Everything You Want to Know
About AI at Reach

Submit a ticket through the IT ticketing system. Describe what you're hoping to use AI for — drafting emails, reviewing documents, building a tool — and IT will set you up with the right platform (Gemini or Claude), any needed training, and ongoing support. Gemini uses your existing Google Workspace account; Claude access is granted separately. Both use your Reach University single sign-on.

For day-to-day tasks using approved platforms (Gemini and Claude), you don't need formal approval — just access. Submit an IT ticket to get set up, and you can start using AI in your regular work.

For Domain 3 projects — building tools for external audiences like employers, candidates, or NCAD partners — you do need to partner with IT before starting. These projects have more complexity and require IT scoping from the beginning.

The easiest starting point depends on your role:

• Faculty: Open Gemini in Google Docs and ask it to help you draft a discussion question for an upcoming class. Review and improve what it gives you.
• Staff/Advisors: Paste the text of a recent candidate email into Claude and ask it to draft a professional, warm follow-up. See how it does.
• Operations: Take a meeting summary or data export and ask Claude to turn it into three clear bullet points for your next team meeting.

Start small. The goal is to get a feel for how AI works before relying on it for important tasks.

Yes. When you submit an IT ticket to get access, you'll be connected with any training materials and guidance IT has prepared. Training is being developed alongside the rollout — early adopters will help shape what training looks like for their departments.

If you have a specific training need — for yourself, your team, or a department-wide session — mention that in your IT ticket and the IT team will work with you on it.

ChatGPT is being phased out for work tasks at Reach. The three main reasons: it has no Google Workspace integration, no university-wide single sign-on, and it falls outside our enterprise agreement — which means we can't guarantee data privacy for anything you put into it.

Submit an IT ticket and note that you're currently using ChatGPT. IT will help you figure out the right migration — either Gemini or Claude, depending on what tasks you're using it for. The good news: both Gemini and Claude are generally more capable than ChatGPT for the tasks people use it for at Reach.

It depends on the deployment type.

Gemini is available at no additional cost through your existing Google Workspace account — everyone has access.

Claude for Teams / Enterprise is $25 per user per month, funded by department budgets. The funds flow to IT, which manages licensing centrally — so there's no personal out-of-pocket cost, but the department must budget for it.

SaaS AI features (Canvas, Element, etc.) are included in existing vendor agreements — no separate cost.

Marketplace apps and large-scale solutions (Claude Code / Azure AI Foundry) vary — IT will clarify costs when you submit a ticket for those.

ChatGPT is not covered under any Reach agreement, which is part of why it's being phased out.

Use Reach's IT ticketing system. When you submit, describe: (1) what you want to accomplish with AI, (2) whether you're looking for access/training, or scoping a specific project, and (3) any urgency or timeline. The IT team will respond and help you get set up. Contact Brad Luhta (Director of IT) with any questions about the process.

IT reviews your request and reaches out to: (1) grant access to the appropriate platform(s), (2) share any relevant training materials, (3) for Domain 3 projects, schedule a scoping conversation to understand what you want to build. For simple access requests, the turnaround is quick. For Domain 3 projects, there's a more involved scoping process before work begins.

IT has organized AI support into five types:

1. Google Gemini — Standard Workspace AI available to everyone (candidates, faculty, staff, contractors). Already in your Google account.
2. Claude for Teams / Enterprise — Licensed, IT-managed deployment at $25/user/month. Requires department budget and IT provisioning. Tied to Reach SSO.
3. SaaS AI Functionality — AI embedded in existing vendor tools (Canvas Ignite AI, Element BoltAI, future SIS AI). IT activates these within platforms you already use.
4. Marketplace Apps — Third-party AI add-ons for Google Workspace, Zoom, and similar platforms. IT approval required before use — submit a ticket.
5. Claude Code / Azure AI Foundry — Large-scale, infrastructure-level AI projects. Must run through IT from the start due to security and infrastructure requirements.

The common thread: anything new goes through IT via a ticket. This ensures security, compliance, and sensible coordination across the university.

Gemini is Reach's standard AI — built into Google Workspace and available to everyone (candidates, faculty, staff, contractors) at no additional cost. Best for day-to-day tasks in Gmail, Docs, Sheets, and Drive.

