Amidot AI User Guide & Documentation

Comprehensive step-by-step documentation for enterprise admins and daily users.

I. System Hierarchy & Workspace Architecture

Amidot AI structures organizational data in a 5-tier logical hierarchy to preserve data boundaries and prevent contextual pollution:

Level	Entity	Scope & Isolation Guarantee
1. Region	Regional Cluster (EU, US, APAC)	Routes requests to designated data centers (e.g. `europe-north1` in Finland) ensuring GDPR and data residency compliance.
2. Tenant	Enterprise Organization	Mapped to dedicated Google Cloud Storage buckets. Domain whitelisting (`@company.com`) automatically routes all company employees to the tenant.
3. Project	You or Group	Enclosed project area. AI System Instructions, keywords, and document repositories inherit at the Group level without cross-project leakage.
4. Session	Conversation Thread	Persistent turn-based chat threads with automatic short- and medium-term memory management.
5. Artifacts	Files & Vector Chunks	Uploaded or AI-generated documents (JSON, Markdown, HTML, XLSX) and semantic vector embeddings stored securely in tenant GCS containers.

II. Onboarding & Single Sign-On (SSO)

A. Enterprise Administrator Setup (3 Minutes)

Subscribe to Amidot AI in Google Cloud Marketplace.
Upon redirect to cloud.amidot.com, click Complete Setup with Google.
Confirm your Legal Company Name, Contact Phone, and Corporate Domain (e.g. acme.com).
Click Activate Workspace & Enter. Behind the scenes, the system approves Google Procurement entitlements and provisions your private GCS containers.

B. Coworker / Employee Access (Zero Friction)

Once the Administrator has whitelisted the corporate domain, any employee can join instantly:

Navigate directly to https://cloud.amidot.com/ (no registration tokens or invite codes needed).
Click Continue with Google using your corporate email account.
The system matches your email domain and places you directly into the company workspace under the corporate billing entitlement.

III. AI Chat Interface & Reasoning Levels

The primary interaction dashboard is the AI tab. You can configure reasoning depth dynamically per prompt:

Prompt Submission: Type your query into the input box and press Shift + Enter or click Send.
Real-Time SSE Streaming: Responses stream instantaneously with live tool-execution progress indicators.
Rich Copy Actions: Hover over any AI response to copy clean formatted HTML (pasteable into Google Docs / Word with tables intact) or raw Markdown.

Configuring Reasoning Levels: In Settings, select your default model and thinking level (MINIMAL for fast factual queries, MEDIUM for balanced tasks, and HIGH for deep mathematical and coding deductions).

IV. Built-In Autonomous Agents

Amidot AI features built-in function-calling tools that execute autonomously during conversations:

Document Generator Agent (generate_and_store_file): Direct the AI to generate structured reports, data tables, or code. The AI writes the file directly into your secure GCS bucket, marked with a green source tag in your Files tab.
HTTPS Web Researcher Agent (handle_url_context): Include public HTTPS links in your prompt (e.g. Summarize the key findings from https://example.com/report). The agent fetches, cleans, and analyzes the live page content inline.

V. Document & Spreadsheet Transformations

To ensure optimal reasoning quality without token bloat, Amidot AI features automatic client-side transformations:

Word Documents (.docx): Converted automatically into clean, semantic HTML before transmission to the model.
Spreadsheets (.xlsx, .xls, .csv): Converted to structured JSON sheets for precise tabular analysis. Users can click the grid_on icon in the Files tab to convert JSON tables back to Excel format at any time.

VI. Proactive Budget Gatekeeper & Spend Caps

Corporate administrators and individual users have fine-grained control over usage costs:

First-Month Onboarding Cap: All new enterprise accounts start with an automatic $150 USD spend cap in their first 30 days to protect against accidental prompt loops or misconfigurations.
Personal Spend Limits: Users can select their preferred monthly limit ($50, $100, $150, $200, $300, $500, or Default) in the Settings tab.
Pre-Emptive Request Filtering: The backend calculates estimated token loads before calling AI models, returning HTTP 429 if the request would exceed monthly limits.

VII. Technical Support Channels

Technical Support Email: support@amidot.com
General Inquiries & Feedback: info.com
Google Cloud Console: Accessible at console.cloud.google.com