Module 02: Daily AI Tools

"Technology is best when it brings people together." — Matt Mullenweg (useful reminder: tools should serve your outcomes, not the reverse)

Opening scenario: It is Monday morning. You need client talking points, a lesson plan tweak, a budget email your CFO will actually read, and a summary of a 40-page PDF before a noon call. Colleagues name-drop ChatGPT, Claude, Gemini, and Copilot. You have twenty minutes to pick a workflow. This module is your field guide: what each family of tools tends to do well, how to sign in and run a first serious session, and how to choose without religious wars.

Learning Objectives

By the end of this module, you will be able to:

• Sign in and complete productive first sessions in ChatGPT, Claude, Gemini, and (where available) Microsoft Copilot.
• Match tasks to tools using practical heuristics—not hype charts.
• Apply prompt habits that improve clarity, tone, and verifiability for office work.
• Compare free vs. paid tiers at a decision level (not feature laundry lists).
• Describe data-boundary basics: what to paste where, and when to use enterprise accounts.
• Design a personal default stack (e.g., drafting + research + calendar) you can explain to a skeptical IT lead.

1. The Big Four Families—and What "Good" Looks Like for You

ChatGPT (OpenAI): Broad general assistant; ecosystem of integrations varies by plan and region. Often strong for iterative drafting, role-play coaching, and structured outputs (tables, outlines).

Claude (Anthropic): Often chosen when users paste long documents (policies, briefs, transcripts) and want careful rewrites or critiques. Still requires verification for facts.

Gemini (Google): Tightly coupled to Google accounts and Workspace in many setups; search grounding (when available) can help with recency. Check your organization's Google AI / workspace policies.

Microsoft Copilot: Appears in Edge, Windows, and Microsoft 365 experiences depending on license. Value jumps when organizational data is appropriately and transparently connected—admin controls matter.

Real World Example

A small bakery owner uses ChatGPT for caption ideas but keeps today’s prices on a paper clip by the register — the model never invents specials. Lesson: AI for language, human for facts that change daily.

Real-world examples

• Lawyer (non-advice workflow): Uses Claude to tighten a client-safe blog post; verifies every statute reference manually.
• Teacher: Uses ChatGPT to generate wrong answers for a "find the error" activity; uses Gemini to compare reading levels of two news articles.
• Agency account manager: Uses Copilot inside Word to reshape meeting notes into a client email; strips internal names before prompting if policy requires.

Table: Quick selection heuristic

Try This Now

1. Open two tools you rarely use. Ask both: "Summarize how you should be used responsibly in a corporate setting in 8 bullets."
2. Highlight the best caution each tool gives—did either omit privacy?

Discussion prompts

• Which tool defaults does your organization allow today—and who decided?

2. ChatGPT: Walkthrough for Non-Technical Professionals

Step 1 — Account: Go to https://chatgpt.com. Create or sign in. If your company offers enterprise, use that email—data handling may differ.

Step 2 — New chat hygiene: Start a fresh chat per project to reduce cross-talk. Name the chat in the sidebar if your app allows.

Step 3 — Prompt skeleton:

• Role: "You are an experienced editor…"
• Task: "Turn these bullets into a coherent email."
• Audience: "Recipient: CFO; tone: concise, respectful."
• Constraints: "No exclamation marks; max 180 words."
• Format: "Use subject line + 3 paragraphs."

Step 4 — Iterate: Ask for two variants (A/B). Ask what could confuse the reader.

Step 5 — Verify: For numbers, dates, laws, confirm outside the chat.

Scenario: HR drafts a policy clarification. ChatGPT produces smooth prose. Risk: old regulations mixed in. Fix: paste only the current excerpt from the official handbook and ask for rewrite without adding requirements.

Table: ChatGPT habits

Try This Now

• Upload or paste a sanitized internal paragraph. Ask: "Improve clarity only; do not add facts; flag any sentence you cannot verify from the text."

Discussion prompts

• When is "uploading a document" not worth the convenience?

3. Claude: Walkthrough for Documents and Delicate Tone

Step 1 — Sign in: https://claude.ai with your account. Confirm whether your workspace is personal or organization-managed.

Step 2 — Long context: Paste a table of contents first, then sections one at a time if you hit limits. Ask for a running summary you approve before the next chunk.

Step 3 — Tone work: Useful prompts:

• "Preserve voice; reduce jargon; keep empathy."
• "Mark any claim that would need a citation in a policy memo."

