how OpenAI turned 13k leads into millions

👋 What’s up closers!

This is Prompt Punk — the best AI sales newsletter in the omniverse! 😉

TLDR; inside today’s newsletter

Let’s rip…

Last week, I wrote about “how to get into AI”. This week I thought I’d give a practical list of how I used AI in the last 24 hours.

I am currently on parental leave, so not much “business” being done. But you can get a sense of how AI has permeated my every day life.

I mostly use ChatGPT on my phone in voice mode. I go to Claude or Gemini if ChatGPT fails me. But it rarely does. I pay for all these services.

My wife keeps laughing at how much I use AI. Fair, because it works.

Interestingly I use AI much more than I previously used search, because I get answers faster.

Cognitively, I am switching from “problem solving” to “asking good questions”.

Here’s how I put it to work in a single day:

These are just some examples.

What’s the point in sharing this? To help you think broadly about using AI.

How do you use AI?

Marketplace app Too Good To Go fed sales/support notes into Mistral on Bedrock to decode why partner stores churn.

Result: Processed ~14k monthly touch points with 85–90% multilingual accuracy. Discovered a new store segment needing different onboarding. Learned about confusion retailers were having with scheduling sales thus tweaked their comms and product. Significant impact on retention playbooks.

How they win
They didn’t just “add AI” they wired it into workflows. Bedrock handles data + ops, Mistral delivers cost/perf and strong multilingual categorization inside AWS.

Why it matters
The potentially messy “voice of the store” across thousands of customers, becomes a prioritized backlog. Fix onboarding, fix scheduling, keep stores longer.

OpenAI wired an AI sales agent (“Tailor Assist”) into its inbound sales “contact us” forms. They get 13,000 inbound leads a month, but can only touch 1,000 with a human. Here’s how it works…

The agent ingests internal policies, sales docs, and case studies, then evaluates each form submission to recommend the right product and write a personalized email that offers help across support or purchasing.

Result

How OpenAI won
They operationalized the sales process.

Why it matters
Inbound becomes a programmatic, high-trust revenue engine. Buyers of large, complex products get answers that feel handcrafted, at scale—so they reply faster and buy sooner.

What it does: Natural-language data analysis that ships answers. Upload CSV/XLSX/PDFs or connect Google Drive, then ask questions and get charts, forecasts, and reports. No SQL required.

Example: “What’s our CAC by channel last quarter and a forecast for Q4?” → Julius cleans the data, runs the math, and returns a chart + commentary you can paste into the deck.

Why it’s valuable: Turns ad-hoc analysis from a 3-hour favor into a 3-minute chat, so sales, ops, and finance stop queueing behind the data team.

Website: julius.ai

What it does: Lightning-fast AI dictation that works in any Mac app. Hit a hotkey, speak, and Willow writes polished text. Auto-punctuation, grammar fixes, filler-word cleanup, even adapts to your writing style.

Example: Draft a prospect email, meeting notes, and a LinkedIn post in one take—Willow formats it as you talk so you can hit send before your coffee cools.

Why it’s valuable: If your job is words for example outbound sales, doc writing, tickets. This saves a tone of time.

I actually built my own similar app before I knew about Willow - https://no-more-typing.com/

Website: willowvoice.com

Let’s be honest. OpenAI and Anthropic sit in the same mental bucket as the Mag7. Apple, Microsoft, Alphabet (Google), Amazon, Nvidia, Meta and Tesla.

They are no longer “scrappy labs.” Not potential M&A targets. Big Tech. Full stop. Despite the best efforts of some (cough cough - Meta).

Why are they Big Tech? Because they crossed the line from products to power.

What changed?

“But don’t Microsoft and Amazon own the joystick?”

No. They’re symbiotic. Azure and AWS monetize the surge, OpenAI and Anthropic keep the product agenda. Both labs hedged (multi-cloud, direct channels), and both hyperscalers hedged (multiple model catalogs). Everyone needs everyone, by design.

