AI email drafts and chat with business context
Try responses from manually submitted messages and stored context. Your team checks them before use. Mailbox automation and live knowledge connectors are not available yet.
AI Email Assistant
The available function accepts a manually submitted customer message and context, then drafts a reply using a stored business profile, knowledge text and examples. It also returns a hint that a person may be needed; it does not route the case. It does not read a mailbox or send email. Gmail, Outlook, IMAP, conversation history and an auto-send threshold are not implemented.
What it does
- Draft from a manually submitted customer message and context
- Stored business profile, knowledge text and reply examples
- Output with provider identification
- Human-review hint, without automatic routing
- Czech or English according to settings
- Available draft API; mailbox connections and sending are planned
Where it helps most
Repeated support questions
Submit a message and approved context; the assistant prepares a draft for your team to check.
Sales inquiries
Prepare a reply from a submitted inquiry. The assistant does not read your inbox or qualify leads by itself.
Onboarding reply drafts
Use routine questions as draft inputs; your team remains responsible for accuracy and sending.
Configured language and tone
Replies use configured Czech or English and tone. Automatic language detection is not included.
How it works
- 1
Submit a message
Submit the message text and optional context through the API or console preview. Mailbox access is not included.
- 2
Store business context
Set the business profile, knowledge text, tone and reply examples. External sources are not fetched automatically.
- 3
Send with your own tools
After approval, send the reply yourself. Mailbox connections and automatic sending are on the roadmap.
AI Knowledge Agent
The available chat processes one question with a stored business profile, instructions and knowledge text. It does not retrieve live data from Notion, Slack or Drive and does not return source citations. You update the context yourself and should check the answers. Conversation history, indexing, streaming and ready-made widget or Slack-bot deployment are not implemented.
What it does
- Single question with stored business context
- Manually maintained knowledge text and instructions
- Output with provider identification, without citations
- Console preview and chat API
- Czech or English according to settings
- Live connectors, indexing and streaming are planned
Where it helps most
Questions about business context
Get a draft answer from stored text. A ready-made website widget and citations are not included.
Internal FAQ
Check answers to common questions in the preview or API; a live Slack connection is not available.
Manually maintained knowledge
Update context in settings. Changes in Notion or Drive are not retrieved automatically.
Onboarding questions
Submit one question and check the draft against approved context. Conversation history is not available.
How it works
- 1
Prepare knowledge text
Add approved business context to settings. Live retrieval from external sources is not available.
- 2
Configure responses
Set the profile, instructions, tone and default reply language in the console.
- 3
Check and update
Check answers and update context manually. A widget, Slack bot and live connectors are planned.
AI PR Review
The GitHub App reviews a requested diff within size limits and posts a summary and any comments. The diff and PR context are processed by an external AI provider through OpenRouter. Approval and merge stay with a person; rule-based auto-approval is only planned.
What it does
- GitHub App triggered by a review request
- Diff and PR context processed through OpenRouter
- Diff size limit; large files may be omitted
- Summary and any inline comments on the PR
- Approval and merge performed by a person
- Rules and auto-approval are planned
Where it helps most
Dependency and formatting changes
AI can add review comments on changes. It does not replace tests or human approval.
Docs and copy
Comments may highlight content issues. PRs are not approved or merged automatically.
A second view for your team
A requested PR gets an AI summary and any inline comments for a reviewer to check.
Asynchronous review
Request the bot as a reviewer and check the results when ready. A human decision is still required.
How it works
- 1
Request a review
Install and link the app, then request praktickai[bot] on a selected PR.
- 2
Read AI comments
An external AI provider processes the submitted diff and context. The output is an additional review, not a safety guarantee.
- 3
Decide as a team
Check comments, tests and permissions. The bot does not approve or merge.
Calculate your savings
Model potential savings from your inputs; these are not measured results or available automation.
Hours saved / month
106 h
Cost saved / month
3696€
This is a rough estimate based on the values you enter, not a guaranteed result.
Why our AI, not a generic chatbot
Context you maintain
Responses use stored context. You check accuracy; source citations are not available.
Clear integration scope
Email drafts and single chat questions are available. Mailbox and live MCP connectors are planned.
Configured reply language
The runtime uses Czech or English according to organization settings. The website is also available in Russian.
Human review
Your team checks answers, sends email and decides on PRs. Automatic routing, sending and approval are not included.
Ready to start?
Start a free course or explore training options for your team.
Beginner courses are free. Team training includes workshops, custom programs, and consulting.