Scenario: A principal pastes a parent letter about a safety drill. Claude smooths language. Human step: confirm times, doors, and contacts against the official schedule.

Try This Now

• Paste two paragraphs you wrote. Ask: "Where might a reader feel blamed? Suggest neutral alternatives without softening accountability."

Table: Claude vs. ChatGPT (heuristic, not permanent)

Discussion prompts

• Should tone-polishing of sensitive HR or student communications require manager review every time?

4. Gemini: Walkthrough with Google Workspace in Mind

Step 1 — Sign in: https://gemini.google.com with the Google account you are allowed to use for work.

Step 2 — Grounding: If available, enable features that tie answers to search or your content per admin policy. Read the disclosure about what gets sent where.

Step 3 — Recency tasks: Good for:

• "Summarize major vendor changes to Gmail labels in the last year—point me to official help pages."
• "What are sensible defaults for Google Drive sharing in schools?" (Still verify locally.)

Scenario: A marketer compares platform feature claims. Gemini with grounding can suggest primary sources; human clicks through and confirms.

Try This Now

• Ask: "What should I verify before trusting an AI answer about tax deadlines?" Then actually verify one item at irs.gov or your national authority.

Step-by-step: collaborative lesson plan

1. Prompt: "Outline a 50-minute workshop on media literacy for adults; include activities."
2. Prompt: "Replace any activity that requires paid tools with free alternatives."
3. Prompt: "Add instructor notes: common misconceptions and discussion prompts."
4. Human: align with your institution's standards.

Discussion prompts

• How should schools and small businesses decide one paid assistant vs. several free ones?

5. Microsoft Copilot: Walkthrough When You Live in M365

Step 1 — Find Copilot: Windows taskbar, Edge sidebar, or inside Word / Outlook / Teams—availability depends on license.

Step 2 — Know the boundary: Ask IT: Does Copilot reference your emails and files? Under what retention rules?

Step 3 — Meeting-to-action: Typical pattern:

• Summarize meeting notes → extract decisions → draft follow-up email → you verify attendees and dates.

Scenario: A program manager converts a Teams transcript into a RAID log (Risks, Assumptions, Issues, Dependencies). Copilot proposes entries; PM removes hallucinated owners.

Table: When Copilot shines vs. when to export

Try This Now

• If you have access: "Turn these bullet notes into an agenda with time boxes." If you lack access, simulate in ChatGPT and note what metadata you would need (attendees, goals).

Discussion prompts

• What audit trail should exist when AI drafts email in a regulated environment?

Practical Tips That Transfer Across Tools

Privacy discipline: Replace real client names with Client A; redact account numbers; prefer enterprise tiers for work secrets.

Brand voice: Maintain a one-page voice guide (words to use / avoid). Paste it at the start of important chats.

Reproducibility: Save final prompts that worked in your team's wiki—not the raw outputs, which may age poorly.

Accessibility: Ask tools to produce plain language and shorter sentences for public notices; still have a human read aloud test.

Try This Now

• Create a voice guide snippet (10 bullets). Run the same email draft through two tools with that snippet—pick the better human-edited blend.

Activities

1. Account audit: List which assistants you can log into today; note work vs. personal separation.
2. Same task triathlon: Draft a 200-word volunteer recruitment post in ChatGPT, Claude, and Gemini; score clarity, warmth, and accuracy after your edits.
3. Failure hunt: Deliberately ask each tool a trick question (niche local ordinance). Document wrong answers—practice graceful doubt.
4. Prompt library: Write five reusable prompts for your role (email, summary, lesson hook, meeting minutes, FAQ).
5. IT conversation: Schedule 15 minutes with IT or compliance using questions from this module—bring notes, not anxiety.

Your Challenge

Build a "Daily AI Stack" one-pager for your real week:

• Three recurring tasks you will run through AI (e.g., meeting recap, rubric drafting, client FAQ).
• Which tool is first choice for each—and backup if access fails.
• Verification rule for each task (what you will never trust without checking).
• Red lines (data you will not paste).
  Share the one-pager with a peer; they red-team one rule.

Expanded challenge: step-by-step

1. List three real weekly tasks (not aspirational). For each, write first-choice tool and backup if blocked or down.
2. For each task, write a verification rule in one line (“I never send dates without checking the calendar,” “I never cite law without the statute PDF”).
3. List red-line data: what you will not paste into consumer accounts; mirror your IT policy if one exists.
4. Add one “failure mode” per task (wrong tone, stale law, hallucinated name) and how you’d catch it.
5. Peer red-team: They try to convince you to break a red line for “speed”; practice saying no with a better path (enterprise tool, redacted excerpt, IT ticket).
6. Revisit the one-pager in 30 days—tools and policies change; date your version.

