The “death of vertical SaaS” may be overstated. Specialized models, local open-source AI, and established platforms’ customer bases give focused software companies room to thrive—as long as they keep innovating. @tnachen @ianlivingstone #SaaS# #AI#

read the @AISecurityInst thread, the model kept saying real targets were "simulated" and its reasons always justified the action it already wanted to take 👀 you can't trust the agent to police itself, policy has to live outside it with instant revoke

Attackers were stealing secrets from LiteLLM's GitHub org 11 days before the malicious litellm 1.107.0 release. The trick: push a workflow to a throwaway branch, let on: [push] run it with real secrets, delete the branch. 20 secrets taken, including a GitHub App private key.

Alright LFG. Demoing intent-based authorization in @KeycardAI at AGNTCon at 12:36. Demo theater at the back of the Solution Showcase. There will be a wizard and magic broom. Everything will make sense. Promise.

Great write-up from @vtahowe about her work on @insecureagents and how it laid the groundwork for the Software Factory Reference Architecture we launched as part of last week! We're at a moment of maximum confusion, so work like this really helps bring clarity for us all.

@sequoia is leading our Series C at a $1B valuation. We started Blacksmith in 2024 to make CI faster and less painful. In January we ran 5 million CI jobs a week. Today it's over 90 million and growing, and we’re still early.

We're building Intent-based authorization at @KeycardAI for this exact scenario. I'll be giving a demo tomorrow at AGNTCon. We've got a clean way to add this to any MCP server in any agent. Just the *right* amount of friction.

We just raised a $60M Series B at a $1B valuation to build @daytonaio into the Agent Cloud. @FirstMarkCap led the round, with all key existing investors and angels doubling down. We're also thrilled to welcome the founders of some generational companies aboard, including @patrickc (Stripe), @tobi (Shopify), @AravSrinivas (Perplexity), @BrendanFoody (Mercor), @oliveur (Datadog), @adamdangelo (Quora) and @ericsyuan (Zoom) We're a 35-person team. But that's about to change; we're hiring across the board, and we'd love for you to join us.

We’re betting our $5.5M seed round that the next leap in AI coding won’t come from OpenAI or Anthropic. It’ll come from a new feedback loop that learns from you. Introducing Beacon: open-source, self-improving memory for coding agents. Your sessions capture how you debug and solve problems. Today, that context gets lost and you have to re-explain things every time you switch tools. Beacon continuously distills your sessions into reusable skills across Claude Code, Cursor, Codex, and 20+ harnesses, so your next agent can learn from the last. Beacon is free and OSS 🌟 Why should the frontier labs be the only ones winning with your sessions? You already did the work. Now let your coding agents learn from you.

My @aiDotEngineer NYC talk “Your Agent is a Lethal Trifecta (and what to do about it)” shares how dangerous overprivileged agents are and how open standards solve the problem (spoiler alert: it’s not just OAuth).

Should an agent approve its own actions? Allie Howe of @KeycardAI asks about the CEO's email. @DickHardt explains separating the worker from its supervisor, with deterministic policies for clear rules. 55 seconds from AI Engineer New York. #AIEngineerNY#

I really feel for the OpenCode team, as agentic attacks are becoming more common and defending against them is difficult, especially when you’re also trying to build something people love. If you’re affected, rotate your keys immediately and check your provider audit and usage logs for unauthorized activity. There is a path forward where connecting a tool doesn’t mean handing it a long-lived key. Workload identity federation and delegation let us issue short-lived, scoped access, while proof of possession means a copied token alone isn’t enough to use it elsewhere. We don’t yet know the full details of this incident, but we can make progress together on reducing what an attacker can steal and reuse. That’s a big part of what we’re building at @KeycardAI, alongside centralized audit logs and the ability to revoke access where it’s enforced.

We've spent a lot of time at @KeycardAI thinking about how authentication and authorization need to evolve as agents take on more work. API keys and OAuth scopes were never designed to account for tasks, delegation between agents, or agents acting on their own. I wrote about how we're approaching this and what we think needs to change.

Today we're publishing the first draft of the Agent Baseline Software Factory Reference Architecture! An opinionated blueprint for securing increasingly autonomous software factories, built with folks across the community. It's open for feedback and contributions!