(RFI 69-MFL0002) - UNCG Enterprise Chatbot.pdf

RFI – #

Version 11.2024

REQUEST FOR INFORMATION (RFI) # 69-MFL0002

University of North Carolina Greensboro Issue Date: October 16th, 2026 Due Date: November 6th, 2026 no later than 4:00 pm EST

Refer ALL Inquiries regarding this RFI to:

Michael Logan Director of Procurement Services mflogan@uncg.edu Phone – 336.334.4104

Description: Enterprise AI Chatbot Websites Solution Commodity No:

RFI SUBMITTAL: Vendor’s response to this RFI must be submitted through the NC eVP submission portal. For additional information or training on how to register and submit a response in NC eVP, responsive vendors should visit the following site: https://eprocurement.nc.gov/training/vendor-training.

Questions or issues related to using the tool itself can be directed to the North Carolina eProcurement Help Desk at 888-211-7440, Option 2. Help Desk representatives are available Monday through Friday from 7:30 AM EST to

5:00 PM EST.

NOTICE TO VENDOR

Request for Information (RFI) will be received at this office until 4:00 pm EST on the day of opening.

QUESTIONS

Written questions shall be submitted to Michael Logan via email at mflogan@uncg.edu and are due no later than October 23rd, 2026 5 pm EST. Please insert “RFI # 69-MFL0002- Questions” in the subject matter of your e-mail.

A summary of all questions and answers will be consolidated into a Q&A Addendum and posted to NC eVP, on or about October 30th, 2026.

EXECUTION

VENDOR NAME: E-MAIL:

STREET ADDRESS: P.O. BOX: ZIP:

CITY & STATE: TELEPHONE NUMBER: TOLL FREE TEL. NO:

TYPE OR PRINT NAME & TITLE OF PERSON SIGNING: FAX NUMBER:

AUTHORIZED SIGNATURE:

DATE:

mailto:mflogan@uncg.edu https://eprocurement.nc.gov/training/vendor-training mailto:mflogan@uncg.edu

Version 11.2024

1.0 OBJECTIVE & PURPOSE

The University of North Carolina Greensboro's Department of Information Technology Services is actively seeking information regarding enterprise AI chatbots for University websites. The intent of this Request for Information (RFI) is to gather information on viable solutions, prior to potentially issuing a future solicitation which may result in a contract.

This RFI is issued for information and planning purposes only and does not constitute a formal solicitation.

Responses to the RFI will not be returned nor will the University confirm receipt of the RFI response. Whatever information is provided in response to this RFI will be used to assess tradeoffs and alternatives available for determining how to proceed in the acquisition process. Responses to this RFI are not offers and cannot be accepted by the Department to form a binding contract.

All interested vendors are requested to provide a written response to the questions below. A response to this RFI is necessary to assist the University in determining the potential levels of interest, adequate competition, market maturity, and service capabilities within the business community.

The Department requests detailed vendor specific and industry standard information on how your firm would address the items in the following sections of this RFI.

RESPONSIVE VENDORS ARE CAUTIONED THAT THIS IS A REQUEST FOR INORMATION ONLY, AND NOT A REQUEST TO CONTRACT. THE UNIVERSITY MAKES NO COMMITMENT TO ACQUIRE ANY GOODS OR SERVICES

UNDER THIS RFI.

THE UNIVERSITY WILL NOT PAY FOR INFORMATION RECEIVED IN RESPONSE TO THE RFI. THE VENDOR SOLEY

BEARS ANY COSTS INCURRED IN PREPARATION AND SUBMISSION OF A RESPONSE.

2.0 INTRODUCTION

The University issued this RFI to conduct market research in determining variables to consider for an AI chatbot enterprise solution which may be used for University websites, as defined herein.

Based on the responses to this RFI, the University may further solicit for the requirements that most meet its needs regarding cost, performance, delivery method, growth potential, scalability and least disruptive to operations.

