thanks @gradypb
crisp as a pringle
The @BostonCollege Investment Committee (an LP and my beloved alma mater) asked for a few thoughts on what's happening in AI. I recorded a test run yesterday morning and then shared it with my partners, who encouraged me to share it more broadly... so here you go!
This is not a sales pitch, it's just a reflection on what we're seeing. And it wasn't intended to be shared, so please pardon the rough edges.
loom.com/share/c016702964a04…
41
People used to vote for who they want in office.
Today, they vote against who they don’t want in office.
This is why the polls are getting more wrong over the past few elections
Asking who you would never vote for is more telling than who you’d consider and why
12
All roads lead to FDEs
Good post on how to think about process redesign in an enterprise with AI Agents.
Unfortunately for most workflows, there’s no “easy” button. The best way to get real efficiency gains is you have to actually reengineer the process to take advantage of what agents are actually good at and can do differently and faster than people.
One of the challenges of course is that the most important workflows actually space many functions at once, so someone needs to be able to go in and retool how work happens.
“Nobody in that chain is empowered to walk into finance, for example, and say that their 14 step process should actually only be 5 steps. So nobody does it. Instead, you 'apply AI' on what you currently have, and you end up with faster sh*t.”
This is the road ahead for AI diffusion in enterprises. This also presents the clear opportunity for companies - both at the applied AI layer (regardless of the article’s title) and teams or companies that can go and implement transformation into the enterprises. And all roads lead to FDEs one way or another.
108
This is the most true thing I’ve read today.
If you have the fortunate place of being a SaaS with distribution, you better be building that MCP and simplifying the UI into a SaaS-specific harness that allow you to master the 0 to 1 context problem in your domain.
This is not financial advice, just survival tips.
The fact that SaaS co's are giving us agents instead of MCPs shows just how self-centered / narcissistic they are.
I never wanted to "live in your product." Your product was the best way to get something done. If the best way to get it done is now headless, please let me do that.
98
Had a similar moment this year.
There are still some things I like to do locally, but agentic coding via cloud based sandboxes has so many advantages
98
Kevin Denman retweeted
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.
The world needs both frontier closed models and frontier open models.
images.nvidia.com/pdf/Open-W…
16,028
29,391
171,953
66,428,678
Agentic Capital lives in the harness and the data
I couldn’t agree more with @chamath on Harnesses being the place where value accrues
Business need to own their entire AI stack around a model as their core business asset but also to drive unique outcomes and applications
I wrote an entire thesis on this exact topic here: x.com/shaughnessy119/status/…
I also agree with him on the death of tokenmaxxing and that negatively impacting frontier model revenues (and margins) but I felt it affects the entire AI funding engine and stock prices and so far it has
Thesis on this topic: x.com/shaughnessy119/status/…
6
1,544
When you’re building an agent (or assistant) that you expect other people to use, you have to wire it for cross-harness deployment.
You still have your own harness for the CI/CD pipeline, but you need to assume significant use will come from elsewhere.
4
195
If you are building service-adjacent software, make sure it comes with services.
The age of “make one copy of software and sell it to everyone” is coming to an end.
You need to deliver the full stack - enabling software and services.
The software aspect used to be the leverage point, now it’s everything else.
Harvey is valued at $11B. Legora just raised at $5.5B. I built their entire web application in two weeks and I'm making it open-source and free for everyone to use. Say hi to Mike: mikeoss.com.
When I got the chance to try Harvey and Legora, I was surprised by how simple they were. A thought came to mind: I could probably build something similar in no time at all with Claude. And so I did.
Assistant, project, tabular review and workflows. You get it all without vendor lock-in.
Mike offers law firms an alternative, where they own the application layer and aren't stuck with a vendor they're renewing forever.
You can try Mike in the demo on the website, or go to the GitHub link on the site to download the code and run a local version yourself.
156
Thoughtful design and implementation of agentic systems is much harder to find than you’d think.
The AGI-pilled think the work just magically happens.
The real builders understand that agentic systems need to be thoughtfully designed with business process, existing systems integration, and workforce change management as major factors.
The foundations of the agentic enterprise are not the agents as much as they are the enterprises ability to empower agents to add value.
As they say, ask not what Claude can do for you, ask what you can do for Claude.
Getting requests from clients for real AI implementation partners.
All I'm finding are Vibe Code Bros or Zapier shops.
I want firms that:
• Diagnose the actual business problem
• Bring PMs + Product + AI talent
• Build + integrate into real workflows
• Care about security and stability
• Ship and iterate
Who’s best in the world at this?