Claude for Teams / Enterprise is a licensed, controlled deployment at $25/user/month. It's managed through IT, tied to Reach SSO, integrates with Google Workspace and Slack via IT-approved connectors, and is better suited for deep document analysis, complex drafting, and building custom Domain 3 tools. Department budgets fund access; IT handles billing.

Think of Gemini as your everyday productivity tool and Claude for Teams as your licensed specialist for complex, document-heavy, or tool-building work.

Gemini is the best choice for email. It integrates directly into Gmail and can draft, improve, and summarize emails without you leaving your inbox. For routine candidate communications, follow-ups, and employer correspondence, Gemini is fast and effective.

Claude for Teams is significantly better for long documents. It can read full strategic plans, 40-page MOUs, accreditation standards, and grant applications — then give you a clear summary of what matters most. Gemini is capable, but Claude's document analysis is substantially more thorough and reliable for longer, complex texts.

Yes. A common workflow: use Gemini for day-to-day email drafting and Workspace tasks, then use Claude for Teams when you need deep document analysis or a more complex project. Gemini is available immediately through your Google account. For Claude for Teams, your department needs to budget for it and route funds through IT — submit a ticket to start that conversation.

Gemini's advanced Google Workspace features (the AI integration inside Gmail, Docs, Sheets, etc.) require users to be 18 years or older. This is a Google platform policy — it is a hard limit that Reach University cannot override.

This is particularly important for instructional contexts: if your course includes candidates who are under 18, they cannot use Gemini's Workspace features for class assignments.

Gemini's advanced features are currently optimized for English. Candidates or staff who primarily use languages other than English may find Gemini's performance less reliable. Claude for Teams generally handles multilingual content better. If language accessibility is important for your use case, mention it when you submit your IT ticket.

Claude for Teams is the preferred choice for Domain 3 because it has strong capabilities for building interactive applications, processing complex data, and generating structured outputs that can power custom tools. It can be integrated into employer portals, candidate self-service tools, and NCAD resources in ways Gemini is not designed for. The most sophisticated Domain 3 work may also involve Claude Code or Azure AI Foundry — all of which must be scoped with IT first. Contact Brad Luhta to get started.

Claude for Teams / Enterprise is a licensed, institutional deployment of Claude managed by IT. It costs $25 per user per month. Key features:

• Tied to Reach University SSO — no separate personal account needed
• Runs through Reach's security infrastructure
• Includes connector tools for Google Workspace, Slack, and other IT-managed apps
• IT manages all connector approvals
• IT is rolling out additional access controls (Reach accounts and managed devices only) — allow some runway for that rollout

If your team is interested, department budget flows to IT, which handles licensing and billing centrally. Submit a ticket to start the conversation with Brad Luhta.

Several of Reach's existing enterprise vendors are embedding AI into their platforms:

• Canvas Ignite AI — AI features within our LMS
• Element 451 BoltAI — AI within our enrollment and student engagement platform
• Future SIS features — As our student information system evolves, AI capabilities may follow

You don't install these separately — they appear as enhanced features within tools your team already uses. IT manages which features get activated and reviews each for data privacy compliance before enabling them. Contact IT if you're unsure whether a feature in an existing system is IT-approved.

For personal exploration outside of work, yes. For work-related tasks — all tools need IT approval first.

IT manages two categories that might feel like "finding your own tools":

• Marketplace apps (Google Workspace, Zoom add-ons): submit an IT ticket before using any add-on, even popular or free ones. IT will review for privacy and security.
• SaaS AI features from existing vendors: check with IT before enabling — they may already be in the pipeline, or IT may need to review before activation.

If you find a tool you think would genuinely help Reach, bring it to IT. They can review it for privacy, security, and enterprise suitability.

These are tools for building large-scale, infrastructure-level AI applications — the kind of custom systems that would power things like a university-wide employer portal or a sophisticated candidate advising tool. They're different from using Claude for Tasks or Gemini for drafting; they require proper software architecture, security planning, and infrastructure management.

The capabilities are real and available to Reach, but any project using these tools must go through IT from day one. This ensures the right security and governance are in place and prevents uncoordinated technology growth across the university. If you have a project in mind that might require this level of capability, start a conversation with Brad Luhta or Dr. St. Roseman.