3.0 BACKGROUND & CURRENT STATUS

The University of North Carolina Greensboro (UNCG) has several Artificial Intelligence (AI)-enabled chatbot solutions currently in place. This includes solutions embedded with existing software applications as well as several University websites, including

• Spartan Central (https://spartancentral.uncg.edu/)

• Housing & Residence Life (https://hrl.uncg.edu/)

• Procurement Services (https://procurement.uncg.edu/)

The website-based chatbots are sourced from multiple vendors and have different capabilities and administrative controls.

This RFI is focused on website chatbots and not AI-capabilities embedded within specific software applications.

See Appendix A for a full list of all known requirements and questions pertaining to this RFI.

https://spartancentral.uncg.edu/ https://hrl.uncg.edu/ https://procurement.uncg.edu/

Version 11.2024

4.0 RFI PROCEDURES

The University recognizes that considerable effort may be required in preparing a response to this RFI.

However, respondents are advised that this is a request for information only, and not a request to contract.

Respondents shall bear all costs for preparing their RFI response. Respondents to this RFI are asked to be thorough and concise.

Entities who submit documentation in response to this RFI may be invited to present their ideas. An in-person presentation by an entity that submits documentation or participates in discussion or demonstrations in response to this RFI will not serve as a basis for precluding that entity from responding to any future solicitation regarding handling and disposition services.

All responses, inquiries, or correspondence relating to or in reference to this RFI, and all documentation submitted by the various respondents shall become the property of the University when received. The responses to this RFI, any follow-up questions and presentations to the University may be used to develop any future solicitation.

5.0 SCHEDULE

Respondents will have 3 weeks to prepare their submissions to this RFI. Responses must be received by the date and time specified in this RFI. Respondents may also be invited to present and discuss their solutions. This could be virtually or in person. Respondents will be notified of the specific date and time at least two (2) weeks in advance of their presentation.

6.0 CLARIFICATION QUESTIONS

Clarification questions regarding this RFI will be accepted until specified time on the cover sheet of this RFI. All questions must be submitted in writing to the email address specified on the cover sheet of this RFI. An addendum containing clarification questions, and their answers will be issued as an addendum to this RFI, though eVP.

7.0 VENDOR RESPONSE

The Department is not interested in brochures or “boilerplate” responses. Instead, responses should clearly define how the Vendor’s proposed solution(s) would meet the Department’s business requirements. Any issues or exceptions to the Department’s requirements should also be identified and explained in detail. The Vendor shall provide the following in their response:

1. Transmittal Letter

The transmittal letter should be in the form of a standard business letter and should be signed by a respondent-authorized individual. It should note the following:

a) A statement that deviations are included, if applicable.

b) A statement that proprietary information is included, if applicable.

2. General Information

Respondent information:

a) Name of Company

b) Contact Person

c) Address, Telephone Number, Email Address

d) Brief history of company

Version 11.2024

3. Content

Interested firms shall submit a response to this RFI together with:

• Marketing materials including information on your firm

• Detailed description of the product(s), services and capabilities available to the Department

• Representative cost estimates

• Recommended implementation strategy

• Timeline estimates

• References from current clients who received the same services as described in the RFI response

The response should define all services that will be required by the proposed solution. The response should also include:

• The vendor’s understanding of the project and services by addressing the Department’s business requirements;

• An estimated total cost of ownership for the solution including continued compliance with emerging industry standards.

4. Cost and Schedule Estimates

These cost estimates are not formal quotes and have no bearing on any future solicitation for solutions.

This information is for research and planning purposes only.

a) Cost Estimates for initial solution implementation to include hardware, software and services.

b) Cost Estimate for total cost of ownership. Be sure to include any specific LLM token charges if this is charged separately. UNCG prefers a flat-rate subscription.

c) Define any limitations to cost estimates

d) Describe DEPARTMENT resources required to implement and support solution

e) Describe all related charges or fees, for a solution proposed.

5. Multiple Responses

Multiple responses will be accepted from a single vendor provided that each response is comprehensive, meets all of the state’s requirements, and is truly unique. Please submit responses separately and clearly indicate responses as “Response #1, Response #2, etc.

