MSBAi Standard Tools Reference

Audience: Faculty, course designers, and TAs. This page defines the standard toolset for all MSBAi courses. Students receive a streamlined setup guide during onboarding.

Course-specific tool requests (faculty poll). Faculty may request specific tools for their own courses — these are treated as “case pack” course materials on top of the shared base stack, per the “Student AI tool cost model and GitHub Copilot update” decision (2026-06-04). Course-by-course requests are gathered in the MSBAi — Instructor Software & Tool Needs poll (a faculty survey; document owner: Maria Rodas), which informed that decision. The responses are parsed weekly into reference/faculty-tool-poll.md (repo only — not published here) — see it for current per-course demand and a tool-frequency rollup. This page defines the shared program baseline, not a ceiling.

Tool Stack Overview

1. VS Code — Primary IDE

What: Free, open-source code editor from Microsoft. Extensible with thousands of plugins.

Why VS Code (not JupyterLab): VS Code is where professional developers and data scientists work. It supports notebooks, Python scripts, terminal, Git, debugging, and AI tools in one environment. Ron Guymon: “This is more complex than JupyterLab, but ultimately more useful. I think the students will have more appetite for it.”

Install: code.visualstudio.com

Required extensions for MSBAi:

2. AI Access Model — Ranked Options

MSBAi AI access model (confirmed 2026-06-04): AI access follows a student-responsibility model. A base AI capability is a program requirement, treated like a textbook cost; course-specific premium tools are treated like “case pack” costs. The program does not publish specific dollar amounts to students, and total spend is expected to stay within the $500 program-wide out-of-pocket cap (see design/assessment_strategy.md and the program cost policy).

Ranked access options (Vishal, 2026-08-14). Point students at these in order — the two free tiers cover normal coursework; the paid tiers are the escalation path for heavy use:

1. GitHub Copilot — free for verified students; the best free option. Free via GitHub Education (monthly eligibility re-check by GitHub). Students sign up at github.com/education using their @illinois.edu email, then install Copilot and Copilot Chat in VS Code for completions, inline chat, and multiple models (Claude, GPT, Gemini). Faculty can assume Copilot access for students who complete verification — build it into course setup instructions.
2. Lumen — free campus-hosted models (NCSA). The University of Illinois NCSA’s AI gateway at lumen.ncsa.illinois.edu: NetID/CILogon login, per-student API keys, an OpenAI-compatible endpoint (https://lumen.ncsa.illinois.edu/v1) that plugs into Continue and similar tools, and campus-hosted open models — currently including GLM-5.2 (the top open-weights model), qwen3-coder-next, muse-glimmer-30b, and deepseek-v4-flash. Cost to student: $0, metered by a coin budget (10 coins to start, +0.05/hr refill, confirmed 2026-08-14). Some models require a one-time consent acknowledgment. Canonical student setup guide lives in BADM 554 (Fall 2026 online, “Using campus AI models in VS Code with Lumen”; posted to that course’s Canvas).
3. Continue + OpenRouter (recommended paid base layer, ~$10–20/month). One key, many models (Claude, GPT, Gemini) at roughly 3–4× the capability per dollar of a single Pro plan. No campus caps. This remains the recommended student-paid base AI capability.
4. Single-vendor plan (Claude Pro or ChatGPT Plus, ~$20/mo). Simplest billing, fewer models — see §4. Faculty cannot assume every student has a specific paid plan.

⚠ Free-tier constraint: both free options carry usage and model restrictions — Copilot has monthly premium-model limits, and Lumen’s coin budget refills slowly (~1.2 coins/day). Comfortable for coursework; a real constraint for heavy use. Course designs that assume sustained heavy AI usage should assume students escalate to options 3–4.

• Most MSBAi instructors list Claude as their preferred AI tool. Because students may use a different base system, course-specific premium tools are treated as individual course materials, not a program-provided subscription.

What changed: Student free-plan signups were paused in May 2026, and the program briefly treated Copilot as a paid optional add-on (2026-06-04). GitHub restored free Copilot for verified students, so as of 2026-07-26 Copilot is once again free and a standard (non-optional) part of the tool stack. The student-paid base AI layer, the OpenRouter recommendation, and the $500 program-wide cap are unchanged by that restoration. Decision record: “GitHub Copilot free for verified students — restored” (2026-07-26; internal registry — ask K-ai).

How to set up Copilot (free for verified students):

1. Create a GitHub account at github.com
2. Verify your student status at github.com/education using your @illinois.edu email (GitHub re-checks eligibility monthly)
3. Install the GitHub Copilot extension in VS Code

What students get with Copilot:

Low floor / high ceiling progression:

Week 1-2:   Inline completions only (ghost text while typing)
Week 3-4:   Copilot Chat for debugging and explanations
Month 2+:   Agent Mode for multi-step tasks
Capstone:   Custom instructions, agent skills, MCP integrations (see courses/practicum.md)

Faculty note: Copilot is a tool, not a shortcut. All courses include AI attribution requirements — students must document what AI tools they used, what prompts they gave, and how they validated outputs. See design/assessment_strategy.md.

