In partnership with

We close out 2025 with a seismic shift: Meta acquired Manus for $2B, all major AI labs now have production browser agents, and enterprise AI spending hit $37B. Here's your final intelligence brief of the year.

From PromptHacker

Everyone Says Apple Is Losing the AI Race. They're All Missing the Point. Read the full analysis →

⚡ Quick Hits

  • Google's Gemini 3 Flash is now the default fast model, delivering 4x performance boost for consumer apps and real-time queries. Details →
  • Zapier Central updates with no-code AI agents for connecting apps like HubSpot to ChatGPT for automated emails and CRM updates. Learn how →
  • ChatGPT's "Your Year with ChatGPT" launched its Spotify Wrapped-style recap showing your message count, most-used features, and AI-generated poetry based on your 2025 conversations. Get yours →
  • Google Translate launched live speech translation in headphones supporting 70+ languages while preserving original intonation. Try it →
  • Notion webhook actions now connect databases directly to Zapier, Make, and 1,000+ tools from page buttons or automation rules. Set up →
  • Enterprise AI spending hit $37 billion in 2025, with 80% adoption in productivity tools from Microsoft and Google. Full report →
  • Apple Intelligence expands across apps with 25+ new Shortcuts actions, auto-categorized Reminders, and Wallet-style order summaries. See features →
  • Microsoft 365 Copilot Chat now runs GPT-5 by default with smart routing that selects the best model variant per task. Details →

The Future of Shopping? AI + Actual Humans.

AI has changed how consumers shop by speeding up research. But one thing hasn’t changed: shoppers still trust people more than AI.

Levanta’s new Affiliate 3.0 Consumer Report reveals a major shift in how shoppers blend AI tools with human influence. Consumers use AI to explore options, but when it comes time to buy, they still turn to creators, communities, and real experiences to validate their decisions.

The data shows:

  • Only 10% of shoppers buy through AI-recommended links

  • 87% discover products through creators, blogs, or communities they trust

  • Human sources like reviews and creators rank higher in trust than AI recommendations

The most effective brands are combining AI discovery with authentic human influence to drive measurable conversions.

Affiliate marketing isn’t being replaced by AI, it’s being amplified by it.

🚀 Top AI Updates

Agentic AI: Your Guide to Autonomous Workflow Orchestration

Meta acquired Manus for $2 billion, validating what's now clear: autonomous AI agents are the next platform war. These aren't just chatbots. They're multi-step, multi-app orchestration systems. Here's what each offers.

OpenAI Operator / ChatGPT Agent Mode

Executes 10-15 step workflows across web apps: research, form filling, booking, purchases. Available in ChatGPT Plus/Pro. Autonomy: Semi-autonomous; you initiate tasks but it executes multi-step sequences independently with periodic check-ins. Best for: General-purpose web automation without coding.

Anthropic Claude Skills + Chrome Extension

Teach Claude custom workflows that chain actions across Atlassian, Canva, Figma, and Notion. Chrome extension ($100-200/month) adds browser control. Autonomy: Configurable per skill; you set granular site-level permissions and approve which actions run automatically vs. require confirmation. Best for: Teams needing repeatable multi-app workflows with permission controls.

Google Gemini Deep Research Agent

Autonomously researches dozens of sources and outputs cited reports to Docs, Sheets, and Slides. Gemini Advanced required. Autonomy: High for research; it plans its own search strategy and synthesizes findings, but outputs require your review before sharing. Best for: Market research and competitive analysis in the Google ecosystem.

Gemini Agent for SMBs ("Operations Manager")

Handles 5-10 step workflows: draft contracts, email clients, create Drive folders, update CRM, schedule calls. Requires Workspace Business/Enterprise + Gemini add-on. Autonomy: Semi-autonomous (human-in-the-loop); plans entire workflow but requires a "Proceed" click for each external action like sending emails or signing contracts. Best for: Small business owners who want end-to-end client workflows.

Google Ads Advisor & Analytics Advisor

Autonomous marketing agents that monitor thousands of keywords/bids simultaneously. Set "bounded permissions" (e.g., reallocate up to 10% budget without asking). Autonomy: High within your defined guardrails; acts independently on optimizations below your threshold, escalates decisions above it. Best for: Transparent ad optimization with guardrails you control.

Google Premade "Role" Gems

Pre-configured AI personas (Hiring Consultant, Writing Editor, Sales Pitch Ideator) for deep-dive sessions. Autonomy: Low (interactive only); these are consultation tools that advise and draft but do not execute actions in other apps. Best for: Consultation and content refinement, not execution.

Google Workspace Studio

No-code agent builder at studio.workspace.google.com. Drag-and-drop triggers and actions (up to 20 steps). Autonomy: Medium (trigger-based); runs fully in background when conditions are met, but you can add "Ask for approval" steps for human checkpoints. Best for: Custom automated workflows across Gmail, Drive, Sheets, Calendar.

