How to sell AI

👋 What’s up, pipeline punks —

Welcome to Prompt Punk — the AI business newsletter for people that actually have to sell stuff.

TLDR; inside today’s newsletter

Let’s rip…

How’s your AI pipeline?

I am not talking about selling GPUs to hyperscalers. That’s easier than selling crucifixes to Catholics.

I am talking about selling enterprise AI transformations. A change sale, not a tech sale.

MIT says 95% of generative AI pilots are failing, but is it accurate?

Why selling AI is hard

Even the kid-geniuses are giving up on building “AI Transformation Consultancies”.

Brendan Falk, who exited his last company to AWS explains it all:

So what’s the wining formula?

You need a “compelling event” + “tech sales” + “clear ROI” + “change management” (what happens to people afterwards).

A rare combination.

When it does work, I see this:

MIT agrees with me. The most successful transformations buy off the shelf software and start with “how will this integrate into our existing workflows”.

What’s working for you? Reply and let me know.

A few years ago, people duct-taped Zapier + OpenAI to run “agents.” Most quit. n8n didn’t.

Now it’s valued at $2.3B on ~$40M ARR — a wild ~58× multiple. Their play? Automating back-office drudgery:

Why it matters:
n8n proves buyers want off-the-shelf AI SaaS, not DIY API spaghetti. If you’re selling AI, which camp are you in?

Databricks is raising at a $100B valuation to fund:

CEO Ali Ghodsi pegs DB TAM at $105B, declaring: “The new user is an AI agent.”

Why it matters:
More proof for AI sellers, that customers want apps, not APIs. Databricks wants to power apps.

For technical folks, if agents create most new DBs, the control plane shifts from humans to software. That rewires vendor choice, pricing, and workload design — and Databricks wants the stack.

An M&A firm asked me to build an AI-native CRM for $20k. Funding Stack CRM was too expensive, Salesforce too complex. It was an easy build with Lovable, no code needed. Not enterprise-grade, but workable.

Reminds me of the 90s when kids built websites for local businesses. There’s even a subreddit full of “AI agent hustlers.” Check it out, great ideas there.

Why it matters:
We’re in the agent-agency era. Non-technical founders can launch vertical AI tools overnight. Just like early web shops, a few will become empires.

Bacardi just turned the internet into a dashboard. They wanted real-time competitive market intelligence.

What drinks are bars serving?

They use startup Jsonify’s AI agents to continuously crawl the internet and scan 180,000 online bar menus. My fav AI prospecting tool.

The result? Now they can spot gaps in venues and cities, aiding their sales machine. Create better pricing and promos. Make decisions in weeks not quarters.

Gotta love AI, now it has found a way to get us drunk (Bacardi 37.5% ABV 😋)

Why it matters:
If you’re a strategist, seller or marketer that needs to capture and analyze information on the internet like a human, check out Jsonify.

The UK government has 1,700 people in 22 locations running around trying to collect debts owed to the state.

Since the UK owes £2.7 trillion, maybe GenAI can help pay down the debt. They have awarded a contract to help employees track down debtors and work efficiently with AI.

Don’t laugh, but debt collection is an unexpected high growth AI use case. A16Z just put $60M into Salient. Well-known UK entrepreneur Joshua March, just launched Veritus Agent.

You can guess how it works: An AI voice agent calls you a million times until you pay up 😂

Why it matters:
Don’t forget the public sector, they need AI Agents too!

In July, Microsoft’s Chief Commercial Officer stated the company had saved $500m in call center costs thanks to AI. The company also laid off 9,000 workers across various teams, geographies and levels. We’ve seen a slate of these announcements over the year, but it’s unclear if AI is replacing workers or a narrative for budget reallocation.

Let’s look at some data.

Why it matters:
Core AI (foundational models, high-growth apps, AI consultants) are growing headcount. “Legacy” big tech, is largely reallocating budget, using AI as a narrative to drive change.

AI is substituting jobs at the margin (Level-1 customer support, transcription) but also eliminating entry level positions that could have existed.

If you’re looking for a new job in AI, focus on teams within companies receiving investment and winning customers.

For everyone else, avoid being a new grad or mid-level professional “stranded” in a non-investment area.

OpenAI might be pulling an AWS move. CFO Sarah Friar expects they could rent out their AI-optimized data centers — the same way Amazon turned spare capacity into a $100B+ business. It’s “a business down the line, for sure”.

Right now, all capacity is eaten by ChatGPT and enterprise demand. But with Sam Altman talking about spending trillions on infrastructure, you gotta expect some spare capacity.

They’re already in deep with Oracle, SoftBank, and private equity to finance mega–data centers under “Project Stargate.”

Why it matters:
If OpenAI becomes an infra landlord, it’s no longer an app — it’s a platform. That shifts what we build, buy, and sell across tech.

Paradigm is reinventing the spreadsheet with AI. Instead of formulas, you ask plain-English questions, pull in live data, and run agents for pipeline analysis or competitor tracking.

Microsoft’s reply? COPILOT() in Excel — neat for cleaning and summarizing, but stuck inside your sheet.

Why it matters:
If you’re in sales, revops or strategy, these tools cut grunt work and give you faster answers.

Last summer my neighbor bought a Chinese car. Everyone freaked.

Now half the neighborhood drives one.

Will AI follow the same path?

It’s already starting. Startups are quietly deploying on-premise Chinese models for high-volume jobs like doc classification — saving 90%+ on costs.

The Chinese menu:

Western benchmarks:

IMPORTANT: Chinese APIs may send your data to the Chinese Communist Party! Local downloads are safer. Expect censorship (no Tiananmen questions for your customers).

Why it matters:
Chinese AI is nearly as good, and orders of magnitude cheaper.

Western players still lead in raw quality today, but by 2028 China could leapfrog (unless Elon/Google pull a rabbit out of the hat).

If you’re selling AI, being the “China model expert” could be your edge.

👉 Have you even tried Qwen? chat.qwen.ai

That’s it peeps.

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