Gemini is available through your existing Google Workspace account — it's the standard AI deployment available to all Reach users, including candidates, faculty, staff, and contractors. Once IT activates it for you, you'll see it integrated into Gmail (Help me write), Google Docs (Gemini in Docs), and you can access the Gemini interface directly. Submit an IT ticket and they'll activate your Gemini access and share quick-start guidance.

Claude for Teams / Enterprise requires IT provisioning — it's a licensed deployment at $25/user/month, not a self-service signup. To get started:

1. Confirm your department has budget for this (funding flows to IT, who manages licensing centrally)
2. Submit an IT ticket requesting Claude for Teams access
3. IT will provision your access via Reach SSO and share connector options for Google Workspace and Slack

Note that IT is also rolling out additional access controls (Reach accounts and Reach-managed devices only) — your IT contact will keep you informed on timing.

The following must never be entered into any AI tool — including our approved platforms:

• Student educational records (grades, transcripts, enrollment details, academic history)
• Social Security numbers or other government identifiers
• Financial account information (bank details, credit card numbers, financial aid records)
• Health or medical information (including behavioral health information)
• Passwords or login credentials
• Sensitive employer information protected by confidentiality agreements

These restrictions are grounded in FERPA, general data privacy best practices, and Reach's AI Guidance Policy. If you're unsure whether something is safe to share, ask IT before you do.

FERPA (Family Educational Rights and Privacy Act) is a federal law that protects the privacy of student educational records. At Reach, FERPA means that student academic information — grades, enrollment status, advising notes, academic history — cannot be shared without the student's consent, including with AI tools. Even our approved platforms don't have the right data handling agreements to receive protected FERPA information. The simplest rule: don't put anything in AI that identifies a specific student and relates to their education.

For IT-approved platforms (Gemini, Claude for Teams, and IT-activated SaaS AI features), general internal documents — meeting notes, draft SOPs, program descriptions, strategic memos — are generally fine. What you're protecting against is entering specific personal data about individuals, not all internal information.

That said, exercise judgment: don't enter documents that contain personal data, confidential legal information, or details specifically protected by confidentiality agreements. When in doubt, ask IT.

Yes, within limits. You can use AI to help you review or summarize employer MOUs (removing specific personal data first), draft employer communications, or analyze general partnership information. What you should not do: enter information that an employer has shared under an explicit confidentiality agreement, or personal information about specific employees at an employer partner. When working with employer documents, remove personal names and identifying details before using AI.

Contact IT and Dr. LeBoeuf immediately. Don't wait or hope it's fine — report it promptly. Depending on what was entered and into which platform, IT will assess the potential impact and take any needed steps. Prompt reporting allows Reach to respond appropriately and document the incident correctly. Early disclosure is always better than delayed disclosure.

Neither Gemini nor Claude should be used to process protected health information (PHI) as defined by HIPAA. This is especially important as the ACH (Behavioral Health) program launches in September 2026. For any behavioral health-related AI use in the ACH program, contact IT before beginning. IT will work with you to identify what's permissible and how to configure tools appropriately.

That's a question for faculty and academic leadership. Reach's AI Guidance Policy sets university-wide guardrails (privacy, disclosure), but individual faculty decide how AI is used in their courses — what's allowed, what requires disclosure, and what counts as academic integrity violations. If you're a candidate, check your course syllabus or ask your Professor of Practice.

Yes — in most cases. If AI substantially shaped formal work (an email you sent, a document you submitted, a report you delivered), you should disclose that. The standard isn't "did you use AI at all" — it's "did AI meaningfully shape the output." Using AI to check your spelling is different from having AI draft a 500-word policy memo that you sent with minor edits. When in doubt, disclose. Trust is foundational at Reach.

Faculty decide this for their individual courses — not the university centrally. Some courses may permit AI assistance with disclosure; others may restrict it; others may have nuanced policies depending on the assignment type. Faculty should communicate their expectations clearly at the start of a course, ideally in the syllabus. Candidates should ask if unclear.