Version 11.2024

8.0 CONFIDENTIAL INFORMATION

In accordance with 09 NCAC 06B.0103 the State may maintain confidentiality of certain types of information described in N.C. Gen. Stat. 132-1 et. seq. Such information may include trades secrets defined by N.C. Gen. Stat.

132-1.2. Respondents may designate appropriate portions of its response confidential, consistent with and to the extent permitted under the Rules and Statutes set forth above, by marking the pages containing confidential information with boldface type at the top and bottom of each such page stating “CONFIDENTIAL.” By so marking any page, the Respondent warrants that is has formed a good faith opinion, having received such necessary or proper review by counsel and other knowledgeable advisers that the portions marked confidential meet the requirements of the Rules and Statutes set forth above. The State may serve as custodian of Respondent’s confidential information and not as an arbiter of claims against Respondent’s assertion of confidentiality. If an action is brought pursuant to N.C. Gen. Stat. 132-9 to compel the State to disclose information marked confidential, the Respondent agrees that it will intervene in the action through its counsel and participate in defending the State, including any public official(s) or public employee(s). The Respondent agrees that it shall hold the State and any official(s) and individual(s) harmless from any and all damages, costs, and attorney’s fees awarded against the State or official or individual in the action. The State agrees to promptly notify the Respondent in writing of any action seeking to compel disclosure of Respondent’s confidential information. The State shall have the right, at its option and expense, to participate in the defense of the action through its counsel. The State shall have no liability to Respondent with respect to the disclosure of Respondent’s confidential information ordered by a court of competent jurisdiction pursuant to N.C. Gen. Stat. 132-9 or other applicable law.

All information received in response to the RFI that is marked Confidential will be handled accordingly. The Department shall not be liable for or suffer any consequential damages for any proprietary information submitted and not properly identified. Proprietary information will be safeguarded in accordance with the applicable state regulations.

Appendix A: Additional RFI Information

Provide the following information about your solution and describe how your chatbot solution will support the following known requirements:

General Requirements

1. The chatbot should support a federated model where individual University departments can utilize the chatbot platform, each with its own bot. Chatbot knowledge should be able to be isolated or shared with other bots, as needed. This federated model should allow for central oversight by ITS but delegate individual chatbot administration to departments. Describe how this works.

2. Provide branding customization (UNCG logo, color scheme, greeting message, avatar/name, etc.) consistent with the University’s public-facing communication standards.

3. Describe how your solution interacts with a WordPress CMS system. Does the chatbot need to be installed on each individual webpage, or can it be installed on an entire departmental subdomain to ease the installation burden?

4. Does your platform support multi-channel deployment beyond a standard website (e.g. SMS, WhatsApp, MS Teams, native mobile apps) using the same central knowledge base?

5. Describe the customer’s interaction with the chatbot, including how the bot would appear on a University website, the greeting that a customer would receive, and how the bot converses with the customer.

6. The chatbot should be able to make use of content (e.g. webpages and related media / documents) on specified websites as well as additional information and documents that can be added to the chatbot’s knowledge base. Specify how the knowledge base is maintained, how often websites are crawled, etc.

7. Does your platform provide industry specific content prepackaged in the chatbot’s knowledge base? For instance, a Financial Aid chatbot may have information from crawled US Department of Education websites.

8. Can your solution review information available in the chatbot’s knowledge base against higher education-specific knowledge metrics to proactively suggest when potential gaps in knowledge exist. Are there specific functional areas where this gap analysis exists? If so, specify which functional areas are included in the gap analysis capability.

9. In lieu of having a direct knowledge base support for a customer engagement, describe the generative AI capabilities for your solution.

a. How are these conversations flagged as potential knowledge gaps or specific conversations to review?

b. Describe how the generative AI answers can be confirmed to provide consistent answers across time.

c. Describe how the platform prevents hallucinations.

10. Describe how your solution supports multi-turn conversation capability (i.e., chatbot can ask clarifying follow-up questions when user intent is ambiguous). The chatbot should be able to maintain context throughout a customer engagement.

