What if you could ask a question about a client’s infrastructure and get a useful answer without hunting through tabs, folders, or half-remembered notes? That’s exactly the kind of experience Weavestream’s built-in AI Chat is designed to deliver — and it’s a lot more thoughtful than a simple chatbot bolted on for show.
Let’s take a tour.
It’s Right Where You Need It
The AI Chat panel lives inside every company page in Weavestream. There’s no separate app to switch to, no copy-pasting snippets into an external tool. Just click the chat icon in the company sidebar, and the panel slides in from the right.
That placement is deliberate. When you’re working inside a tenant — checking assets, reviewing documentation, managing passwords — the AI is already there in context. You’re not breaking your flow to get help; the help comes to you.
You can even resize the panel by dragging its edge if you need more room. Small detail, but the kind of thing you appreciate when you’re in the middle of something.
Your Model, Your Rules
This is where Weavestream does something refreshingly different from most “AI-powered” tools: you choose the language model.
Under Admin → Settings → AI, you plug in any OpenAI-compatible endpoint. That could be OpenAI’s own API (with GPT-4o or GPT-4o mini), or it could be a self-hosted model running locally via Ollama or LM Studio. As long as it speaks the OpenAI chat completions API, it works.
Why does this matter? A few reasons:
- Cost control — you’re not paying a per-seat AI subscription on top of your Weavestream licence.
- Privacy — if you’re running a self-hosted model, your client data never leaves your own infrastructure.
- Flexibility — as the AI landscape evolves and better models emerge, you swap the endpoint, not the platform.
It’s a bring-your-own-brain approach, and it suits managed service providers and security-conscious teams very well.
Smart Context, Not Spray-and-Pray
Here’s where the real magic happens. The AI only knows what you tell it — but Weavestream makes it remarkably easy to tell it the right things.
Auto-attach means that if you open the chat while you’re already looking at an asset or an article, that record is automatically pulled in as context. You can see exactly what’s attached in a strip above the message input, and you can remove it if you don’t want it included.
@-mention lets you explicitly search for any article or asset in the current company and attach it. Type @, pick from the list, and it’s added. You can stack up multiple items for richer context — handy when a client issue touches several pieces of infrastructure at once.
This explicit context model is important. The AI doesn’t silently crawl your entire database. It only sees what you’ve chosen to share with it. That keeps responses focused and keeps your data handling predictable.
Draft, Edit, and Save — Without Leaving the Page
The AI Chat isn’t just a Q&A tool. It’s genuinely useful for documentation work.
When an article is attached, you can ask the AI to rewrite a section, expand on something, fix awkward phrasing, or restructure content. The proposed changes appear as diff cards right inside the chat — you can review exactly what would change before accepting. Hit accept, and the edits land in the article immediately. No more copy-pasting between a chat window and a text editor.
There’s also a Save as article action on any assistant response. If the AI drafts something useful — a troubleshooting guide, a network topology summary, a set of onboarding steps — you can turn that response into a proper article in a few clicks, choosing the title, folder, and format before it’s created.
For teams who document client environments as part of their workflow, this is a genuine time-saver.
Privacy First, Always
Because Weavestream is self-hosted and uses your own LLM endpoint, the privacy story is clean: data goes from your server to your configured endpoint, and nowhere else. Weavestream doesn’t proxy your requests through any Weavestream-operated infrastructure. What gets sent is the text of attached articles and the field values of attached assets — nothing more.
That means your data-handling obligations are between you and your LLM provider. If you’re running Ollama locally, that means your data never leaves your premises at all.
Give It a Try
If you haven’t explored AI Chat yet, the best starting point is simple: open a company page, click the chat icon, and attach an article you’ve been meaning to improve. Ask the AI to tighten it up, add a missing section, or rewrite the intro. See the diff. Accept what looks good.
It takes about two minutes to get a feel for it — and once you do, it’s one of those features that quietly becomes part of how you work every day.