A useful test: if you removed the AI contribution, would the work be meaningfully different? If yes, disclose. Examples that generally warrant disclosure:

• AI drafted the first version of a document or email and you made minor edits
• AI generated the structure or outline that organized your work
• AI wrote paragraphs or substantial sections that you included largely unchanged
• AI summarized a document and you presented that summary as your own analysis

Examples that generally don't require disclosure:

• Using AI as a grammar or spell checker
• Using AI to brainstorm ideas you then developed independently
• Using AI to search for information you then verified and wrote about in your own words

No — and this isn't just a policy position, it's a structural reality of how Reach works. The Reach Method is built on inquiry, dialogue, collaboration, and practice. Those require human judgment, workplace expertise, relationship-building, and mentorship. AI can help Professors of Practice do their jobs better — draft faster, give richer feedback, research more efficiently — but the job itself requires human presence and expertise. AI is a tool for faculty, not a replacement.

Add a clear AI use policy to your syllabus. Consider covering: (1) what AI use is permitted or restricted, (2) disclosure requirements for AI-assisted work, (3) what happens if a candidate violates your AI policy. Dr. LeBoeuf's office can provide guidance and model language. When in doubt, be explicit — ambiguity leads to problems. Contact Academic Affairs to discuss your specific course context.

This is an academic integrity matter — handle it the same way you would any other academic integrity concern, following Reach's existing policies and processes. Contact Dr. LeBoeuf's office for guidance specific to AI-related academic integrity situations. Documentation is important: note what was submitted, what AI use is suspected, and your reasoning.

AI is most useful for first drafts. You bring the expertise and judgment; AI handles the formatting, structure, and initial language. Specific things AI can do well for course building:

• Draft a syllabus structure given your learning objectives
• Generate 10 discussion questions on a topic — you pick the best 3
• Write a rubric aligned to specific competencies
• Create a case study scenario from a real-world workplace context you describe
• Suggest additional readings or resources on a curriculum topic
• Adapt existing materials for a new cohort or workplace sector

Use Gemini inside Google Docs for the most integrated experience.

Yes. AI can help you give feedback that is more detailed, consistent, and faster to write — especially valuable as your cohort grows. Common approaches:

• Paste a candidate's submission and ask AI to identify 3 strengths and 2 specific areas for improvement
• Ask AI to help you write feedback in a more constructive or encouraging tone
• Use AI to generate example feedback you then personalize

Important: AI should never evaluate a candidate on its own. You review and own all feedback — AI is a drafting assistant, not an evaluator. Never enter personally identifiable student information when using AI for feedback.

Once your Gemini access is activated by IT, you'll see a Gemini icon or "Help me write" prompt in your Google Docs toolbar. You can highlight text to get AI help improving it, ask Gemini to draft a new section, summarize a document, or suggest edits. Gemini is also accessible via the sidebar (Gemini in the right panel) for a more conversational experience while you work in Docs.

Yes, this is a practical and valuable use case. You can use Gemini (with its multimodal capabilities) or Claude to process a Zoom recording transcript and generate a summary, key takeaways, or a list of action items. This is useful for candidates who couldn't attend a session, or for faculty who want to create a follow-up resource after class. Be mindful not to include personally identifying student information in what you share with AI.

Yes — when used thoughtfully. The Reach Method involves inquiry, dialogue, collaboration, and practice. AI can support each:

• Inquiry: Generate provocative questions for candidates to explore; research background on industry topics
• Dialogue: Prep for class discussion by asking AI to anticipate common questions or counterarguments; create structured debate prompts
• Collaboration: Draft group project frameworks or evaluation criteria for collaborative work
• Practice: Generate practice scenarios tied to candidates' real workplace contexts

AI works best as a prep and support tool for faculty — the Reach Method itself still requires the human relationships and dialogue that define Reach's model.

The fastest path: (1) Submit an IT ticket for Gemini access. (2) While you wait, open claude.ai and try asking it to help you draft something for your next class — no account needed for a quick test. (3) Once Gemini is activated in your Workspace, open a Google Doc and ask "Help me write a discussion prompt for [your topic]." That first interaction — seeing what AI gives you and then improving it — is the fastest way to build intuition for how and when to use it.

