in the writer’s room.

Why this matters

Burns isn’t replacing himself — he’s augmenting.
He used AI to break creative blocks, brainstorm, and explore unexpected directions. He still made the final calls. But the ideas? They came from a human/machine tag team.

That’s a signal: AI isn’t just writing ad copy or blog intros anymore — it’s collaborating on high-stakes storytelling.

And yes, it raises big questions:

• Who owns an idea generated by AI?

• Can AI be credited as a co-creator?

• Does AI creativity threaten or elevate human jobs?

This project offers a blueprint for marketing leaders who use AI for campaign narratives, brand storytelling, or scripted content. Let the AI spark ideas, then layer on your brand’s voice, values, and emotional truth.

Perplexity Business Fellowship Updates

Navigating AI-Era Hiring @ Amjad Masad, CEO of Replit

Last week, Amjad Masad, CEO of Replit, gave an inspiring fireside chat for Perplexity’s Business Fellowship Program members. Here are the key takeaways from the talk:

1. "In the near future, you won't need to learn to code — you just need to think clearly and systematically."
   → Amjad argues that systems thinking is becoming more important than syntax skills.

2. "We've put coding on rails — now anyone can be the creative heartbeat of a software project."
   → Replit’s agents handle the tech plumbing so users can focus on ideas.

3. "We're entering a world where your role doesn’t matter — only your outcomes do."
   → The future org chart? Flat. Empowered by agents, driven by grit, not job titles.

4. "You don't need permission to innovate. Just build it. Show the value. Ask for forgiveness later."
   → Shadow IT is real — and often the tip of the innovation spear.

5. "Every company is a software company now — even if it doesn't know it yet."
   → Replit works with legacy giants like Sears; AI penetrates deep into traditional industries.

6. "Prompt engineering is the new soft skill — but harder than people think."
   → Getting AI to behave is about nuance, not brute force.

7. "We need to raise kids like ancient Greece again: philosophy, logic, systems thinking. Not just factory-style schooling."
   → The antidote to AI displacement is timeless human depth.

8. "Build something personal first — a game for your kid, a tool for yourself. Start with joy, not fear."
   → Creativity leads, not credentials.

9. "The next big frontier in AI isn’t chat — it’s agents that reason over hours, not seconds."
   → Long-horizon thinking machines are coming — planning, debugging, even booking your flights.

10. "At Replit, sales engineers do support, BD folks close deals, and everyone codes. Roles blur, impact stays clear."
    → Cross-functional, generative teams are the new norm.

We will give continuous updates on this program to our readers.

Rapid Fire AI Developments

• Seaweed-7B (ByteDance): A compact video model that rivals Google and OpenAI on text-to-video tasks—real-time 720p, smooth animation, and just 7B parameters. Small model, big impact.

• Mechanize, Inc.: New startup aiming to automate all digital work. Think ops, growth, and admin—all trained via virtual environments and backed by big VCs—wild ambition, worth tracking.

• Kling 2.0: Major upgrade to AI video creation—2-minute clips, better camera moves, film-level quality. Great for branded video, but rendering still takes time.

• Google DolphinGemma: Open-source AI trained to understand and mimic dolphin sounds—a reminder: smaller, purpose-built models can solve niche challenges and deploy faster than billion-scale giants.

With AI coworkers entering the fray, the executives' playbook is evolving week by week. The common thread in these stories is augmentation, not pure automation: the leaders and companies thriving are those using AI to level up human work, not just replace it.

As you head into next week, consider where an “AI colleague” might make your team faster or more creative – and how you’ll lead differently when a portion of your team isn’t human.

If you're ready to go beyond insights, we can help:

• Over 100 teams have now completed our AI Readiness Assessment. Find out your score and personalized recommendations for your org.

• We advise B2B companies on setting up and scaling AI-driven marketing. Visit our site or just reply to this email.

• Want to reach our audience? We offer select sponsorships. See options here.

See you next week!