Today, we’re bringing you the latest in AI-powered marketing and business strategies. Here’s what’s inside:

🚨 AI Top Story: Human-level AI by 2027? We break down the forecast shaking up the industry

🎥 AI YouTube Resource Of The Week: How To Use Sora and ChatGPT to Create Cinematic AI-Generated Videos

💬 Killer Marketing Prompt: Get AI To Help You Decide If That Project Is Worth Your Time.

🎯 AI Use Case Of The Week: See how KPMG quietly used AI to cut HR support calls by 20%—and how you can build your own onboarding agent.

🌟 Creator Spotlight: Will Francis on how to spot the signs of AI generated text—and how to make them disappear.

What Happens If They’re Right About 2027?

Are We Ready for AGI This Soon?

Another week, another wild AI prediction…but this one’s gaining traction.

According to a new roadmap from a group of well-connected researchers, human-level AI could arrive as early as 2027 (some with connections to OpenAI and the Center for AI Policy).

A new roadmap has been released predicting that within the next 24 months, we might see models capable of reasoning, adapting, and performing across virtually every domain just like a human would. Even more mind-blowing, they suggest that artificial superintelligence (ASI) could follow only months later.

Now, you might be thinking: isn’t AGI already here? In some ways, yes.
Today’s top-tier models can code, write, summarise, plan, and even reason across a wide range of tasks. That’s a level of generality we hadn’t seen before; what some are starting to call “early AGI.” But the kind of AGI we have in 2025 still relies heavily on prompting, has limited autonomy, and doesn’t yet possess a full understanding of context or the real-world grounding that humans do.

What this forecast is pointing to is not just AGI in name, but human-level AGI - an AI that can think, learn, and operate with the depth, nuance, and independence of a person. That’s a much bigger leap.

Of course, not everyone agrees with the timeline. While the forecast maps out a detailed, quarter-by-quarter sprint to AGI, some experts are skeptical. They argue that continued progress depends on breakthroughs in reasoning, memory, and real-world interaction, not just scaling more compute.

Ali Farhadi from the Allen Institute for AI has voiced concerns that the forecast might be more aspirational than grounded. Others, like Jack Clark from Anthropic, see it as a plausible technical scenario that deserves serious attention.

But progress at this scale also comes with serious responsibility. With every leap forward, the stakes get higher. Human-level AGI and ASI carry risks - especially if they’re not aligned with human values or deployed without guardrails. As the people actively using and integrating these tools, marketers need to stay not just curious, but critically informed.

Whether true AGI becomes a reality by 2027 or arrives later, one thing is clear: we’re approaching a turning point in what AI can do. The coming years could completely reshape how we think about intelligence, creativity, and strategy. The question isn’t really just about how advanced AI becomes, it’s how ready we are to meet it.

🧠 OpenAI Interested in Buying Google's Chrome Browser if Court Orders Spin-Off- bloomberg.com - OpenAI says it would consider acquiring Chrome if regulators force Google to spin it off

📬 Your next marketing challenge may be winning over the AI in charge of the customer’s inbox - Marketers now have to think about not just the customer, but the AI screening their emails first. A new layer of gatekeeping is here.

🔍 YouTube is testing AI Overviews in its search results - Search on YouTube may start looking a lot more like ChatGPT—with AI-generated summaries sitting above video results.

🧪 OpenAI rolls out a ‘lightweight’ version of its ChatGPT deep research tool - OpenAI’s pro research assistant just got a simplified version, making advanced AI analysis more accessible to everyday users.

WILL FRANCIS - ChatGPT leaves hidden watermarks in the text that it generates. This post breaks down how you can both spot AI generated text, and also how you can remove these blatant giveaways. Perfect for marketers, writers, or anyone blending AI into their content.

(OR if you want to know whether you’re favourite creator is using AI 😬)

How To Use Sora and ChatGPT to Create Cinematic AI-Generated Videos

David Sheldrick walks through his full process for creating visually rich music video-style edits using Sora and ChatGPT. From creative ideation and world-building to rendering, editing, and assembling final cuts.

New Hire Questions? KPMG Taught AI to Handle Them

The Overview

KPMG is rethinking how new hires get up to speed, with a little help from Microsoft’s AI.

Instead of sending people hunting through folders or chasing down answers, they’ve built an internal onboarding agent using Azure OpenAI. It gives new team members quick access to templates, key documents, and the kind of context that usually requires a few emails and a couple of calls.

The impact’s already showing: onboarding is more consistent, and follow-up calls to HR have dropped by 20%. It’s a practical use of AI that’s quietly improving how things get done

The Implementation

Built on Azure OpenAI:
The onboarding agent runs on Microsoft’s enterprise-grade AI platform, giving it secure access to KPMG’s internal knowledge and the ability to respond with clarity and context.

Answers without the back-and-forth:
Instead of waiting on HR or digging through internal wikis, new hires get what they need—templates, policies, helpful links—right when they need it.

Learns from what’s worked before:
The agent draws on historical onboarding flows and internal docs, making it easier to surface what’s actually useful instead of just pointing to a knowledge base.

Replicate it

Step 1: Gather your internal resources
Make a list of the most frequently asked onboarding questions, helpful links, templates, and HR documentation. Ask HR what they’re tired of answering. That’s your starting data.

Step 2: Build a structured knowledge base
Organise this content in a way that’s searchable and consistent. This can live in a simple Notion doc, SharePoint, or internal wiki (but make sure it’s clean and structured).

Step 3: Use a secure LLM platform
Plug that content into an AI tool with retrieval capabilities—like Microsoft Azure OpenAI, ChatGPT with custom GPTs, or Google Vertex AI. Use embedding + retrieval to keep things accurate and contextual.

Step 4: Create a conversational layer
Use a simple chatbot interface (think Power Virtual Agents, Slack bots, or a custom front-end) so employees can ask questions naturally, and get relevant answers without guessing keywords.

Step 5: Pilot and refine
Start with a small cohort of new hires. Track the most common queries, gather feedback, and tweak your system to surface better answers or cover gaps.

“Waiting for the end to come” — of the interview process 😂