Candidate communication is one of the highest-value starting points. With enrollment growing from 3,200 to 7,500, every team will need to maintain or improve communication quality at higher volume. AI helps by:

• Drafting personalized check-in emails given context you provide (stage of program, industry, recent interaction)
• Writing milestone reminder messages in a warm, clear tone
• Drafting responses to common questions (you review and personalize before sending)
• Creating communication templates for repeatable scenarios

Gemini in Gmail is the most integrated option for this. Always review and personalize before sending — AI gives you the first draft, not the final word.

Yes. You can export a summary or snapshot from Domo (without including personal data), paste it into Claude, and ask it to: write a 3-bullet summary for your team meeting, identify what the data suggests about trends, draft a narrative explanation of the numbers, or write an email to leadership summarizing the key findings. Claude is better than Gemini for this kind of structured data narration. Don't paste raw candidate-level data — aggregate summaries and de-identified outputs are fine.

Yes. Claude is excellent at this. You can paste or upload an employer MOU and ask Claude to: summarize the key obligations for both parties, flag anything unusual compared to standard language, identify what Reach needs to do by specific dates, or highlight provisions that might cause operational challenges. This can turn a 30-minute manual review into a 5-minute AI-assisted review that's often more thorough. Remove personally identifying information from the document before uploading if possible.

AI is excellent at first-draft documentation. The best workflow:

• Describe the process you want to document in a few sentences or bullet points
• Ask Claude or Gemini to turn it into a structured SOP with numbered steps
• Review, correct inaccuracies, and add your institutional knowledge
• Ask AI to format it cleanly for sharing

For FAQs: describe the common questions you get, ask AI to draft clear, plain-language answers, then review and improve. This is one of the fastest ways to capture institutional knowledge that currently only lives in people's heads.

The fastest workflow for most staff: email drafting with Gemini in Gmail. Once access is activated:

• Compose a new email and click "Help me write"
• Describe who you're writing to and what you need to say
• Gemini drafts the email — review, edit, and send

Most people save 10-15 minutes per email this way. Over a week of heavy communication, that adds up to hours. Once you're comfortable with this, branch out to document drafting, data summaries, and MOU review.

Yes. If you have survey results — aggregate responses, not individual identifiable answers — you can paste them into Claude and ask it to: identify the top 3 themes in the feedback, summarize what respondents said about a specific topic, or write a one-page summary of the results for leadership. For open-text responses, remove any personally identifying information before sharing with AI. Claude handles large amounts of qualitative text well.

Domain 3 is AI-powered tools built for external audiences — people outside Reach's core staff who interact with Reach in some way. This includes:

• Employer partners who want to track their candidates' progress
• NCAD partner universities who need access to Reach's frameworks and playbooks
• Prospective candidates exploring the program
• Policymakers, funders, or community stakeholders who consume impact data

Domain 3 is distinct from Domains 1 and 2 because the end user isn't a Reach employee — it's someone outside the organization. That's why these projects require IT partnership from the beginning.

Any department can identify the need for a Domain 3 tool, but building one requires IT partnership. This isn't a bottleneck — it's a governance safeguard. External-facing AI tools carry risks that internal tools don't: misinformation to candidates, privacy exposure to external parties, or brand/reputation risk if the tool malfunctions. IT partnership ensures these tools are built, tested, and governed correctly. Start by submitting a ticket describing what you want to build and why.

Submit an IT ticket and describe: (1) who the external audience is, (2) what the tool would do for them, (3) what problem it solves that can't be solved with existing resources. IT will schedule a scoping conversation with you, ask questions about requirements, and help design an approach. The scoping conversation is free and informative — even if a project doesn't move forward, the conversation helps IT understand your department's needs.

Based on Reach's FY26-28 strategic priorities, the highest-impact Domain 3 opportunities are:

• Employer progress portals — giving employer partners real-time visibility into their candidates' progress. Builds trust, reduces check-in calls, strengthens partnerships.
• NCAD replication resources — tools that let universities exploring apprenticeship degree programs access Reach's frameworks and playbooks without requiring Reach staff at every meeting.
• Candidate self-service Q&A — an AI tool that answers common questions about Reach programs 24/7, reducing load on advising staff as enrollment scales.
• Stakeholder impact reports — dynamic, data-driven reports generated for different audiences (legislators, funders, community partners).

