Help & Guides
Your guide to understanding and using the Alkimi platform.
Getting Started
Welcome to Alkimi AI, the infrastructure for grounded, academic AI. Our platform empowers you to transform your proprietary data (documents, curriculum, and research) into secure, intelligent agents that guide learning and automate workflows. Unlike generic AI tools, Alkimi ensures your assistants are strictly grounded in the knowledge you provide, offering control, accuracy, and pedagogical value.
This documentation is designed to help you get up and running quickly. Whether you are a professor building a Socratic tutor for your students, an administrator setting up a compliance bot, or a developer integrating our API, these guides will walk you through every step of the process, from ingesting your first document to deploying your fully customized agent.
How We Work
The Alkimi platform operates in three high-level stages:
Ingest & Ground
Connect proprietary data sources
Build & Customize
Configure persona & controls
Deploy & Interact
Launch to employees or customers
- Ingest & Ground: Clients securely connect their proprietary data sources (such as documents, curriculum materials, or support wikis) to the platform. Alkimi processes and indexes this knowledge to create a secure, private foundation for the AI.
- Build & Customize: Organizations use the platform to configure their intelligent assistants. This includes defining parameters, setting access controls (i.e., private vs. public), and customizing the assistant's persona.
- Deploy & Interact: The resulting AI assistant is deployed to its intended audience. This can be internally for secure access by employees or students, or externally by embedding it directly onto a public website to serve as a customer support or help agent.
Key Concepts
- Organizations: The top-level container for your team. An organization holds all your members, billing, and security settings. You can belong to multiple organizations.
- Workspaces: Subdivisions within an organization that let you scope agents, knowledge, and member access. Every organization starts with a Default workspace, and you can create more to separate projects or teams.
- Agents: Customizable AI entities that can perform tasks, answer questions, and automate workflows.
- Knowledge: Collections of documents and data that you can provide to your agents, giving them a unique context and knowledge base.
- Blocks: When you upload a document, it is read page by page and broken down into logical blocks (paragraphs, headings, list items, tables and figures), each anchored to its position on the page. The retrieval system searches these blocks to find the most relevant information for a user's query, and citations point back at the exact block.
- Chats: The primary interface for interacting with your agents.
Navigation
The platform uses a left-side navigation rail (icon bar) to switch between different tools and settings.
- Navigation Rail: The icon bar on the left side of the screen provides quick access to all major areas of the platform. From top to bottom: Workspaces (home), Chat, Agents, Knowledge, and Organization Settings.
- Workspace Switcher: Each area operates within a workspace context. You can switch between workspaces using the workspace dropdown in the sidebar. Workspaces allow you to organize agents, knowledge, and members into separate scopes within your organization.
- Quick Search: Press Ctrl+K (or Cmd+K on Mac) to open the command palette, which lets you quickly search for and navigate to any page, agent, collection, or action across the platform.
- Account Menu: Click your profile picture at the bottom of the navigation rail to access personal settings, including your Profile, Security, Preferences, Help & Guides, and Logout. See the Account guide for more details.
Your First Agent: A 5-Minute Tutorial
Let's create a simple agent that can answer questions about a specific topic.
- Create a Knowledge Collection: Click Knowledge in the left icon rail. Click the plus (+) button next to the collection dropdown in the sidebar and give it a name, like "Project Research."
Create Collection
Collection nameWhat is this collection about? (Optional)0 / 512 - Add a Document: Select your new collection in the sidebar. Then, click the plus (+) button next to "Content" or hover over a folder, click the three dots (...), and select "Add File(s)." In the dialog, click
"Click to select files" to upload a PDF or text file.
Add File(s)
Select files to upload.
Click to select filesor drag & drop files here
Max 100MB & 1000 pages per file.
- Create the Agent: Click Agents in the left icon rail. Click the plus (+) button next to the agent dropdown in the sidebar. Give it a name, like "Research Assistant."
Create Agent
Choose how to build your agent.
Agent Name
e.g. Customer Support...Leave blank to auto-generate
How to Build
AI Agent Builder
Describe what you need and AI will build it.
Browse Catalog
Pick from pre-configured templates.
Import from File
Upload an agent configuration (.json).
Create Blank Agent
Start with defaults and configure later.
- Connect the Knowledge: In the agent configuration screen, go to the "Knowledge" tab. Select your "Project Research" collection to connect it to the agent.
- Start Chatting: That's it! Click Chat in the left icon rail. Select your new "Research Assistant" from the sidebar, and start asking it questions about the document you uploaded.
Tip: Creating an Agent from a File
You can also create a new agent from a configuration file. In the "Create Agent" dialog, click the "Import from File" button in the footer. This will open a file picker for you to select a JSON configuration file. This is a great
way to quickly set up a new agent from a backup or a shared template. You can override the name of the agent from the file by entering a new name in the dialog before clicking import.