Dan Krieger

co-founder & cpo @raisi_ai

building infrastructure for private capital markets

[Dan Krieger @dankrieg]

Karpathy said people who don’t use LLMs are already losing

The next step is learning how to build with them

At first you ask the model for help
Then you give it context
Then you give it tools
Then you give it a goal
Then you design the loop around it

That loop can run through a whole company:

- marketing watches market signals
- product learns from users
- customer success monitors account health
- sales learns from objections
- operations catches bottlenecks
- management coordinates the system

Each loop has the same shape:
1. observe what changed
2. decide what matters
3. execute the routine
4. verify the output
5. escalate the judgment
6. learn from the trace

This is where LLMs become more than assistants.
They become part of the company’s operating system.
That’s what my research paper is about: loop engineering for autonomous companies.
Watch the clip, then read the paper below.

[Dan Krieger @dankrieg]

https://t.co/xebQuN3yys

[Dan Krieger @dankrieg]

Boris Cherny: “I don’t talk to an agent anymore. I talk to a loop.”

That idea gets much bigger once you apply it to a company.

- Marketing can run as a loop
- Product can run as a loop
- Customer success can run as a loop
- Sales can run as a loop
- Management can run as the loop that coordinates the others

Each loop has the same basic shape:
1. observe the signal
2. set the goal
3. execute the routine
4. verify the output
5. escalate the judgment
6. learn from the trace

That’s how recurring work keeps moving without a human pushing every next step.

[Dan Krieger @dankrieg]

https://t.co/xebQuN3yys

[Dan Krieger @dankrieg]

Prompt engineering was about asking better questions.

Loop engineering is about designing systems that keep working after the prompt.

A company already runs on loops: marketing learns from campaigns, sales from objections, product from users, CS from account health, management from the operating rhythm.

The autonomous company won’t be built by adding agents to every workflow.

It happens when the company’s real loops become explicit AI-human systems: market signals, customer health, product feedback, execution, verification, learning.

[Dan Krieger @dankrieg]

https://t.co/xebQuN3yys

[Dan Krieger @dankrieg]

https://t.co/xebQuN3yys

[Dan Krieger @dankrieg]

2026 is the best!!! I loove agents

[a16z @a16z]

Against all odds.

Congrats @elonmusk and @SpaceX.

[Dan Krieger @dankrieg]

@a16z @elonmusk @SpaceX go @elonmusk!!

[Chamath Palihapitiya @chamath]

Institutions are inefficient.

Experts are overrated.

Capital should flow toward builders.

Technology compounds.

Independent thinking wins.

[Raisi @raisi_ai]

🏙️ TECH WEEK by a16z NY Tech Week is always a good reality check.

You can spend months on Zoom and convince yourself you know what's happening in venture.

Then you spend a few days in rooms with founders, VCs, family offices, operators, LPs, and institutional investors and realize the market has already moved again.

We believe private capital markets need better infrastructure. Better ways for founders, funds, investment banks, and investors to discover opportunities, build relationships, and move capital more efficiently.

Tonight we're excited to bring together another incredible group of founders, investors, operators, and builders:

Joining us:

→ @brookeleblanc LeBlanc — Head of Community at @UseCorgi ($2.6B valuation)
→ Jeff Frommer — CEO of OWM (Creator Ownership) and former co-founder of MALKA Media
→ Linda Marmora — Managing Partner at a global family office
→ Kevin Mergruen — Managing Director at Bright Data

Most importantly, if you're attending tonight, come say hello to the Raisi Team!

@RobertHarary, @holdbru, @dankrieg , Caroline Fichthorn, Bryan Yankton, Felicia Fox, and Michael Edwards 🤝

[Dan Krieger @dankrieg]

I feel like white-collar workers are becoming LLM wrappers

[Dan Krieger @dankrieg]

You're wasting your time when AI models aren't running for your own benefit

[Andrej Karpathy @karpathy]

This is the the quote I've been citing a lot recently.

[Dan Krieger @dankrieg]

@naval @grok summarize takeaways

[Dan Krieger @dankrieg]

@RobinhoodApp @vladtenev Robinhood is the best

[Dan Krieger @dankrieg]

@Polymarket @elonmusk when is the xAI device coming?

The most secure AI device out there

[Dan Krieger @dankrieg]

@NoLimitGains Phone turning into a real autonomous operator instead of a notification machine feels like the next big shift

[Dan Krieger @dankrieg]

Curious when private equity starts rolling up roofing + plumbing companies at scale…

[GREG ISENBERG @gregisenberg]

What are the best businesses to be in a post-AGI world?

[Dan Krieger @dankrieg]

Bottom line for anyone raising:

The bar is definitely higher than 2021, but real capital is still there for strong teams with actual progress in big markets.

Private capital markets tech especially feels well-positioned as AI starts reshaping how funds and finance actually work.

What are you seeing in your own fundraising conversations lately?

Drop it below 👇#Startups #VC #Fintech #PrivateMarkets

[Dan Krieger @dankrieg]

I just did a deep dive on Seed and Series A fundraising trends from 2020 through mid-2026.

What a wild ride — total boom, then that brutal funding winter, and now this AI-powered rebound.

Pulled the data from Crunchbase, PitchBook, Carta, etc.

Here’s what actually stands out 👇

[Dan Krieger @dankrieg]

Where the market’s active right now (mid-2026):

AI is absolutely eating everything — infra, agents, vertical apps.

But B2B fintech and private markets infrastructure is quietly having a strong comeback — cap tables, fund admin, tokenization, RegTech stuff.

It’s pretty split: the top deals get huge checks fast, everyone else has to show clear proof.