3. Google Colab — Notebook Environment

3a. Colab in Browser (Low Floor)

What: Free, browser-based Jupyter notebook environment with cloud compute (including free-tier GPU).

Why it’s the floor: Zero install. Students open a browser, sign in with Google, and start coding. No Python setup, no dependency management, no terminal.

URL: colab.research.google.com

Free-tier includes:

• Python environment with common libraries pre-installed (pandas, sklearn, matplotlib, etc.)
• NVIDIA T4 GPU and TPU v5e accelerators (limited session time)
• Google Drive integration for file storage
• Shareable notebooks via link

When to use browser Colab:

• Pre-program orientation and bridge modules
• First 1-2 weeks of any course (before VS Code setup is complete)
• Quick prototyping and experimentation
• Students who need a fallback when VS Code has issues

⚠ Colab’s built-in Gemini AI is not suitable for class use. Faculty testing (2026-08-20) found that 3 simple single-sentence coding prompts consumed 10% of the free daily Gemini credit limit. Colab’s notebook runtime is still recommended; for AI coding assistance, use GitHub Copilot or Lumen (see §2).

3b. Colab Extension for VS Code (High Ceiling)

What: Official Google extension that connects .ipynb notebooks in VS Code to Colab cloud runtimes. Launched November 2025.

Why it matters: Students get VS Code’s full IDE features (Copilot, debugging, Git, extensions) while running code on Colab’s cloud GPUs. Best of both worlds.

Install: VS Code Marketplace

How it works:

1. Open any .ipynb file in VS Code
2. Click “Select Kernel” → choose “Colab”
3. Sign in with Google account
4. Code runs on Colab runtime; results display in VS Code

Current limitations (as of Feb 2026):

• Cannot mount Google Drive (use Git repos or cloud storage for datasets)
• Colab Secrets not available (use environment variables or .env files)
• Session timeouts same as browser Colab (reconnect as needed)

Faculty note: Assignments should be .ipynb files stored in GitHub repos. This works identically whether students use browser Colab or the VS Code extension. Do not design assignments that depend on Google Drive mounting.

4. Google AI — Gemini Advanced + NotebookLM Plus (Free Student Offer)

What: Google’s full AI suite, free for 12 months for college students. Includes Gemini Advanced (Google’s top-tier model), Deep Research, NotebookLM Plus, and 5TB Google One storage.

Cost: Free for 12 months via the Google AI student offer.

How to claim: blog.google/innovation-and-ai/products/gemini-app/student-offer-google-ai/

MSBAi directive: Students will be asked to claim this offer during onboarding. Faculty should assume students have access to Gemini Advanced, Deep Research, and NotebookLM Plus throughout the program year.

Note: The Gemini app subscription does NOT resolve the Colab built-in Gemini credit limitation. Colab’s AI credits are a separate pool. For AI-assisted coding in Colab, use GitHub Copilot or Lumen (§2).

5. Claude / ChatGPT — General AI Assistants

What: General-purpose AI assistants for explanation, debugging, brainstorming, and writing.

Cost: Free tiers available for both. Students may also access these through Copilot’s model selector, or via a full subscription through their Google AI plan (Gemini Advanced).

MSBAi policy: No vendor lock-in. Course materials should reference “AI tools” generically. When demonstrating specific features, show at least two platforms. Students choose their preferred tools and document usage in AI attribution logs.

6. GitHub — Version Control & Portfolios

What: Code hosting platform with version control, collaboration, and CI/CD.

Cost: Free. Students also get the Student Developer Pack with additional tools.

MSBAi requirements:

• All project code in GitHub repos (public for portfolio)
• Professional README files for all projects
• Commit history shows individual contribution (used in team project assessment)
• GitHub profile becomes the student’s professional portfolio

Student Developer Pack includes:

• GitHub Copilot (free for verified students, see §2)
• GitHub Codespaces (60 hours/month cloud dev environments)
• Free .me domain (Namecheap)
• Various cloud credits and tool access

7. BI / Visualization Tool — Course-Level Choice

No program-wide BI tool is mandated. Faculty select the tool appropriate for their course:

• Power BI Desktop — free, academic license via Illinois. Good for executive-style dashboards.
• Tableau — available via Illinois site license.
• Python (plotly, streamlit, dash) — preferred in code-first courses (BADM 554, FIN 550, BADM 576).
• AI-native visualization — BADM 557 uses AI-generated charts and frameworks; no dedicated BI tool required.

Faculty should document their tool choice in their course syllabus.

8. WRDS — Financial & Accounting Data

What: Wharton Research Data Services. Access to Compustat (firm financials), CRSP (stock returns), and other research databases.

Cost: Program-paid annual license (College expense, not student expense).

Used in: FIN 550 (primary), potentially other courses.

Faculty note: WRDS access should be used alongside open datasets and APIs (yfinance, SEC EDGAR, Kaggle, Census Bureau) so students learn both licensed and open data workflows. Having WRDS as a program-wide resource is a branding differentiator — “Gies provides access to institutional-grade financial data.”

