Summary
DeepThought-Internal Technician Bot is a USNH-internal AI assistant that helps Client Services and other support staff answer customer questions faster and more consistently. It is trained on:
- Published USNH TeamDynamix Knowledge Base (KB) articles
- Prior 12 months of historical TeamDynamix ticket data (resolutions, notes, and related correspondence from the ticket body and feed entries)
This training data is refreshed weekly, so newly published articles and recently resolved tickets may not appear in DeepThought's answers until the next refresh cycle.
DeepThought-Internal Technician Bot is not a replacement for your own judgment. Always verify critical steps against the linked KB article before relaying instructions to a customer.
Who Can Access DeepThought-Internal Technician Bot?
Access to DeepThought-Internal Technician Bot is restricted to members of the following Active Directory group:
- App-USNH-ETSCS-CSDeepThought-usr
Key Capabilities
- Natural language questions. Ask DeepThought the way you'd ask a coworker. You do not need to phrase questions as keyword searches.
- Cited answers. Most responses include a Sources Used section that links directly to the KB articles and ticket numbers DeepThought pulled from.
- Conversational follow-ups. If the first answer is incomplete, too generic, or missing a platform-specific detail, keep asking in the same conversation thread. DeepThought maintains context within the conversation.
How-To
Task: Use DeepThought-Internal Technician Bot to assist with troubleshooting
Instructions
Step 1 – Open DeepThought-Internal Technician Bot at https://deepthought.usnh.edu/?id=helpdesk-int
You will land on the Helpdesk Internal Chatbot (DeepThought-HD-Internal) home page with a How can I help? prompt and a chat input box labeled Chat with DeepThought.
- Note: If you see a conversation in progress when you come to this page, refresh the browser page to start a new, blank conversation.

Ask DeepThought-Internal Technician Bot a Question
Step 2 – Before you begin, plan your question.
- The more specific you are, the better the answer.
- Know the customer's device type, operating system, role (student, faculty, staff), and campus if relevant to the issue. You will often need to include these details to get the right answer.
Step 3 – In the Chat with DeepThought input box, type your question in plain language. For example:
How do I connect to Eduroam?

Step 4 – Press Enter or click the send arrow on the right side of the input box.
Step 5 – DeepThought will return an answer, usually formatted with a heading, step-by-step instructions, and a Sources Used section at the bottom listing the KB article(s) and ticket(s) it pulled from.

Step 6 – Review the entire answer. If the answer fully resolves the customer's question, you can:
- Walk the customer through the steps directly, or
- Send the customer the linked KB article from the Sources Used section so they have a reference.
Important! Remember to follow the Best Practices below!
Refine an Answer Using Follow-Up Questions
DeepThought's first answer is often general. If the customer has a specific device, a specific error, or a more unusual situation, keep the conversation going in the same thread to narrow the answer down.
Step 7 – After reading the initial answer, type a follow-up question into the same chat thread. For example:
Can you tell me the steps for my Chromebook?
Step 8 – DeepThought will use the context of the earlier exchange and return a more specific answer.

Step 9 – Continue asking follow-up questions as needed. Useful follow-up patterns include:
- "What if they get a certificate error?"
- "Is there a related ticket where this was resolved?"
- "What are the troubleshooting steps for [specific symptom]?"
- "Does this apply to [student / faculty / staff] accounts?"
Step 10 – Check the Helpful Links and Sources Used sections at the bottom of each response. These often include related KB articles that DeepThought considered.


Step 11 – When you have a useful answer, you can:
- Send the customer the answer from DeepThought directly and/or
- Send the customer the linked KB article(s) so they have a reference.
Important! Remember to follow the Best Practices below!
Step 12 - To start a new conversation with DeepThought on a new topic, refresh the browser page to return to the original Helpdesk Internal Chatbot (DeepThought-Internal Technician Bot) landing page.
Best Practices
- Always verify links before sending to customers. If DeepThought's phrasing differs from the KB article, trust the KB article.
- Use ticket references for context, not as customer-facing material. Related ticket numbers are helpful, but do not send ticket links to customers.
- Ask follow-up questions rather than re-typing the whole question. DeepThought works best as a conversation. If the first answer is off, refine in the same thread.
- Provide device and account context up front when the issue is platform-specific. For example, "A faculty member on a personal Chromebook cannot connect to Eduroam" will get a better first answer than "Eduroam isn't working."
- Remember the weekly refresh. If a KB article was just published or a ticket was just resolved, DeepThought may not know about it yet.
- AI responses may contain incomplete or incorrect information. If something looks wrong, check the cited KB article. If the KB article itself is wrong or out of date, flag it for update through the normal KB review process.
Back to Step 6 above
Limitations
- DeepThought-Internal Technician Bot only knows what is in the USNH TeamDynamix KB and the prior 12 months of TeamDynamix ticket history.
- It does not have access to information outside of the USNH data it was trained on.
- Its training data is refreshed weekly, so very recent changes may not be reflected yet.
- Answers are generated by an AI model and can occasionally be incomplete, outdated, or incorrect. Always verify against the cited sources.
Further Readings
DeepThought also has a personal version, available to anyone in USNH at https://deepthought.usnh.edu/
AI at USNH - Category with Public KB Articles including USNH's DeepThought Artificial Intelligence tool - secure for use at USNH
Need additional help?
If you have questions or need additional help with these topics, please reach out to your Team Lead or supervisor for assistance.