If what you've built doesn't clearly link to ROI, it is not going to survive the upcoming cutbacks
It's easy to build working demos with broken unit economics
Commodity operational agents will move to managed service providers because cost management will be the discipline that matters
Even with employer caps, the spend on AI tokens dramatically exceeds any other historical spend on software.
Typically, companies maybe would spend on the order of $10-50 for a software license per month per employee, but now will pay hundreds or thousands on tokens to augment their productivity.
This shows you how big the TAM for intelligence is in the enterprise. The markets for AI are going to dramatically expand the size of the traditional software markets over time.
agent-first is the way - if i am onboarding to your product and can't ask an agent to complete my goal... that's the last you'll see of me in your product
(unless you're slack. or hubspot. or google docs 🤔)
I wish Slack was:
- Agent-first
- Beautiful to use
- Integrated with agents natively so your Hermes or OpenClaw lives inside it
- Huddles worked seamlessly and were fun
- Built for teams of 1-3, not just teams of 300
- Truly a second brain similar to Obsidian
- Searchable without wanting to throw your laptop
- Designed around async, not constant interruption
- Voice first for mobile
- A place where I could see who's working on what right now without asking anyone
- Smart enough to know the difference between "I need you right now" and "whenever you get to this"
- A workspace where my agent could tap someone else's agent on the shoulder and coordinate without involving either human
- Designed so the new hire on day 1 has the same context as the person who's been there 3 years
-Something that felt like walking into a room of people building, not walking into a room of people typing
- A place where decisions are first-class objects
- Able to auto generate SOPs, skills, agents etc from conversation history
- Something that rewards deep work instead of punishing it with 47 unread notifications
@dexhorthy@walden_yan@tobi@karpathy Requiring that your workflows be MECE is exhausting and doesn't scale for real work. Try building appointment scheduling and you'll see what I mean
Workflows that are agent-first but have deterministic tools is how you can get all those edge cases handled for free
If your AI solution providers are markups on token costs, you're in for a rude awakening
Credit models will stop making sense if you can't cleanly tie token value to your target business metric
AI is getting expensive. Fast.
Microsoft just cancelled internal Claude Code licenses after costs spiraled.
Uber's COO publicly said it's getting "harder to justify" AI spending with no clear link to useful product improvements.
And we're still in the early innings.
This reminds me of Uber and Airbnb's early days.
They stayed private. Kept prices artificially low. Burned VC money. Won over consumers. Took out legacy businesses.
Then the VCs wanted returns. Prices went up. Fast.
Here's the wild part: we haven't even hit that moment yet with AI.
OpenAI and Anthropic are still burning cash at a historic rate. No public market pressure. No mandate to profit.
OpenAI is projected to lose a cumulative $100B+ through 2028. Holy ****.
The prices you're paying today are the subsidized version.
So what happens when they flip the switch?
The companies winning right now are the ones building ROI discipline before it's forced on them.
Not waiting for a CFO mandate. Not scrambling when the bill triples.
IMO - the AI efficiency era is just starting.
Thoughts!?
@levie This is the consequence of products that aren't built for deployments. If you choose the right agent orchestration platform, you don't need FDEs - they just slow things down
Bloomberg: OpenAI launches a $ 10Bn joint venture called “The Deployment Company” to help businesses use its AI.
The new company, The Deployment Company, has raised more than $ 4B from 19 investors, including TPG, Brookfield, Advent, Bain, SoftBank, and Dragoneer.
The basic bet is that AI adoption is no longer mainly a model-quality problem, because many companies already want AI but lack the teams, workflows, data access, security rules, and operating discipline to install it safely inside real business processes.
Private equity firms are useful here because they control or advise large webs of companies, and the report says OpenAI’s partners can reach more than 2,000 portfolio companies and clients.
That turns enterprise AI selling from one-company-at-a-time pitching into a routed distribution system, where OpenAI can package software, consulting, deployment playbooks, and sector-specific use cases across finance, healthcare, coding, operations, and support.
The deeper technical point is that LLMs do not create value just by answering prompts, because they need to be connected to company data, permissions, tools, evaluation systems, and human review loops before they can affect revenue or cost.
Anthropic also is building a similar PE-backed route for Claude, which suggests the next AI race may be less about demos and more about who can industrialize deployment fastest.
---
bloomberg. com/news/articles/2026-05-04/openai-finalizes-10-billion-joint-venture-with-pe-firms-to-deploy-ai
Investor update I just got:
“We’ve finally internalized that people are bad at using software. To fix this, we are building agents that use [our product] for them.”
It’s working.
The company is growing rapidly.
64 vs. 33. A lifetime of hard work vs. a silver spoon. The results speak for themselves.
The weight of the job is too heavy for “Mamscrawny.” The only thing he can lift is your taxes.
@n_dof_droyd@pdhsu Yeah it's probably high friction to switch to a business account, but you can get itemized receipts that way
I do think if you actually provide enough value (real or perceived), something in this category will take off now. People want narratives, not data