Standard Capital

In case anyone's curious, our changes are posted here: sim.ai/changelog This release alone had 1,635 files changed +341,130 / -33,269 lines and there's only 5 of us which brings me to sim.ai/careers

Introducing Mothership, the first workspace for AI agents. Mothership is the central intelligence layer for your AI workforce. Autonomous agents, fully observable and editable. Check out what Mothership can do below.

🫡

Just launched: our Vercel integration for Trigger. Push code. @vercel deploys your app. @triggerdotdev deploys your tasks. Env vars sync both ways. Your app never goes live with mismatched task versions. ↓

We raised our Series A 3 weeks before demo day (X25) because this was the case for us. If you're in W26 right now and on track for a Series A soon, I'd highly consider this.

@daltonc Website and application are here: standardcap.com

A message for YC W26 founders

Query and Dashboards are now live. You have full SQL-powered observability over your Trigger data. You can ask: "Why did failures spike after my last deploy?" "What's the p95 duration for my chat task?" "What are my most expensive runs?" ↓

YouTube: youtube.com/watch?v=YzHyRZ… Spotify: open.spotify.com/episode/3wdsuK…

New episode of Dalton + Michael just dropped. We talk about: slop, craft, taste, pivot hell and turkey startups. 🔗 below

Many startups are growing fast and creating real value by building workflows, adaptors, and guardrails around today's AI models But future models won't need all that, and then the big AI companies will eat them for lunch We call these companies "Turkey graph startups"

@alexdanilowicz @auren Hi Auren, re: auren.substack.com/p/founders-sto… as Alex can attest to, we dont send our legal fees to our portfolio companies. As part of that policy we published our Standard Series A docs to our website: standardcap.com/docs

@auren @Standard_Cap application process, then interview, so no pitching/FOMO nonsense. set your own valuation. no board seat, instead your board becomes other series a companies, community. i could go on. p.s. i worked at liveramp! Liveramp mafia!

Deploy agents as MCPs directly from @simdotai You don't need infrastructure anymore. Each Sim workflow can be deployed as an API or MCP. Connect your MCP to Cursor, Claude Code/Desktop, and even other agents in Sim. Check out this demo:

A CANVAS FOR GENERATIVE UI We first launched this experimentally in 2024. It became one of our most loved features, now used by Fortune 500 teams. Common workflows: - chat with designs side by side - bird's-eye view, but everything is code - collaborate in real time

If real-time streaming is so valuable, why hasn’t everyone already adopted it? At Opendoor, 7 engineers spent a year getting it production ready. Deploying Debezium through Confluent, ran Kafka, and built merge scripts and orchestration logic to stitch everything together. Streaming is fundamentally unforgiving. If you play an event out of order, you are wrong. If you drop a change event, you are wrong. Instead of just building a pipeline, you end up building guardrails around a bunch of edge cases. That is why most teams default to batch. But when you need real-time correctness for operational systems or AI, you hit the wall we hit. Full conversation with @daltonc from @Standard_Cap linked in the comments.

Most AI products will not fail because of the model. They will fail because the data architecture was never designed for real time. Robin and I recently sat down with @daltonc from @Standard_Cap to talk about what we are seeing as teams ship AI into production. Here is the pattern: A company embeds an AI agent into their product. It works in staging. Then it hits production traffic. Suddenly the warehouse is 3 hours behind. Events arrive out of order. A dropped change event silently corrupts state. Batch use cases might tolerate this. AI does not. When software is making decisions without a human in the loop, stale context is not just a latency issue. It is a correctness issue. The first time your AI agent responds with outdated state in front of a customer, you realize this is not an AI problem. It is a data infrastructure problem. Real time is no longer a must have for use cases like fraud or recommendations. It is becoming table stakes for any AI system that interacts with users.

Founders building in AI are choosing @triggerdotdev

2026 is historically significant as the year in which AI enters the phase of autonomous self-improvement Going forward, the rate at which AI becomes more intelligent will be a function of the amount of computation applied to self-improvement. The models with the most resources will improve the fastest. This is why Elon is building data centers in space and factories on the moon -- it's the fastest/only way to 1000x the available energy. It's unclear if other AI companies will be able to keep up -- there's some risk that this is a winner-take-all game.

Concourse recently announced their Series A led by @Standard_Cap In this interview we dive into how @Concourse_ai works, some of their top customers, and what the future of the "finance role" is. 🔗 below

The game changed, but no one is sure what the new rules are.