Industry snapshots: daily stacks (illustrative)

Comparison table: “one tool everywhere” vs. “horses for courses”

Discussion Corner

1. Onboarding: What will new hires do on day one so they don’t reach for the wrong account (personal vs. work)?
2. Accessibility: How do assistants change expectations for plain language and translation—who checks quality for your audience?
3. Fatigue: When does switching tools hurt more than it helps? Where could you standardize?
4. Managers: What should a manager never ask a direct report to do with AI without written policy cover?

Try This Now (added)

1. Stack stress test: For each of your three tasks, ask your backup tool: “What would you get wrong if I used you for [task]?” Note one genuine limit per tool.
2. Voice lock drill: Paste your voice guide; generate one paragraph; strikethrough any phrase that isn’t you—what rule was missing from the guide?
3. Calendar block: Schedule three 20-minute “tool hygiene” slots this month: account check, prompt library tidy, IT question written down.

Key Takeaways

• Tool choice is task- and policy-dependent; there is no single winner for all work.
• Prompt structure (role, task, audience, constraints, format) beats clever tricks.
• Long documents and delicate tone workflows deserve extra grounding and human review.
• Recency issues mean you must verify laws, prices, and news externally.
• Enterprise controls change what "safe" means—ask IT, do not guess.
• Your personal stack should be documented, not improvised fresh every Monday.

Resources

• ChatGPT
• Claude
• Gemini
• Microsoft Copilot consumer — compare with your work tenant documentation

Bonus: Coping with Plan Confusion

When pricing pages change monthly: Anchor decisions on your needs: document length, privacy, integrations, and support. Run a 30-day pilot with metrics (time saved, error rate, employee satisfaction)—not vibes.

When a tool is blocked at work: Maintain a non-sensitive sandbox on personal accounts for skill-building, while respecting employment agreements. Never circumvent security controls with shadow IT.

Seminar Discussion (Optional)

• What single policy would make AI tools safer without killing innovation in your team?
• Who owns prompt libraries—communications, IT, or each function?
• How will you onboard new hires to tools without normalizing careless pasting?

Try This Now: Meeting Minutes to Actions (90 Seconds per Tool)

1. Paste fictional bullets: "Attendees: Ana, Ben. Decided: move launch; owner: Ana; due: Friday."
2. Ask each tool: "Produce a follow-up email and a task checklist."
3. Compare ownership clarity and missing info prompts (did the tool ask for Friday's date?).

Rubric: Judging an AI Draft Quickly

Cross-Tool Feature Map (High Level)

Verify current features in-product; this space evolves rapidly.

Reflection Journal

1. The tool I underestimated: ___
2. The risk I will watch most: ___
3. One prompt I will reuse weekly: ___
4. One thing I will stop doing (e.g., pasting raw client data): ___

Extended Scenario: The Conference Workshop

You run a 60-minute community workshop on AI tools for small-business owners.

Minute 0–10: Live demo—draft a grand opening social post with constraints.
Minute 10–25: Compare two tools side by side on the same prompt.
Minute 25–40: Table exercise—participants write red lines for customer data.
Minute 40–55: Q&A on myths (Module 01).
Minute 55–60: Handout: three safe starter prompts for their trade.

Use this outline as Your Challenge extra credit if you teach or train.

Appendix: Plain-Language Prompt Starters

• "You are a concise executive coach. I will describe a conflict. Ask one clarifying question, then suggest three response options with trade-offs."
• "Convert this bulleted brainstorm into a project plan with milestones—not tasks invented beyond the bullets."
• "Explain this paragraph to a curious 14-year-old; then to a skeptical expert; keep both under 100 words."

Quality Control Checklist Before Sending AI Text Externally

• Removed placeholders and obvious model boilerplate
• Fact-checked numbers, names, dates, legal claims
• Matched brand voice and inclusive language standards
• Confirmed no confidential or personal data leaked
• Attached human sign-off per policy

When to Escalate to a Specialist

Escalate (legal, IT security, accessibility, subject-matter expert) when outputs affect rights, safety, money, or compliance and you lack personal expertise to verify. The tool is a sparring partner, not a signatory.

Closing Thought

The goal is not to collect tools like trophies. The goal is to finish your Tuesday with better communication, clearer thinking, and documented judgment—using assistants where they earn their place.