A Plain-Language Guide to Our AI Features
Provided by Nomisma LLC, a Wyoming limited liability company.
We believe you deserve to understand how AI features in Actaa work. This guide explains what our AI does, what data it uses, and what safeguards we have in place — all in plain language.
---
What Actaa AI Is
Actaa AI is a workspace assistant that helps you find information and get work done faster. Think of it as a smart search and writing helper that knows about your company's projects, tasks, and documents.
What It Can Do
- Answer questions about your work ("What's the status of Project X?")
- Find information across your workspace ("Show me tasks assigned to Sarah")
- Generate content like summaries, reports, and drafts
- Prepare for meetings with relevant context
- Help with writing emails, documents, and updates
What It Cannot Do
- Make decisions for you
- Take actions automatically (like sending emails or completing tasks)
- Access information outside your company's workspace
- Remember conversations across different sessions (unless your admin enables history)
---
Two Modes of Operation
Open Mode (General AI)
When you ask general questions — like "What's the capital of France?" or "How do I write a for loop in Python?" — the AI answers using its general knowledge. It does NOT look at your company data.
Example: "Explain what agile methodology is"
Result: General explanation from AI's training data
Org Mode (Workspace AI)
When you ask about your work, the AI searches your company's data to find relevant information and generates a response based on what it finds.
Example: "What tasks are due this week?"
Result: AI searches your tasks and lists what's due
---
What Data the AI Can Access
Your company administrator controls what data AI can search. Here's what can potentially be enabled:
| Data Type | Default | Notes |
|---|---|---|
| Tasks & Projects | ✅ On | Respects visibility settings |
| Documents | ⚪ Admin choice | Must be explicitly enabled |
| Wiki Pages | ⚪ Admin choice | Must be explicitly enabled |
| Meeting Notes | ⚪ Admin choice | Private notes excluded |
| Time-Off Records | ⚪ Off | Requires explicit enablement |
Data the AI NEVER Accesses
These categories are never made available to AI, regardless of settings:
- Direct messages between individuals
- Private channels you're not a member of
- Salary and compensation information
- Passwords or API keys
- Social Security numbers or government IDs
- Health information
- Payment card numbers
---
How We Protect Your Data
Three Layers of Protection
- Before indexing: Sensitive patterns (like SSNs, credit cards) are automatically detected and removed before AI can ever see them.
- When you ask: Your questions are scanned and sanitized before being sent to the AI.
- In responses: AI answers are checked for accidentally leaked sensitive information before you see them.
Encryption
- All data is encrypted when stored
- All data is encrypted when transmitted
- API keys (for those who bring their own) use military-grade encryption
Isolation
- Your company's data is completely separate from other companies
- AI can only access data from your company
- No cross-company data sharing ever occurs
---
AI Providers We Use
All AI features are powered exclusively through Amazon Bedrock (Amazon Web Services), in the us-east-1 region. We do not send your data to any AI provider's own API — no OpenAI, no direct Anthropic API, no Google AI, or similar:
| Provider | What They Provide | Their Commitment |
|---|---|---|
| Amazon Bedrock (AWS) | Managed AI model inference, using the Anthropic Claude family of models (and other foundation models available in our AWS account) | Inputs are used only to generate a response, are not retained beyond the request, and are not used to train foundation models |
Important: Your data is processed within our AWS environment to generate responses and is then discarded. It is not used to train any AI model.
---
Conversation History
What Gets Saved
If your admin enables conversation history:
- Your questions and AI responses are stored
- History is tied to your user account
- History is retained for the period your admin sets (30-365 days)
What Doesn't Get Saved
- Conversations are not used to train any AI models
- Conversations are not shared with other users
- Conversations are not analyzed for advertising
Your Control
- You can view your conversation history
- You can delete individual conversations
- You can export your history
- You can delete all your history
---
Understanding AI Limitations
AI Can Make Mistakes
AI is not perfect. It can:
- Misunderstand your question
- Miss information that exists in your workspace
- Generate incorrect details (called "hallucinations")
- Provide outdated information if data hasn't been re-indexed
Always verify important information before acting on AI responses.
AI Doesn't "Think"
AI generates responses based on patterns in language. It doesn't truly understand context the way humans do. It works best for:
- Finding information quickly
- Generating first drafts
- Summarizing long content
- Answering straightforward questions
It works less well for:
- Nuanced judgment calls
- Highly specialized expertise
- Situations requiring common sense
- Emotional or sensitive topics
---
HR Data Protections
If your company enables HR-related data sources (like time-off or performance data), additional protections apply:
What AI Will Do
- Tell you factual information ("Sarah has 10 PTO days remaining")
- Summarize data ("The team has 3 open positions")
What AI Will NOT Do
- Recommend who to hire, fire, or promote
- Rank or score employees
- Predict employee performance
- Make any employment-related suggestions
These restrictions are enforced automatically and cannot be bypassed, ensuring compliance with employment laws.
---
Your Rights
Access
- View what data AI has access to (Settings → AI → Data Sources)
- See your conversation history
- Export your AI data
Control
- Disable AI for your account (ask your admin)
- Delete your conversation history
- Opt out of specific data sources
Transparency
- Ask questions about how AI works (we're here to help!)
- Request our data processing documentation
- Review our security practices
---
How Billing Works
Token System
AI usage is measured in "tokens" — roughly 4 characters of text. Both your questions and AI responses consume tokens.
What Your Admin Controls
- Monthly budget limits
- User-level limits (if any)
- Alert thresholds
- Overage settings
What Happens at Limits
When budget is exhausted, AI politely lets you know and suggests waiting for the next cycle or contacting your admin.
---
Frequently Asked Questions
"Can AI see my private messages?"
No. Direct messages and private channels are never accessible to AI, regardless of admin settings.
"Will my questions be used to train AI?"
No. Neither Actaa nor our AI providers use your data for training purposes.
"Can other people see my AI conversations?"
Your conversations are private to you. Admins can see usage statistics but not conversation content (unless legally required).
"What if AI gives me wrong information?"
AI can make mistakes. Always verify important information from the original source. AI responses include links to source documents when available.
"Can I turn off AI completely?"
Yes. Ask your admin to disable AI for your account, or avoid using AI features. The rest of Actaa works normally without AI.
"Is my company's data safe?"
Yes. Your data is encrypted, isolated from other companies, and processed only to answer your questions. We never sell or share your data.
---
Questions or Concerns?
We're committed to transparency. If you have questions about how AI works:
- In-App Help: Click the help icon and search for "AI"
- Email: info@actaa.ai
- Privacy Concerns: info@actaa.ai
---