It depends on the complexity. Simple tools — like a Q&A bot that answers questions about a specific program using a defined knowledge base — can be scoped and built in weeks. More complex tools — like an employer portal integrated with Reach's data systems — require significantly more time (months), especially if they involve integration with Domo, Sonis, or other Reach systems. The scoping conversation with IT will give you a more specific estimate based on what you want to build.

Three structural reasons — not about quality:

• No single sign-on (SSO): Individual ChatGPT accounts aren't tied to Reach's identity management. We can't verify who has accounts, ensure they're closed when someone leaves, or enforce security policies.
• No Google Workspace integration: ChatGPT doesn't connect to Reach's primary tools — Gmail, Drive, Docs. This creates workflow friction and makes it hard to build integrated processes.
• Outside enterprise agreement: ChatGPT usage by Reach staff isn't covered by any enterprise data privacy agreement. We cannot guarantee that information entered into ChatGPT is handled according to our privacy requirements.

Both Gemini and Claude meet all three requirements. The phase-out is about compliance and governance, not capability preference.

ChatGPT conversation history belongs to your personal account. You can export your ChatGPT history from the ChatGPT settings before phasing it out. However, anything you've already shared with ChatGPT stays in OpenAI's systems per their terms of service — there's nothing Reach can do about what's already been entered. Going forward, using approved platforms ensures your work is covered by appropriate enterprise agreements.

Both are AI assistants that generate text, but they have different strengths. Claude is notably better at:

• Reading and analyzing very long documents (Claude can handle entire book-length documents)
• Being careful and transparent — Claude tends to acknowledge uncertainty rather than state things confidently that it isn't sure about
• Following complex multi-step instructions precisely

ChatGPT has historically been stronger at image generation (via DALL-E integration). For the text-based tasks Reach uses AI for — drafting, analysis, document review — Claude is generally more capable and more appropriate for enterprise use.

• Email drafting: → Gemini in Gmail
• Document drafting (SOPs, guides, memos): → Gemini in Google Docs or Claude
• Reading/summarizing long documents: → Claude
• Research and background information: → Gemini (with web search) or Claude
• Data analysis and narratives: → Claude
• Quick questions and brainstorming: → Either — try both and see which you prefer
• Building a custom tool or application: → Claude (via IT partnership)

If you have a specific workflow that doesn't fit neatly into the above, submit an IT ticket describing it and IT will help you find the right approach.

Yes, for work-related tasks. The AI Guidance Policy (effective July 1, 2025) specifies approved platforms for work use. Using ChatGPT for Reach work tasks — especially tasks involving any Reach information — is not compliant with the policy. Personal use of ChatGPT on personal devices for non-work purposes is your own choice. But for anything related to your role at Reach, please migrate to Gemini or Claude. Submit a ticket and IT will help you make the switch.

The ACH (Apprenticeship College of Health) program launches its pilot cohort in September 2026. AI will be used for curriculum development, operational support, and potentially candidate-facing tools — but all of this needs to be scoped with IT before implementation. Contact the IT team well before the September 2026 launch to begin the conversation. The goal is to have AI tools ready and tested before cohort one begins.

Claude is designed to be careful and transparent — which is generally a strength. In behavioral health contexts, however, this carefulness can tip into over-caution: Claude may decline to engage with certain behavioral health topics, scenarios, or clinical language that faculty and staff in the ACH program need to work with routinely. This isn't a fundamental flaw — it's a configuration challenge. IT can work with the ACH team to configure Claude appropriately for behavioral health instructional use. The key is surfacing this early so there's time to test and adjust before the September 2026 pilot.

This needs careful handling. Behavioral health advising involves sensitive personal information about candidates, potential crisis situations, and professional obligations that AI is not equipped to handle on its own. AI should never be the sole responder in a behavioral health advising interaction. Potential appropriate uses include: helping advisors draft follow-up communications, summarizing resources, or preparing talking points for meetings. All such use must be scoped with IT and reviewed with Dr. LeBoeuf before implementation in the ACH program.

Start with IT (Brad Luhta) for platform access, configuration, and tool scoping. For academic and policy questions about how AI can be used in behavioral health instruction and advising, contact Dr. Lisa LeBoeuf (Vice Provost, Academic Affairs). For the ACH program launch more broadly, work through the program leadership and include IT in conversations about AI from the beginning.