2
83
Cursor may be the most import RL environment there is right now.
We believe Cursor discovered a novel solution to Problem Six of the First Proof challenge, a set of math research problems that approximate the work of Stanford, MIT, Berkeley academics. Cursor's solution yields stronger results than the official, human-written solution.
Notably, we used the same harness that built a browser from scratch a few weeks ago. It ran fully autonomously, without nudging or hints, for four days.
This suggests that our technique for scaling agent coordination might generalize beyond coding.
1
105
I could not emphasize this enough to builders.
I’m about 3000 hours in to agentic coding.
If your product (API / CLI / MCP / Docs / Templates) can’t talk to my agents it won’t get into the stack.
Edge cases: writing C++ directly on Ethernet cables to optimize AI throughput
Karpathy is telling you something most product teams haven’t internalized yet.
The new distribution channel for software is agents. Agents don’t browse your marketing site, watch your demo video, or click through your onboarding flow. They call your CLI. They hit your MCP server. They read your docs programmatically. If none of those surface areas exist, your product is invisible to them.
Look at how fast this moved. MCP went from zero to 97 million monthly SDK downloads in twelve months. 10,000+ active servers. OpenAI, Google DeepMind, Microsoft, and Cloudflare all adopted it. By December 2025, Anthropic donated MCP to the Linux Foundation because the standard had already won. Running an MCP server is now compared to running a web server.
That’s the new baseline for product discovery.
85% of enterprises are expected to have AI agents deployed. Those agents need structured, programmatic access to your product. They need CLIs, MCP endpoints, and machine-readable documentation. A beautiful React dashboard is worthless to an agent trying to pull data into a workflow at 3am.
This tells you everything about why Karpathy’s framing of CLIs as “legacy” technology is so precise. Legacy means battle-tested, standardized, universally parseable. stdin/stdout, flags, JSON output. The entire Unix philosophy was accidentally designed for AI agents decades before they existed.
Your competitor ships an MCP server and suddenly every Claude Code user, every Cursor session, every autonomous workflow can discover and use their product. No human ever visits the website. No sales call. No onboarding email. The agent just finds the tool and starts using it.
The companies that win the next 24 months are the ones building agent-accessible surface area right now. The ones that lose are still optimizing their landing page above the fold.
12
4
255
Unlike models, agents have access to a dynamic set of data, tools, and context.
In this way, agents can tackle much harder problems than the underlying models if they are architected well.
We rebuilt how our agent uses context.
Instead of stuffing everything into a prompt, Cursor dynamically discovers context via files, tools, and history, cutting token usage by 46.9% and freeing up more space for the agent to work.
1
77
I find this n8n vs LangGraph comparison very helpful.
If you are building agentic apps, you need a LangGraph-like framework.
It you are building a bunch of integration for make existing apps agentic, a visual workflow builder is probably enough.
Everyone is building AI agents.
Very few understand the agentic frameworks that actually power them.
In 2025, two frameworks dominate agent development —
not as competitors, but as complementary layers:
n8n — Visual Workflow Automation
What it does
• Visually connects AI agents with business tools and APIs
• Flow: Trigger → AI Agent → Tools → Action
• Removes integration complexity and speeds up deployment
Think of it as:
The orchestrator that plugs AI into your entire tech stack
—
LangGraph — Graph-based Agent Orchestration (LangChain)
What it does
• Enables stateful, cyclical, multi-step agent workflows
• Flow: State → Agents → Conditional Logic → State (loops)
• Designed for complex reasoning and coordination
Think of it as:
The brain managing advanced agent decision-making
—
When to use n8n
• AI + business tool integrations
• Customer support and ops automation
• No-code or low-code workflows for teams
• Fast shipping with 700+ integrations
When to use LangGraph
• Multi-agent reasoning systems
• Enterprise-grade AI applications
• Cyclical or long-running workflows
• Fine-grained state control and memory
—
Ecosystem strengths
n8n
• Visual builder for non-developers
• Self-hosted, open-source option
• Strong business automation community
LangGraph
• Deep LangChain integration
• LangSmith for observability and debugging
• Advanced state persistence and control
—
The real insight 👇
The best AI systems use both.
n8n → Visual orchestration and tool integration
LangGraph → Agent logic, reasoning, and state
Think in layers: business automation and intelligent decision-making
—
Your turn 👋
What would you build first?
A visually simple, tool-connected agent (n8n)?
Or a deeply orchestrated, reasoning-heavy agent (LangGraph)?
1
91