Perplexity Comet Browser

AI-native browser with assistant that sees your tabs. Background Assistants (Max plan) run parallel tasks: travel booking, price comparison, data entry. Free tier available. Autonomy: Medium-high; Background Assistants work independently across multiple tasks, but prompt for permission before taking actions in authenticated sessions. Best for: Replacing Chrome with an AI-first browsing experience.

Microsoft Copilot Studio + Agent 365

Build agents that create Office documents, respond to triggers, and connect to GitHub/Asana/Jira via MCP. Computer use lets agents control any UI. Autonomy: Fully configurable; set agents to run autonomously on triggers or require approval at any step, with enterprise-grade audit logging for compliance. Best for: Enterprise teams on Microsoft 365 needing governance and compliance.

Notion 3.0 AI Agents

Agents run autonomously for 20 minutes across hundreds of pages. Pull from Slack, Drive, GitHub. Custom Agents run on schedules while you sleep. Autonomy: High within Notion; executes complex multi-page operations independently, but external actions (emails, posts) still require connected app permissions. Best for: Teams using Notion as their workspace hub.

Manus (Now Meta)

Hit $100M ARR in 8 months before Meta's $2B acquisition. Full end-to-end workflow execution across research, coding, and analysis. Autonomy: High; designed for fully autonomous complex task completion with minimal human intervention during execution. Best for: Watch this space as Meta integrates it across Facebook, Instagram, and WhatsApp.

Want the complete breakdown? Read our Complete Agentic AI Overview for 2026 →
Full comparison of every agent, pricing, access requirements, and which service fits your workflow.

🎯 Pro Tip

Multi-Agent Debate Prompting for Better Decisions

Use "Simulator" style prompts where you ask ChatGPT or Claude to simulate multiple executives debating a business decision. This forces the model to consider expert perspectives from different angles and surface blind spots you'd miss with single-perspective prompts.

Example Prompt:

Simulate a 15-minute executive debate between a CFO and CTO about whether we should invest $500K in building a custom AI tool vs. subscribing to enterprise AI software. Each executive should: 1. State their initial position with 3 supporting arguments 2. Challenge at least one of the other's assumptions 3. Propose one risk mitigation for their preferred approach After the debate, synthesize the key decision criteria and provide a recommendation with confidence level.
  • Why it matters: This technique reduces confirmation bias by forcing the model to steelman opposing viewpoints before reaching conclusions.
  • Why it matters: Works especially well with GPT-5 and Claude Opus, which excel at maintaining distinct "voices" across extended reasoning chains.

💡 Productivity Gem

ChatGPT Wrapped for Workflow Optimization

Your "Year with ChatGPT" recap is more than novelty. Use it to identify your most productive patterns and underutilized features for 2026 planning.

Setup Steps:

  1. Open ChatGPT and type "Show me my year with ChatGPT"
  2. Note your "chattiest day" and cross-reference with your calendar
  3. Identify your top themes and evaluate if they align with your actual priorities
  4. Check which features you used most (Projects, Custom GPTs, Vision) vs. which you ignored

Why it matters: Most users discover they only use 20% of ChatGPT's capabilities. The recap reveals whether your AI usage actually supported your goals or just became a time sink.

⚕️ AI-Enabled Health Tip

Gemini Health Check via Google Search

Google's Gemini (via the new AI Search features) can answer health queries using verified sources. Unlike generic AI chat, these responses are specifically tuned for medical accuracy with citations to reputable health sites.

  • Try asking Gemini in Google Search for "best exercises for lower back pain" and it will cite reputable health sites in its answer
  • Search "diets to lower cholesterol" for cited dietary guidance rather than generic advice
  • Use "symptoms of [condition]" queries to get structured information with "When to see a doctor" guidance

Why it matters: This gives you a reliable first-pass for health questions without the hallucination risks of unconstrained AI chat. Always consult a healthcare professional for medical decisions.

🧒 AI for Kids Tip

ChatGPT Education Mode + MIT App Inventor for STEM Projects

Use OpenAI's ChatGPT Education mode for interactive STEM projects. Guide your child through building a simple AI chatbot using no-code tools, fostering coding literacy with safe, step-by-step learning.

  • Age range: 10-14 years old
  • Setup: Open ChatGPT with a parent logged in. Prompt: "Guide my 10-year-old through building a simple AI chatbot using MIT App Inventor. Break it into 30-minute sessions with clear stopping points."
  • Guardrails: Stay in the same chat session to maintain context. Review each step before proceeding. Keep memory enabled so it tracks progress across sessions.

Why it matters: This approach combines AI-assisted learning with hands-on creation, teaching kids how to build with AI, not just use it. MIT App Inventor is free, browser-based, and produces real Android apps they can share.

AI at Work Cartoon

That's a wrap. We are now a couple of days into 2026 (already!). What AI service are you most looking forward to using this year? Hit reply and let me know.

This space changes fast. Stay curious and keep learning daily!

Keep Reading

No posts found