Three leaders own the AI strategy:

• Dr. Paul St. Roseman (Chief Information and Research Officer) — Executive sponsor of AI integration across Reach. His role as Chief Research Officer has expanded to lead strategic AI adoption into business processes and internal tools university-wide.
• Brad Luhta (Director of IT) — All platform access, IT tickets, Domain 3 project scoping, training, and technical governance. He is the day-to-day capacity-building hub for AI at Reach.
• Dr. Lisa LeBoeuf (Vice Provost, Academic Affairs) — Academic policy, faculty guidance, disclosure standards, academic integrity, and the AI Guidance Policy's academic dimensions.

Brad Luhta and Dr. LeBoeuf co-authored the AI Guidance Policy (effective July 1, 2025). For most questions, start with IT. For questions about academic or faculty policy, contact Dr. LeBoeuf. For institution-wide AI strategy and business process integration, contact Dr. St. Roseman.

This depends on the nature and severity of the violation. Minor violations (e.g., using ChatGPT for a non-sensitive task after the phase-out) may result in a conversation and guidance. More serious violations — particularly those involving privacy breaches, FERPA violations, or intentional misrepresentation using AI — would be handled through Reach's standard HR and academic integrity processes. The AI Guidance Policy is a formal university policy, and significant violations carry the same weight as violations of other institutional policies. When in doubt, ask before you act.

Platform decisions (which AI tools are approved for use at Reach) are made by IT leadership with input from Academic Affairs and reviewed against three criteria: (1) enterprise security and data privacy standards, (2) compatibility with Reach's existing systems (especially Google Workspace and SSO), and (3) coverage under Reach's enterprise agreements. ChatGPT failed all three. Gemini and Claude passed all three. If you believe a new platform should be evaluated, bring it to Brad Luhta's attention — IT will assess it against these criteria.

Yes, almost certainly. AI is evolving rapidly — platform capabilities, data handling practices, and regulatory environments are all changing. The AI Guidance Policy will be reviewed on a regular cycle (at minimum annually) to ensure it reflects the current landscape. Platform-specific restrictions (like Gemini's age limit or Claude's behavioral health configuration) should be re-evaluated regularly as platforms update their capabilities. When the policy changes, IT and Academic Affairs will communicate updates to all relevant stakeholders.

NCAD (National Center for Apprenticeship Degrees) is Reach's national replication network, launched in 2025. Its role is to help other universities launch apprenticeship degree programs using Reach's model — the Launch Accelerator and Backbone (LAB). The moonshot goal: 3 million apprenticeship degree starts by 2035.

AI supports NCAD by making Reach's playbooks, frameworks, and resources accessible at scale — without requiring Reach staff to be physically present or involved in every conversation. AI-powered tools can help partner universities access and apply Reach's methodology without proportional demands on Reach's team.

The most valuable AI tools for NCAD would:

• Provide interactive access to Reach's program launch playbook — answering specific questions based on where a partner university is in the process
• Help partner universities understand the ABCs framework (Affordability, Based in Workplace, Credit for Work) and how to adapt it for their context
• Generate customized templates for employer partnership agreements, MOUs, and program documentation
• Offer decision-support for common program design questions
• Connect partner universities to relevant Reach resources without requiring a Reach staff member to facilitate every interaction

These are Domain 3 projects — they'd be built by IT in partnership with the NCAD team.

The 3 million apprenticeship degree starts goal requires Reach's model to scale far beyond what Reach's own staff can deliver. NCAD is the vehicle for that scaling — helping other universities replicate the model. AI is the multiplier that makes NCAD's work possible at the required scale. Without AI tools that can carry Reach's knowledge and frameworks, NCAD partner universities would need proportionally more support from Reach staff — making the 3 million goal functionally impossible. AI-powered NCAD resources make the moonshot realistic.

Start by submitting an IT ticket and mentioning NCAD specifically — Brad Luhta's team will route it appropriately. NCAD-related AI projects are Domain 3 projects, which means IT partnership is required. The NCAD team and IT will need to work together to define what tools are most valuable for partner universities and how to build and maintain them. Early conversations are welcome — the sooner NCAD AI tools are scoped, the more they can contribute to Phase 2 and 3 of the rollout.