What to watch (the real risks)

Why they’re Big Tech now( the checklist)

What it means for you

In a word “more”. More opportunity, more work, more to learn, more to trip you up. More to regret?

What do you do with more? Cease it!

Stop watching, start doing. Stop asking yourself “is this a bubble” and start building.

If you’ve learned anything from this newsletter it is that we still have a LONG way to go with AI. At least 5-8 years in my opinion.

Get after it.

You knew it was coming. AI Ads are here!

The moment you ask Meta AI for “Christmas gift ideas for 4 year olds,” expect your feed to start looking like a kindergarten.

Meta will start using topics from your AI assistant chats to tune what you see across Facebook and Instagram, including ads.

There’s no real opt-out other than not using Meta AI. Sensitive topics are excluded and the rollout skips the EU, UK, and South Korea (for now).

Why it matters:
Clear buying intent is great for advertisers. You get that with AI. It should lower CPAs (Cost Per Acquisition) for all your company’s ad buying. As for consumers….better ads? 🤷‍♂️

Anthropic named Chris Ciauri (ex EMEA Google Cloud President) as MD, International and will triple headcount outside the U.S., with hiring sprees in Dublin, London, Zurich and more.

Who’s paying for this? Your engineers banging on the Claude Code API. Anthropic’s revenue has surged to a >$5B run-rate from $1B in January. That means more local reps, more local hosting, and a lot more competitive pressure in Europe and Asia.

Why it matters:
Local coverage wins regulated, multilingual accounts. Expect more Claude reps calling your customers and more competition.

You have to admire OpenAI’s product velocity. This week they unveiled Sora 2, an AI model that can make video and audio synchronized with sound effects. Think short clip videos like you see in TikTok. This is a major technical breakthrough. Previous models have struggled to master human body physics (running, twisting, diving). Sora 2 is pretty much there.

See a Sora 2 video here.

Included in the launch was the Sora app (invite-only for now) that lets users create and share clips, including cameos. An API is on the roadmap.

Why it matters:
If everyone can create videos at scale with a few prompts, imagine the possibilities. As someone in sales, I’d imagine customized videos for customers, showing how they could benefit from the products I sell.

Or simply, make cool birthday videos for my kids. Incredible.

No more waiting for GPUs! Nebius (Amsterdam-based, European Neo-Cloud) just locked in a five-year, $17.4B capacity deal with Microsoft. Nebius is following the same playbook as CoreWeave popularized in the U.S., now with a European accent.

Buy GPUs wholesale, sell retail.

Why is Microsoft doing this? Skips the risk and brain damage of building and operating a datacenter. Just in case there’s a bust. 🙄

Why it matters:
Neo-clouds mean more compute and more AI. More competition is great for consumers. And hopefully, no more throttling on Claude and ChatGPT.

Instead of googling, clicking five tabs, and hunting through a bunch of random e-commerce sites, you can now buy inside ChatGPT.

OpenAI’s new Instant Checkout has begun its rollout in collaboration with Stripe.

If you ask ChatGPT for “a minimalist leather wallet under €50,” it won’t just show options, it can take your money and place the order right there. Incredible.

Why it matters:
If shopping starts (and ends) in chat, fewer commercial queries detour through Google or marketplaces. For e-commerce merchants this is a brand-new, high-intent channel with less friction, a cleaner attribution story. Gold.

Cinematic AI generated portraits on Google Gemini

“A cinematic close-up editorial portrait of a slim adult man with the user's face as the main subject, dark blazer, and black turtleneck, very low angle chin tilted upward, direct gaze with subtle smirk, seamless vivid orange studio background, 85mm lens, f/4, ISO 100, shutter 1/200s, cinematic dual-tone lighting with warm orange glow and cool blue rim highlights.”

Voila!

Thanks for reading! 👋

If you liked this:

📨 Forward to a biz techie friend
💬 Hit reply and tell me what you want more (or less) of
📊Vote in the poll below 👇

— John
Prompt Punk