11. Proposed solutions should be compatibility with standard web browsers and mobile-responsive display formats.

12. If the bot becomes stumped, it should be able to offer the customer the ability to:

a. Hand off to a live human representative. This capability should only be offered when agents are logged into the system.

b. Submit a support ticket that can be followed up upon. This should generate a ServiceNow ticket.

Describe how both of these support options work in your solution.

13. Describe any multi-lingual (particularly Spanish) support that the chatbot can provide. Note that the administrative function will only provide knowledge base information in English.

14. The solution should provide source attribution and hyperlinking; inline citations and direct links to official

UNCG web pages or knowledge base articles are preferred. Describe how your solution supports this requirement so that customers can verify information / answers.

15. Describe the chatbot’s ability to make use of SSO, including

a. Specify the SSO protocols available for integration with your solution.

b. How does the customer experience change after logging in via SSO? What additional resources are available in the chatbot’s knowledge base and where can those documents reside, such as within WordPress or SharePoint or elsewhere?

c. Can the chatbot use SSO to gain access to enterprise systems (e.g. Ellucian Banner, Slate, Salesforce, ServiceNow, Canvas) to answer customer-specific inquiries?

d. Does the chatbot preserve conversation history and context when the customer logs in mid-chat?

Administrative Requirements

1. Provide the ability for departmental users to customize, edit, and expand chatbot responses, decision trees, and knowledge base articles without requiring vendor intervention for routine content updates.

2. Departmental users should be able to review all customer-chatbot interactions to understand how the bot engaged with customers and successfully answered (or not) their questions. Engagements that did not have successful outcomes should allow administrative users to make edits for how similar questions are answered in the future. Describe how this process works.

3. Describe the training provided to departmental users in order for them to become proficient in using the tool.

What administrative guides and training materials are available to administrative users?

4. Provide information about the administrative burden for both a) implementing the tool and b) ongoing maintenance. How does the administrative time commitment change post-implementation?

5. Describe system and usage metrics available to departmental administrators that support efficient monitoring of the chatbot. How is sentiment analysis (identifying user frustration/drop-off points) and intent clustering (grouping unhandled queries into actionable administrative recommendations) handled by the platform?

6. The chatbot should be able to have controls that specify the verbosity of answers. Clear, crisp answers tend to be preferred over long responses. Additionally, discuss how the chatbot’s tone and voice can be prescribed in its configuration.

7. Identify the underlying large language model (LLM) engine(s) that power your solution. Describe whether your platform supports integration with third-party or institution-specified LLM providers (e.g. OpenAI, Google

Version 11.2024

Gemini, etc.), and describe any limitations, licensing implications, or performance trade-offs associated with substituting your default LLM engine if UNCG has a preferred LLM provider.

8. Web content and university policies change frequently. Does the platform automatically flag outdated knowledge base content or alerts administrators when indexed web pages return 404 errors or altered content?

Security, Reliability & Accessibility

1. Provide a copy of your most recent Voluntary Product Accessibility Template (VPAT) which specifies your products support of Section 508 standards. The chatbot interface should comply with the Americans with Disabilities Act (ADA) and Web Content Accessibility Guidelines (WCAG 2.1 Level AA or most current standard).

2. Provide documentation of data encryption standards (in transit and at rest), data retention policies, and data ownership terms confirming that all data collected through the chatbot remains the property of the Department. Provide confirmation that customer interactions and proprietary knowledge bases will not be used to train public LLM foundation models.

3. Discuss administrative permission depth. Can an admin in one departmental unit be restricted so they can only edit their content and view their interaction logs without accessing other departmental content and transcripts?

4. Specify explicit compliance frameworks that your tool supports, particularly for student privacy (FERPA) and health-related interactions (HIPAA).

5. Higher-ed chatbots regularly encounter sensitive or high-risk inputs (e.g., mental health distress, Title IX concerns, active campus safety threats). How does the system detect crisis keywords and instantly trigger hardcoded emergency responses (e.g. UNCG campus police, direct human intervention, etc.)?

6. How can conversation and knowledge base data be extracted from the system, either for further analysis or for future vendor platform changes?

7. Specify your operational SLAs in terms of system availability.

8.0 CONFIDENTIAL INFORMATION