SquadOS SquadOS
PTEN
Get started
Structured memory for your agent

Upload PDFs, links, and text. The agent searches and answers with precision.

Upload PDFs, links, and text. Everything gets indexed automatically for the agent to search and answer with precision.

Create my knowledge base

An AI agent only answers well about what it knows. SquadOS's Knowledge Base lets you upload the documents, links, and text that define your business — and the agent starts searching them via RAG (Retrieval-Augmented Generation) before answering, instead of relying solely on what the model learned in training. The result is accurate, up-to-date answers grounded in your actual information.

File uploads: PDF, DOCX, XLSX, CSV, TXT, JSON, Markdown (up to 10 MB)

Upload documents in the most common formats. Each file is processed and indexed automatically — no special formatting required.

Text blocks and website crawling

Paste text directly or point a URL for the crawler to follow. The crawler reads the sitemap, follows internal links in scope, and can re-sync manually, daily, weekly, or monthly.

Automatic indexing and embeddings

After upload or crawl, the content is automatically embedded and indexed. The agent searches by semantic similarity — not just keyword matching.

Multiple bases, connected per agent

Create separate bases for different areas (e.g., support, sales, HR) and connect each base only to the agents that should access it. Search, filter, and export items as CSV.

How it works

Create a base, add content, and connect it to the agent — it starts searching before answering.

01

Create a base and add content

Upload files (PDF, DOCX, XLSX, CSV, TXT, JSON, Markdown), paste text blocks, or point a URL for the crawler to follow.

02

The content is indexed automatically

After adding each item, SquadOS processes and indexes the content via embeddings. No action is required on your part.

03

Connect the base to the agent and enable lookup

In the agent settings, enable the "Consult Knowledge Base" tool and select which base it should access. From then on, the agent searches the base before formulating each response.

Frequently asked questions

  • What file formats are supported?

    PDF, DOCX, XLSX, CSV, TXT, JSON, and Markdown, with a maximum size of 10 MB per file.

  • Can I keep my website content always up to date in the base?

    Yes. The crawler can automatically re-sync content from a URL on a daily, weekly, or monthly schedule — or you can trigger a re-sync manually at any time.

  • Can I have different bases for different agents?

    Yes. You can create as many bases as you like and connect each one only to the relevant agents. A support agent can access the FAQ base, while a sales agent accesses the product catalog.

  • Does the knowledge base work with AutoLearn?

    Yes, and it's a requirement. AutoLearn only works when the "Consult Knowledge Base" tool is active on the agent with at least one base selected. Gaps detected by AutoLearn are added directly to the base.

Related features

AutoLearn

Every question the agent doesn't answer well becomes an improvement opportunity. AutoLearn spots them on its own, groups them by similarity, and gives you a button to add them to the knowledge base.

AgentMaker

Say what the agent needs to do and AgentMaker sets everything up: prompt, model, tools, knowledge base, and channels. No AI expertise required.

Give your agent real memory

Upload your documents and the agent starts answering with your information, not guesses.

Start for free

PDF, DOCX, links, and text. Automatic indexing.