Scout

Any MCP client can connect to Scout agents, tables, and documents through a single hosted endpoint. No local setup required. Claude, Cursor, Codex, and more. Connect once. Use Scout anywhere. scoutos.com/changelog/2026…

Claude Sonnet 5 landed yesterday. It's already live in @scoutshq today! Available to every user, on every agent, no waiting for a tier upgrade.

Your API key doesn't have permission levels. It just has access. That's the default. With @scoutshq it doesn't have to be. If you want to transmit your private key, Scout can do that. If you want to segment traffic via keys, Scout can do that. If you want to constrain a credential to a role, Scout will enforce it. If you'd rather sign a JWT and transmit that instead, Scout accepts it. If you want to disable a key temporarily, Scout will enforce it. If you want to rotate a key, Scout can do that...on a timeline you own. Start building your agentic credential architecture @ scoutos.com

Your team already lives in Slack. Now your AI agents can too. Most "AI assistants" make you go to them. Typically a separate tab, a separate login, a separate workflow nobody asked for. We flipped that. With @scoutshq agents in Slack, you deploy an agent straight into the channels where work actually happens. No new habits. No context-switching. Just an agent in the room, ready when you are. Here's what makes it click: ⚡️ Deploy in seconds — connect any agent to any channel from Studio 🎯 Channel-aware by design — set triggers and conditions so agents respond to the right moments, not every message 🧠 Custom instructions per channel — your #support agent and your #revenue agent behave like two different teammates, because they should 🔀 Two ways in — wire it up from the channel list or straight from the agent itself The result: AI that shows up where your team already is, acts on the rules you set, and gets out of the way until it's useful. The best automation isn't the one you remember to open. It's the one already sitting in your channel. Connect your first Scout agent to Slack today @ scoutos.com

You have 30 agents running across your org. Do you know which models they're using right now? Most enterprise AI deployments can't answer that. Not because nobody cares, because nobody built the controls to track it...that's what Model Management in @scoutshq is for. When AI moves from pilot to infrastructure, "whoever built the agent picked the model" stops being acceptable. Security needs an approved list. Finance needs predictable costs. And when a better model ships next quarter, and it will...someone needs to update 30 agents without rebuilding them one by one. Here's what that looks like in practice: → Approved models only — define the models your org has vetted; agents can't run anything outside that list → Cost tiers by design — set a default for everyday work, reserve premium models for the agents that justify the spend → Bulk reassignment — update your entire agent fleet when the model landscape shifts, not one agent at a time → Vendor flexibility — use whatever model you want for the task at hand; stop building your stack around someone else's default The result: a model layer that's governed, auditable, and actually yours. The best AI infrastructure isn't the most powerful. It's the one your org can actually stand behind. Start managing your model stack @ scoutos.com

When a task goes sideways with an AI assistant, the instinct is to blame the model. "It doesn't understand my codebase." "It keeps suggesting the wrong approach." "It's not smart enough for this problem." But most of the time, the issue isn't intelligence. It's context.

We ❤️ Dagster + Colton! One of our earliest customers and incredible partners

Wait a second. Zoom in on that screenshot. Enhance! Is that...? Yeah that's a reference to the Scout codebase on the GPT5 livestream! 😎 Full context: One of the Scout engineers reported a bug with PDF uploads in OpenAI's Python SDK and they tried to get GPT-5 to solve it on their livestream. They didn't even look at the code it gave them (not something we'd recommend)... but hey at least we got our bug fixed.

The exact moment here: youtu.be/0Uu_VJeVVfo?t=…

New: @statsig MCP now available in Scout! Create AI agents that can manage feature flags, run A/B tests, and analyze experiments through natural conversation. From safe feature rollouts to statistical analysis - all powered by AI. #FeatureManagement #AIAgents #ProductDevelopment scoutos.com/blog/new-stats…

🚀 NEW: GitHub MCP integration is live in Scout! Chat with your repos, manage PRs, review code, and coordinate releases through natural conversation. 50+ GitHub tools. Zero context switching. Pure developer magic. @github + Scout = the future of development workflows ✨ Try it: scoutos.com/blog/new-githu…

🚀 We just launched Linear MCP integration for Scout! Now you can create AI agents that manage your @linear issues through natural conversation. Create tickets, update status, assign tasks - all by just chatting with your Scout agent. No more context switching! ✨ scoutos.com/blog/introduci…

Tired of answering the same questions over and over? These LLM prompts transform repetitive support requests into helpful FAQs scoutos.com/blog/top-5-llm…

Better prompts = better answers. Check out these 5 prompt techniques that help dial in your RAG workflows: scoutos.com/blog/top-5-llm…

Explore Anthropic's latest models, Claude Sonnet 4 and Opus 4 on Scout! Built for extended reasoning and powerful coding performance.

A little taste of what's cooking! 🧑‍🍳 Lots more on the way! Let us know what you'd like to see next.