👋 Hey,
Welcome to the latest edition of The AI Driven Marketer!
Every week, I cut through the noise to bring you the most practical marketing AI news, tools, and a battle-tested automation workflow you can plug directly into your business.
Let's get into the build.
⚡ The AI Pulse: What Marketers Need to Know
Anthropic Ships Claude Sonnet 5, Its Most Agentic Mid-Tier Model Yet
On June 30, Anthropic released Claude Sonnet 5 and made it the default model for Free and Pro users. It plans, uses tools like browsers and terminals, and runs multi-step tasks on its own, landing close to flagship Opus 4.8 performance while costing far less. Introductory pricing is $2 per million input tokens and $10 per million output tokens through August 31.
So What? The cheap tier can now do the agentic work that used to require the expensive model. If you run AI research, content, or automation workflows, your cost per task just dropped while output quality went up. This is the week to move your heavier automations onto Sonnet 5 and pocket the difference.
Anthropic Brings Claude Fable 5 Back Online Globally
Fable 5, Anthropic's most capable publicly available model, returned worldwide on July 1, nearly three weeks after a June 12 US export control order forced Anthropic to pull it. The order followed a reported jailbreak that could push the model to write exploit code, so Anthropic redeployed with a new safety classifier that blocks that technique in over 99% of cases and routes some coding and debugging requests to Opus 4.8. It is live again on Claude.ai, Claude Code, and Claude Cowork.
So What? The comeback matters less than the precedent: this was the first time the US used export controls to switch off a commercial AI model, and it happened with about 90 minutes of notice. If your workflow leans on a single frontier model, that is a real continuity risk. Keep a fallback wired into your stack (Opus 4.8 or Sonnet 5) and avoid building a business-critical process around one model you cannot swap out.
Google Makes AI Image and Video Cheap Enough to Run at Ad-Campaign Scale
Also on June 30, Google shipped Nano Banana 2 Lite, its fastest and cheapest image model, which generates a usable image in about four seconds, and released Gemini Omni Flash for video in public preview at $0.10 per second of output. Both handle legible on-image text, character consistency, and quick localization.
So What? Producing 20 ad-creative variations or a batch of product videos is no longer a budget or time problem. You can A/B test visual concepts, localize typography for different markets, and spin up short product clips without a designer queue. Cheap iteration is the whole game in paid social, and it just got cheaper.
Klaviyo Launches Two AI Agents That Share One Customer Profile
On June 30, Klaviyo moved Composer, its marketing agent, into public beta and upgraded Customer Agent, its service agent. Both run off the same real-time customer record, so a support conversation updates the data that shapes your next campaign, and the reverse holds too. Customer Agent can complete returns, apply loyalty points, and run across web chat, email, SMS, and WhatsApp, configured with a single prompt.
So What? For ecommerce brands and solo operators, this closes the gap between marketing and support that used to need two tools and a bigger team. Your service insights now sharpen your campaigns automatically. If you already run Klaviyo, you can stand up a branded service agent with no engineering and let it improve itself daily.
TikTok Opens an Agentic Hub So AI Agents Can Run Your Ad Campaigns
On July 1, TikTok launched Agentic Hub, a marketplace of prebuilt AI Skills that plug into TikTok Ads through its Model Context Protocol. Agents can handle campaign setup, creative generation, performance analysis, audience insights, and catalog management, with no API keys or code. Launch partners include HubSpot, Wix, and Constant Contact.
So What? You can now manage and improve TikTok campaigns by talking to an AI agent instead of clicking through Ads Manager. For a small team, that removes the manual setup and reporting grind and makes TikTok advertising workable without a dedicated media buyer. The same MCP pattern is spreading across ad platforms, so it is worth building your agent stack around it now.
xAI Launches a No-Code Builder for AI Phone Agents
On July 2, xAI released Grok Voice Agent Builder in beta. You describe how a call should go, attach your business files, pick a voice, and test real calls, all without code. It supports tools, guardrails, MCP connections, and call review, priced at $0.05 per minute of audio, with xAI phone numbers at $0.01 per minute and no platform fee during beta.
