jonathanwall

Runloop is excited to be a launch partner for the @stripe Projects developer preview 🚀 One command provisions a full Runloop account with devboxes ready for agents, full access to pro plan features, and real credentials in your environment. No dashboard and no copy-paste keys 🙅 Simply run: $ stripe projects init my-app && stripe projects add runloop Try it today: projects.dev Demo: youtu.be/89qkfRFWhy0

Very excited for @RunloopDev to partner with @stripe to streamline onboarding. Provision accounts via a CLI single command: $ stripe projects init my-app && stripe projects add runloop Try it today: projects.dev Demo: lnkd.in/dJvtFghr

Auditors haven't figured out AI agents yet. But they will. When they start asking questions like who authorized this action, what did the agent have access to, where's the audit trail, you need answers ready fast. And excuses about it being AI simply won';t cut it. Security engineers tend to front-run compliance. That's the play here: get your logging, isolation, and access controls in place now so you have good answers before the questions arrive. Right now the industry is in a "log everything" phase. That's table stakes. The next question is: can you prove what went wrong and whose fault it was when something breaks? Full conversation: our CEO Jonathan Wall on Techstrong TV — techstrong.tv/videos/latest-… #AIAgents #AISecurity #Compliance #AuditTrail #GRC #DevSecOps #AIEngineering #Runloop

Absolutely inside. Most haven't realized this yet. This is where everything is going. If you're a sandbox provider, here's a free idea of what to do:

Running an agent that uses MCP is inherently unsafe. Until now. Protect your API keys and stop MCP-context bloat before it starts with MCP Hub: the first L7 proxy to protect your secrets and consolidate MCP endpoints. Learn more: runloop.ai/blog/use-mcp-f…

Don't let your agent leak your API keys! Introducing Agent Gateway: an L7 proxy that protects your authentication keys at the infrastructure layer. API keys never enter the runtime environment, so your agent can't leak them. Instead, agents receive one-time tokens and can be restricted to approved endpoints with Network Policies. Zero-trust agents are now possible. Read more: runloop.ai/blog/protect-a…

California’s Union Machine: $921 Million and Counting Public sector unions collect nearly $1 billion a year to control Sacramento. Normal citizens? A trickle against a torrent. But there's a legal pathway to break their grip and get back to sanity. garryslist.org/posts/californ…

AI agents are powerful, but security has to keep up. 🔐 @ZDNET explores how exploitable flaws in agentic AI change the threat model. As our CEO @jonathantwall shares in the article: lateral movement between agents makes the blast radius grow fast, a critical risk orgs can’t ignore. Agent-native security is mandatory in 2026 and sandboxing is just the first step. Read more here 👉️ zdnet.com/article/exploi…

Big moment for Postgres! Search has always been Postgres' weak spot, and everyone just accepted it. If you needed a real relevance-ranked keyword search, the default answer was to spin up Elasticsearch or add Algolia and deal with the data sync headaches forever. The problem isn't that Postgres can't do text search. It can. But the built-in `ts_rank` function uses a basic term frequency algorithm that doesn't come close to what modern search engines deliver. So teams end up: - Running a separate Elasticsearch cluster just for search - Building sync pipelines that inevitably drift out of consistency - Paying for managed search services that charge per query - Accepting mediocre search relevance because "good enough" ships faster But this is actually a solvable problem. You can realistically bring industry-standard search ranking directly into Postgres, which eliminates the need for external infra entirely. This exact solution is now available with the newly open-sourced pg_textsearch by @TigerDatabase, a Postgres extension that brings true BM25 relevance ranking into the database. BM25 is the algorithm behind Elasticsearch, Lucene, and most modern search engines. Now it runs natively in Postgres. Here's what pg_textsearch enables: - True BM25 ranking with configurable parameters (the same algorithm powering production search systems) - Simple SQL syntax: `ORDER BY content <@> 'search terms'` - Works with Postgres text search configurations for multiple languages - Pairs naturally with pgvector for hybrid keyword + semantic search That last point matters a lot for RAG apps. The video below shows this in action, and I worked with the team to put this together. You can now do hybrid retrieval (combining keyword matching with vector similarity) in a single database, without stitching together multiple systems. The syntax is clean enough that you can add relevance-ranked search to existing queries in minutes. pg_textsearch is fully open-source under the PostgreSQL license. You can find a link to their GitHub repo in the next tweet.

Friendly reminder the first German Chancellor to initiate phase out of all Nuclear by 2022 went to work for Russian oil and gas firms when he left office. Say his name. Gerhard Schroeder

You can now upload files, directories, and binary resources directly through the Runloop SDK and have them automatically unpacked onto devboxes before the boxes start using our Object store; now generally available. Try it out and let us know what you think! runloop.ai/blog/object-st…

I must say that the most hilarious twist of tech trajectories of 2025 was the fusion industry realizing they could *double* the economics of fusion reactors by using the neutron blanket to.... Transmute Mercury into Gold. Alchemy is so back

@BrandGrowthOS @langchain Cloud hosted sandboxes let you run multiple agents in parallel for as long as necessary wo/ blocking your local dev env. Companies like @SuperconductDev are working on the ergonomics of the product layer to make managing your scalable team of cloud agents graceful.

@langchain Sandboxes solve the security concern, but do they slow the iteration speed enough that your team stops using the agent? I've seen safe execution environments kill adoption because the friction outweighed the protection.

🚀 Building Enterprise Agents with Deep Agents Learn to build and deploy enterprise AI agents using Deep Agents and Runloop. Uses Runloop to run code safely in sandboxes 📺 Watch the full tutorial: youtube.com/watch?v=rj5OhG… Made by the LangChain Community

this was a ton of fun to work on, sandboxes are quickly becoming a default key part of agent infra. lots of use cases like safe code exec, reproducible work, and massively parallel jobs. hype to see what the community builds around this, reach out with any questions from the code, vid, or blog :)

@CobusGreylingZA @langchain @RunloopAI @daytonaio @modal These sandboxes are deployed as containers they are just hosted on a scalable cloud so you can have as many as you want, running as long as you would like.

Honest question...AI Agent sandboxes are exploding in popularity (Anthropic, LangChain, etc.), I understand the underlying premise for it...but online many are arguing 𝘄𝗵𝘆 𝗻𝗼𝘁 𝗷𝘂𝘀𝘁 𝘀𝗮𝗻𝗱𝗯𝗼𝘅 𝘄𝗶𝘁𝗵 𝗮 𝗱𝗼𝗰𝗸𝗲𝗿 𝗶𝗺𝗮𝗴𝗲? For instance, docker images that run the ML services (such as embedding models) and have staging environments with replicas of vector and relational databases, etc?

💻Sandboxes for DeepAgents We're excited to launch Sandboxes for DeepAgents, a new set of integrations that allow you to safely execute arbitrary DeepAgent code and bash commands in remote sandboxes. Supports @RunloopAI @daytonaio @modal Your DeepAgent runs locally (or wherever you want), but when it needs to execute code, create files, or run commands, those operations happen in the remote sandbox. Blog: blog.langchain.com/execute-code-w… Docs: github.com/langchain-ai/d… Video: youtu.be/CejntUP3muU

@codewithimanshu @langchain @RunloopAI @daytonaio @modal The sandboxes run on dedicated containers in microVMs like firecracker so at the compute layer isolation is bullet proof. Network isolation and tool use are configurable so you need to take care w/ how you set these services up.

@langchain @RunloopAI @daytonaio @modal That's a cool launch, LangChain! But I wonder, is security really foolproof with these remote sandboxes?

Very fun collaborating w/ the exceptional langchain team!