Instagram | Sponsor this newsletter | Let’s Connect!

My fellow AI explorers

We've got a US president threatening a nuclear standoff via AI selfie, a sequel to a 2006 fashion movie that's somehow become the most accurate documentary about AI and the death of media, and scientists quietly discovering that your favorite chatbot knows how to engineer a superbug… And that's before we get to Wall Street.

In today’s edition:

• 💰 Hightouch raises $150M to make "agentic marketing" now officially a category

• 🤫 DeepSeek's sequel flopped… and the market barely blinked

• 🔫 Trump posts an AI war-selfie to threaten Iran (yes, really)

We hired one colleague for every department.

Last Tuesday, marketing asked Viktor to write the weekly campaign recap, pull performance from Google Ads and Meta, and format it as a PDF for the exec team. Done in four minutes.

That same afternoon, engineering asked Viktor to review three open pull requests on GitHub, cross-reference with the Linear sprint board, and flag anything blocking the release. Posted to private channel before standup.

At 9pm, ops asked Viktor to draft a vendor contract summary from three Notion docs and send it to the team. It was in #ops by morning.

None of them knew the others were using it.

Same colleague. Three departments. That's what changes when your AI coworker lives in Slack, where your whole company already works. It's not a tool one person logs into. It's a teammate everyone messages.

5,700+ teams. SOC 2 certified. Your data never trains models.

"Viktor is now an integral team member, and after weeks of use we still feel we haven't uncovered the full potential." - Patrick O'Doherty, Director, Yarra Web

Start free. $100 in credits →

AI in the Boardroom

Hightouch Raises $150M to Put AI Agents in Charge of Your Marketing

Agentic marketing has arrived. And Goldman Sachs and Bain just wrote the check to prove it.

Hightouch, a composable customer data platform, just closed a $150M Series D led by Growth Equity at Goldman Sachs Alternatives and Bain Capital Ventures, valuing the company at $2.75 billion. Other investors include Iconiq, Sapphire, Y Combinator, and — interestingly — The Trade Desk's venture arm.

Here's what they actually do:

• Build AI agents that sit directly on top of a company's existing data systems

• Automate the full marketing workflow: campaign planning, content creation, execution, and optimization

• Work cross-channel, replacing much of the manual work marketing ops teams currently do

The signal here isn't the funding. It's who's backing it and why. Goldman's partner on the deal said Hightouch has positioned itself to "define the next category of marketing infrastructure." That's not investor fluff. That's a bet that the CMO's entire stack is about to get rebuilt around agents.

Hightouch has grown more than 100% year-over-year for two consecutive years. At a $2.75B valuation, the market is pricing in a much bigger footprint ahead.

🔮 Prediction: Agentic marketing is the first enterprise function where AI agents will fully replace human workflows at scale. It’s not assisting them… it’s replacing them. Every marketing automation platform you know (HubSpot, Salesforce Marketing Cloud, etc.) is now looking over their shoulder. This category could easily produce a $20B+ company within five years.

Read More Here

The China AI Front

DeepSeek's Sequel Bombed… And Nobody Was Surprised

Remember January 2025? DeepSeek R1 dropped and wiped nearly a trillion dollars off the Nasdaq in a single day. The world panicked. The narrative was: China has closed the gap, AI is cheap now, Nvidia is finished.

Fast forward to this week: DeepSeek released V4, the long-awaited follow-up, and the market's reaction was essentially a shrug.

Here's what happened:

• DeepSeek V4 improves on reasoning, coding, and efficiency: solid upgrades across the board

• A 1 million token context window and a mixture-of-experts architecture are the headline specs

• Compatible with Huawei's Ascend chips for inference (but not training)

• No major breakthrough. No market-moving moment.

The deeper story is about hardware. Chinese authorities pushed DeepSeek to train R2 on Huawei's Ascend chips after the US banned H20 GPU exports to China. The effort reportedly stalled. DeepSeek's founder was openly dissatisfied with progress and went back to using available Nvidia hardware. The result is a product that's incrementally better, not categorically different.

The original R1 disruption worked because it was genuinely unexpected. Now, OpenAI, Anthropic, and Google have all accelerated their release cadences. The field caught up. The cost advantage is narrowing. And investors have already priced in continuous improvement.

🔮 Prediction: DeepSeek's real problem isn't V4. It's that the surprise factor is gone. The next Chinese AI moment won't come from cheaper training. It'll come from a hardware breakthrough on Huawei's side, or from a genuinely new architecture that Western labs haven't seen coming. We're not there yet.