So What? Solo founders and small teams can now run a phone line that answers calls, books appointments, and handles basic support for pennies per minute. That is a front desk you could not previously justify hiring for. Start it on after-hours and overflow calls, review the transcripts, and expand from there.
🛠️ The AI Toolkit
3 Tools specifically curated for digital marketers.
1. Arcads
Best for: DTC marketers and founders who need scroll-stopping video ads without hiring actors or a film crew.
What it does: Turns a script into UGC-style video ads using a library of 1,000+ AI actors, with built-in editing, subtitles, and translation into 30+ languages.
2. Gumloop
Best for: Solo founders and lean teams who want to automate repetitive work without writing code.
What it does: Lets you build and deploy no-code AI agents for research, CRM updates, support triage, and more on a visual canvas, then tag them into Slack or email like coworkers.
3. Fyxer
Best for: Founders and consultants drowning in email who want their inbox handled for them.
What it does: An AI assistant that organizes your inbox, drafts replies in your own voice from past conversations, and takes meeting notes so every reply has full context.
Bonus
The Cheap-Creative Sprint
Google just dropped Nano Banana 2 Lite (images in about four seconds) and Gemini Omni Flash (video at $0.10 per second). That changes the math on paid social: you can produce a full week of ad creative for the price of a coffee. Here is a repeatable Friday-afternoon sprint that turns one offer into 20+ image variations and a product video, for under $10.
Step 1: Lock the brief (10 minutes)
Pick one product, one offer, and one audience. Write the value proposition in a single sentence: "[Product] helps [audience] [get outcome] without [pain]." Everything you generate this session serves that one sentence. A tight brief is what makes 20 variations feel like a test, not noise.
Step 2: Batch your image concepts with Nano Banana 2 Lite
Generate 5 concepts times 4 variations, so 20 images, in one sitting. Use this prompt template and change only the bracketed parts:
A [format: UGC selfie / flat-lay / lifestyle / studio product] ad image for [product], shot for [platform]. [Audience] using it to [outcome]. Include short on-image text that reads exactly: "[hook, 3-5 words]." Bright, thumb-stopping, high contrast, vertical 4:5.
Run it five times with different formats and hooks, then ask for four quick variations of each. Because the model renders legible text, your hook lives in the image, not just the caption.
Step 3: Localize and swap hooks (no reshoot)
For each winning concept, generate the same image with the text swapped for different hooks or languages. One line does it: "Same image, change the on-image text to: '[new hook]'." That is how you test six angles for one product without a designer or a photoshoot.
Step 4: Cut one product video with Gemini Omni Flash
Turn your best still into an 8 to 10 second motion ad. Prompt:
Animate this product image into an 8-second vertical ad: slow push-in on the product, [action: hand picks it up / label turns to camera], soft studio lighting, upbeat background music, end on the product with the text "[offer]." Keep the product and colors consistent.
At $0.10 per second, a 10-second clip costs about $1. Make two versions and keep the stronger one.
Step 5: Label, ship, and let the data pick
Name every file so you can read results at a glance, like product_format_hook_v1. Push three to five variants per platform, run them for 48 to 72 hours on a small budget, then move spend to the two that win. Cut the rest and rerun the sprint next Friday with what you learned.
What it costs (estimated)
Item | Volume | Rough cost |
|---|---|---|
Ad images (Nano Banana 2 Lite) | ~30 | ~$1.00 |
Hook and language swaps | ~10 | ~$0.35 |
Product video (Gemini Omni Flash, 10 sec) | 2 clips | ~$2.00 |
Total | a week of creative | under $4 |
💡 Key insight: Cheap iteration beats one expensive hero asset. When a fresh variation costs pennies, the winning move is to test more angles, not to polish one. Let volume plus a small ad budget find the creative that converts, then put your money behind that one.
That's a wrap for this week in AI.
See you next week,
Rananjay
P.S. If you found this valuable, share it with a fellow marketer who's trying to make sense of AI. And hit reply to let me know—what's the one AI workflow you want to automate this quarter?