9. GCP — Cloud Data Platform

What: Google Cloud Platform, specifically BigQuery (serverless data warehouse) and supporting GCP services (GCS, Cloud Functions).

Cost: Free tier (BigQuery: 10 GB storage + 1 TB queries/month free). No credit card required for basic use; GCP free tier sufficient for course projects.

Used in: BADM 554 (BigQuery for SQL, ETL deployment), BADM 558 (BigQuery + Snowflake + full GCP stack)

Why BigQuery over Cloud SQL: Serverless — no instance to provision or manage. Browser-accessible via GCP console. Columnar warehouse model aligns with analytics use cases across both courses. Consistent GCP thread from first course (BADM 554) through data engineering (BADM 558).

Proposed Tools (Under Evaluation — Not Yet Adopted)

These tools are being evaluated for possible use in MSBAi courses. They are not part of the standard toolset and have not been formally vetted. Faculty should not assign them until they are moved to the confirmed stack above.

10. Mindforum — Collaborative AI Workspace (Proposed)

What: A shared workspace for small groups (2–6 participants) with an AI collaborator that only speaks when explicitly mentioned with @ai. Groups can upload documents as context — readings, datasets, case studies, project briefs.

URL: mindforum.illinihunt.org

Cost: Free (Gies infrastructure). No install required — browser-based.

When to use in class:

Design principle: The @ai mention model enforces the pre-AI/post-AI discipline — students must decide when to invoke AI, rather than having it generate responses automatically. This aligns directly with AIAS Level 2–4 design and the Critical Engagement principle.

Faculty note: Mindforum is most powerful when the group is doing genuine deliberation before invoking @ai. Assign tasks where the human discussion itself is the first deliverable — @ai is called to challenge, extend, or evaluate, not to start.

11. Cognitive Swarm — Real-Time Group Brainstorming (Proposed)

What: A multimodal brainstorming application for large groups. Participants voice ideas; AI extracts concepts and positions them in 3D space by semantic similarity. Three background agents run simultaneously (connection-finder, critic, direction-suggester). Groups vote using quadratic voting, then synthesize top ideas into a diagram.

URL: agentlab.illinihunt.org/cognitive-swarm

Cost: Free (Gies infrastructure). Browser-based, real-time.

Three phases:

1. Explore — participants speak ideas; AI anchors concepts in semantic space; background agents find connections, critique weak reasoning, and suggest new directions
2. Vote — quadratic voting: fixed credit budget per participant; extra votes on the same idea cost progressively more, preventing any single voice from dominating
3. Forge — system synthesizes top-weighted ideas into a shareable diagram (flowchart, mindmap, ER diagram, or journey map — format inferred from topic)

When to use in class:

Pedagogical notes:

• Quadratic voting is itself a curriculum topic — a decision-making mechanism worth naming explicitly. Students learn about it by using it.
• The three background agents (connector, critic, direction-suggester) are a live multi-agent system. In Agentic AI, flip the sequence: use it first, then deconstruct the architecture.
• Voice input lowers the barrier to participation for students who hesitate to type in group settings.
• The Explore → Vote → Forge sequence enforces the diverge-before-converge discipline that Studio sessions aim for.

Student Onboarding Checklist

This is the setup sequence for new MSBAi students during pre-program orientation:

Total setup time: ~60 minutes (excluding bridge module)

Fallback: Students who have not yet provisioned a base AI tool should prioritize setting up GitHub Copilot (free, §2 step 1) or Lumen (free, §2 step 2) immediately. Colab’s built-in Gemini AI has very limited free credits and is not a viable substitute for class activities.

Local install fallback — Illinois Anywhere: Students who cannot install software on their primary work computer (e.g., employer-managed devices) can access a virtual Windows machine through the Illinois Anywhere service, which will have the required tools pre-installed. No action needed for most students; flagged during onboarding if local install is blocked.

Low Floor / High Ceiling: See DESIGN_PRINCIPLES.md Principle 3 for the full progression map from Colab (zero install) to VS Code + Copilot Agent Mode.

Faculty Guidelines

When designing course materials:

1. Assignments must be .ipynb files in GitHub repos. This works for both browser Colab and VS Code users.
2. Do not depend on Google Drive mounting. Use Git repos, direct downloads, or API calls for data access.
3. Reference “AI tools” generically in syllabi. When demonstrating, show Copilot for coding tasks and Claude/ChatGPT for research/writing.
4. Include AI attribution requirements in all project rubrics. Template in design/assessment_strategy.md.
5. Studio sessions should demo VS Code + Copilot workflows — this is where students see the tools in action and build fluency.
6. Bridge modules handle setup. Courses can assume students have VS Code, Copilot, and Colab working by Week 1.

Tool support responsibilities:

This is the standard tools reference for the MSBAi program. All course technology stacks should align with this page. For program-level details, see program/curriculum.md. For assessment policies around AI tool usage, see design/assessment_strategy.md.