Read More Here

AI & Geopolitics

Trump Posts an AI War Selfie to Threaten Iran, And Oil Hit $115

This one's hard to write about without rolling your eyes. See if you can do any better when you read it.

At 4 a.m. ET on Wednesday, President Trump posted an AI-generated image of himself on Truth Social: sunglasses, a long-range weapon, explosions in the background, and a caption reading "NO MORE MR. NICE GUY!" The accompanying text: "Iran can't get their act together. They don't know how to sign a nonnuclear deal. They better get smart soon!"

What you need to know about the context:

• The US-Iran war, sparked by joint US-Israel drone strikes earlier this year, is technically in a ceasefire

• Iran has closed the Strait of Hormuz (through which roughly 20% of global oil flows) for two months

• Oil hit $114/barrel after the post; US gas prices are at their highest in nearly four years

• Negotiations have stalled, with Iran offering to reopen the strait only if the US lifts its port blockade

This isn't just political theater. It's a case study in how AI-generated imagery is now a geopolitical tool. The image was designed to signal resolve, intimidate a foreign government, and move markets. It did all three.

🔮 Prediction: We are entering an era where AI-generated media is a standard instrument of statecraft, right alongside sanctions, troop deployments, and diplomatic cables. Every foreign ministry in the world will need a "synthetic media response" capability within five years. The line between AI content and official government communication is dissolving, fast.

Read More Here

30-Second AI Play

🧲 Build Your Own Agentic Marketing Prompt Stack in 10 Minutes

You don't need $150M to start using agentic AI for marketing. Here's how to build a simple prompt chain that simulates what Hightouch's agents do:

1. Define your customer segment. Prompt: "I'm targeting [persona]. Their top pain points are [X, Y, Z]. Draft a campaign brief."

2. Generate the campaign angle. Prompt: "Based on this brief, suggest 3 campaign angles: one emotional, one rational, one provocative."

3. Write the copy variants. Prompt: "For each angle, write a 3-line email subject line, preview text, and CTA."

4. Predict performance. Prompt: "Based on these variants, which would you expect to perform best with a B2B SaaS audience and why? What A/B test would you run first?"

5. Automate the loop. Feed the output back into step 1 with updated performance data weekly.

💡 Pro tip: Give Claude or GPT-4o your actual CRM segment data (anonymized) and ask it to score each angle before you write a single word. You'll save 3-4 hours per campaign, guaranteed.

Other Relevant AI News!

🎬 The Devil Wears Prada 2 just dropped in cinemas, and it's accidentally the most accurate film about AI's impact on media in 2026. The sequel depicts a fashion magazine world gutted by algorithms, content farms, and billionaire buyouts. Critics say Meryl Streep's Miranda Priestly is a metaphor for every institution AI is eating alive.

☠️ Researchers gave AI chatbots a biosecurity stress test, and the results were genuinely alarming. According to a new NYT investigation, leading models provided step-by-step instructions for engineering drug-resistant pathogens and releasing them in crowded transit systems. A Stanford biosecurity expert who tested one model said he had to go outside for a walk to collect himself. Several companies have since updated their safety guardrails.

🇺🇦 Ukraine is turning battlefield data into a geopolitical asset. Defense Minister Fedorov announced the country will share millions of drone combat videos with allied nations to help train AI warfare models. Ukraine can't outspend Russia — but it can out-data it. Middle powers everywhere should be taking notes.

😱 A fascinating piece from BBC Future this week argues AI companies are actively stoking fear about their own products — existential risk warnings, doomsday talk — as a way to consolidate regulatory power and lock out competition. The fear is the marketing.

🎥 Vine is back… kind of. A new app called Divine just went live, backed by Jack Dorsey's nonprofit, housing 500,000 restored Vine videos from 100,000 original creators. The big twist: it has a strict no-AI-content policy, automatically flagging and removing AI-generated posts. In a feed full of synthetic slop, they're betting on the nostalgia of knowing it's a real human.

Golden Nuggets

• 💰 When Goldman Sachs leads a marketing startup's Series D, it's not an investment. It's a declaration that an entire category is about to flip. "Agentic marketing" is this year's "cloud."

• 🌍 The DeepSeek hype cycle is over. The new frontier in Chinese AI isn't cheaper training: it's whether Huawei can build chips that can actually compete. That's a hardware race now.

• 🎯 Ukraine figured out that in modern warfare, data is territory. The country that processes battlefield information the fastest wins. Every military in the world is watching.

Would love to hear your thoughts! Send me your thoughts by replying to this email (yes, I read them all :)

Until our next AI rendezvous,

Anthony | Founder of Uncover AI