> ## Documentation Index
> Fetch the complete documentation index at: https://docs.testfactors.com/llms.txt
> Use this file to discover all available pages before exploring further.
# FactorBot: in-app AI assistant
> Chat with FactorBot, the TestFactors AI assistant that knows your project, workbooks, and test history. Ask questions and get answers in plain English.
**FactorBot is the fastest way to get an answer about your project.** Faster than searching, faster than asking a colleague, and it cites its sources so you can verify.
## What FactorBot does
FactorBot is an AI assistant built into TestFactors that knows everything about **your current project**:
* Every uploaded workbook (every sheet, every cell)
* Every scenario, test case, and script
* Every defect logged in any cycle
* Every entry in your project's knowledge wiki
* Every membership and permission in your project
You ask in plain English. FactorBot answers with citations — every claim it makes links back to the source it came from.
Examples of what you can ask:
> *"Which scenarios cover hire flows for international employees?"*
>
> *"Show me all defects logged on the Hire wizard this cycle."*
>
> *"What does Iteration 2 mean on the Position Management sheet?"*
>
> *"Who's the assigned tester for the Compensation module?"*
>
> *"What's the difference between Test Cases TC-042 and TC-043?"*
>
> *"Summarize the open defects by severity."*
>
> *"What workbook changes between v2 and v3 of the FO workbook?"*
***
## Who can use FactorBot
| Role | Can chat | Can see all project data | Can run actions |
| ----------------------- | -------- | ------------------------------------ | -------------------------------------------------------- |
| **All project members** | Yes | Scoped to what you can see in the UI | Limited (read-only for now; some safe actions available) |
FactorBot respects your permissions. A Tester asking *"Show me all members of this organization"* will only see members of the projects they belong to.
***
## Where to find FactorBot
The FactorBot chat icon appears in the bottom-right corner of every page when you're inside a project. Click to open a side panel.
***
## Step-by-step: ask your first question
Click the FactorBot icon (bottom-right). The chat panel slides in.
Just write what you want to know in plain English. Don't worry about phrasing — FactorBot understands typos, partial questions, and conversational tone.
Example: *"how many test cases for hire"*
FactorBot streams the answer in real time. Every claim has a citation badge — click it to jump to the source.
Example response:
> You have **18 test cases** related to hire flows across 3 categories:
>
> * **Hire** (12 cases) — `Category: Hire` \[↗]
> * **Rehire** (4 cases) — `Category: Rehire` \[↗]
> * **Concurrent Hire** (2 cases) — `Category: Concurrency` \[↗]
>
> Of those, **15 are Active** and **3 are still Proposed**. Want me to open the Proposed ones for review?
Reply to refine. FactorBot keeps the conversation context:
> *"Yes, show me the proposed ones"*
FactorBot opens the list in the main pane or directly cites the 3 proposals with links.
Click any citation badge `[↗]` to open the source (workbook cell, test case, defect, wiki page) in a new tab.
**That's it.** Your first question. FactorBot remembers the conversation, so follow-ups feel natural.
***
## What FactorBot is great at
*"Where's the defect about the picklist on the Job Code field?"* — beats manual searching every time.
*"Summarize this cycle's progress."* / *"What are the top 3 risks right now?"*
*"What does the AI mean by 'high\_null\_rate' on this column?"* — FactorBot reads the docs so you don't have to.
*"How is test case TC-042 different from TC-051?"* — diff view in plain English.
*"What's the expected behavior for a Position with no manager assigned?"* — FactorBot finds the relevant workbook cell.
*"Who hasn't started their assignments yet?"* — saves you a trip to the dashboard.
## What FactorBot is less great at
*"Will we hit the deadline?"* — FactorBot can summarize progress and velocity, but human judgment makes the call.
*"Should we skip regression testing this cycle?"* — that's a call you make.
FactorBot doesn't browse the internet or know things outside your TestFactors project. For SF help, use SAP's documentation.
By design FactorBot won't delete defects, archive test cases, or remove members on its own — those need explicit human action in the UI.
***
## Tips
*"Show me Critical defects in EC logged this week"* beats *"What's broken?"*. Specifics get crisper answers.
No need for keywords or jargon. Conversational works fine: *"i think there's a defect about the wizard. can you find it?"*
*"Walk me through this project — what's the goal of this cycle?"* — FactorBot gives a great orientation.
*"Show me defect counts by severity"* gets you a table. *"Is the project healthy?"* gets you opinion. Pick the right one.
Don't trust the summary — verify the source. FactorBot is right 95% of the time and wrong 5% — knowing which is which matters.
Long conversations get muddled. Use the **+ New chat** button to reset context when switching subjects.
***
## Conversation history
FactorBot keeps your chat history per project. The history panel (left side of the chat) shows recent conversations grouped by date.
* **Search history** with the input at the top
* **Pin** important conversations so they stay at the top
* **Delete** conversations you don't need
* **Share** a conversation with a teammate (sends a link they can view)
***
## Troubleshooting
Two possible causes:
1. **You don't have permission** to see the data you're asking about (e.g. defects in a project you're not a member of). FactorBot respects your permissions — ask your Admin if you should have access.
2. **The data doesn't exist yet** — e.g. asking about scenarios when none have been generated. Use the relevant module to create the data first.
Click the **thumbs-down** on the response with a one-line note describing what was wrong. This trains the model. The right answer is often one follow-up question away: *"That citation looks wrong — can you double-check against the v3 workbook?"*
Most responses come within 3–10 seconds. If it's taking 30+:
1. Cancel and retry (the **Stop** button in the chat input)
2. Refresh the page
3. If it's still slow, your AI service may be experiencing latency — try again in a few minutes
Ask a follow-up: *"Can you give me more detail on the second point?"* or *"Show me 10 examples instead of 3."* FactorBot calibrates to what you ask for.
Click the **share** icon in the chat header. Generates a link that — when opened by a teammate with project access — shows them the full conversation. They can continue the conversation from there.
Workbooks need to finish processing before FactorBot can read them. Check the Workbooks tab — once it says "Ready", FactorBot will see it within \~30 seconds.
***
## Related guides
Add domain knowledge so FactorBot's answers get smarter over time.
FactorBot's accuracy depends on clean, processed workbooks.
Ask FactorBot to find, summarize, or compare defects.
Use FactorBot to find specific test cases when there are too many to scroll through.