Appendix B — The AI Tool Directory

“but test everything; hold fast what is good.” — 1 Thessalonians 5:21 (ESV)

This is the directory behind three of the busiest weeks in the course — Week 5 (Choosing the Right Tool), Week 9 (Image & Video), and Week 10 (Voice & Audio). When a lab tells you to “reach for an image tool” or “pick a research assistant,” this is where you find the names, sorted by the job they do rather than by the company that makes them.

One habit this appendix is built to break: saying “I use ChatGPT” as if AI were a single product. It isn’t. It’s a shelf of specialized instruments, and the professional who thrives is the one who knows which one to pick up. That’s the whole spine of the book, and it applies to this page more than any other:

You choose the tool. You own the verdict. AI drafts, generates, and accelerates; you decide, verify, and are accountable. AI is an assistant, not an authority.

⚠ Read this before you trust a single line below. Every name, version, price, and free-vs-paid tier in this appendix is a dated snapshot — “as of mid-2026” — and this corner of the world moves weekly. Models are renamed, prices change, free tiers shrink and grow, whole products get discontinued between one term and the next. Never treat this directory as current pricing or a purchase decision. Always open the vendor’s own page and confirm before you sign up, pay, or paste anything in. Trust the category and the judgment; verify the specifics yourself.

Start in BoodleBox

If you’re at Concordia, you already own the toolbox — it’s BoodleBox. Concordia University Wisconsin licenses BoodleBox (box.boodle.ai; sign in with your Concordia account) for exactly the kind of work this directory is about, and it puts many of the tools below inside one secure, FERPA/SOC-2, no-train-on-your-data app. So before you go shopping down this page, look for the job you need inside BoodleBox first — most weeks, it’s already in the room.

The trick is knowing which in-app option matches which category here. Open a chat, and wherever a lab says “open ChatGPT” or “reach for an image tool,” reach instead for the matching piece of BoodleBox. Exact labels move — this is all as of 2026 — so aim for the action, not the pixel:

Because BoodleBox is the vetted, no-train, FERPA/SOC-2 option Concordia stands behind, it’s also the appropriate place to put school or work content — a point Week 14 and Appendix C develop. (Even so: don’t paste needless personal data anywhere, BoodleBox included.)

Beyond BoodleBox. BoodleBox is text- and image-first, so a few categories on this page live outside it (as of 2026): video (B.3), voice & audio (B.4), music (B.5), live meeting notetakers (B.8), full deck-builders (B.7), and live translation (B.9). For those — and any time BoodleBox doesn’t cover what a job actually needs — the sections below are your map: the same job-first, “choose the tool, own the verdict” thinking, just reaching for a standalone product instead. And the honest fallback: no BoodleBox access — off campus, or a job it doesn’t include? Every category below has at least one option with a genuine free tier (a few individual tools are paid, and are marked), so this book still functions without the license. Appendix A walks you through the BoodleBox on-ramp; this directory covers everything else.

How to read this directory

Each section below follows the same shape so you can scan it fast:

• What it’s for — the job this category of tool does.
• The leading options — a short table: the tool, what it’s best at, and free-vs-paid at a glance.
• How to choose — one line to point you at the right one.

A few conventions, all hedged as of mid-2026:

• “Free tier” means there’s a no-cost way to try it, usually rate-limited and often running a smaller model. It’s plenty for learning. It is not the vendor’s best model — the strongest tier almost always sits behind a paid plan.
• Most professionals meet these as ~$20/month consumer plans (the “Plus / Pro / Premium” tier), with heavier “Pro / Max / Ultra” tiers running roughly $100–250/month. ⚠ These numbers drift — the exact figure is on the vendor’s pricing page, not in this book.
• Everything here can be wrong confidently. Every generative tool can hallucinate — state a false thing in a fluent, convincing voice. That’s Week 8’s lesson, and it never stops applying. Verify before you ship.

Coach’s Note — Don’t collect tools like trophies. You do not need an account with every name on this page — you need one general assistant you know well, plus the two or three specialists your actual work calls for. A carpenter owns a lot of tools but reaches for five of them all day. Learn your five cold. Add a sixth only when a real job demands it.

B.1 — General assistants

What it’s for. The everyday workhorse — the chat box you open for writing, thinking, summarizing, drafting email, brainstorming, coding help, and quick questions. If you own exactly one AI tool, it’s one of these. Each is a family of models under one app, and each can also handle images, files, and voice to varying degrees.

Three more worth knowing by name: Perplexity (an answer engine — see research assistants, B.10), Grok (xAI — strong on real-time info via live web/X, and more willing to engage edgier prompts), and DeepSeek (free and capable, but a Chinese service — raise data-residency questions before feeding it anything sensitive).

How to choose. Pick one as your daily driver and learn its quirks; reach for Claude on long documents and careful reasoning, Gemini when your work already lives in Google Workspace, and ChatGPT when you want the broadest all-round default. This category is the subject of Week 2 and Week 3 — go there for how the model families differ.

B.2 — Image generation

What it’s for. Turning a text description (a “prompt”) into an image — marketing visuals, slide art, concept mockups, illustrations, social graphics. Prompting an image tool is its own skill (subject + style + lighting + composition + aspect ratio); that’s the Week 9 lab.

Also around: Imagen (Google, reliable in-image text, inside Gemini) and Grok Imagine (xAI, looser content rules).

How to choose. For anything a client or employer will publish, favor Firefly for its licensing and indemnification; for polished default results reach for GPT Image; for a specific artistic look, Midjourney. And read Appendix C before you publish — provenance, likeness, and “don’t pass a synthetic image off as real” are not optional.

B.3 — Video generation

What it’s for. Generating short video clips from a text prompt or a still image — now with synced audio. This is the fastest-churning category in the whole directory; treat every specific below as provisional.

Also common: Kling (multi-shot cinematic sequences), Pika (consumer-friendly short clips), and Grok Imagine (video in the Grok app).

How to choose. For a quick realistic clip with sound, try Veo or Sora; for shot-by-shot creative control, Runway. ⚠ Version names and even whether a product still exists as a standalone change monthly here — verify on the vendor page first. And because realistic synthetic video is now trivial to make, deepfake literacy is a professional duty, not a curiosity — see Week 9 and Appendix C.

B.4 — Voice & audio

What it’s for. Turning text into lifelike speech (narration, e-learning, accessibility, dubbing), cloning a voice, or turning documents into listenable audio. This is the Week 10 territory.

Also worth knowing: ChatGPT Advanced Voice (real-time spoken conversation) and alternative TTS engines like Hume, Cartesia, and PlayHT.

How to choose. For studio-quality narration or a cloned voice, ElevenLabs; to hear a set of documents as a discussion, NotebookLM. One hard rule that Week 10 drives home: voice cloning is trivial now, so “I recognized their voice” is no longer proof of anything. Clone only a voice you have explicit consent to use, disclose that it’s synthetic, and verify money-movement requests out-of-band. The consent-and-disclosure discipline lives in Week 10 and Appendix C.

B.5 — Music generation

What it’s for. Generating a full song — lyrics, vocals, and instrumentation — from a text prompt. Useful for scratch tracks, background music, jingles, and demos.

How to choose. For a full song with sung lyrics, Suno; for cleaner instrumental output, Udio.

⚠ Caution — the law is unsettled here. As of mid-2026 the AI-music companies are in active copyright/training-data litigation with major record labels (settlements and ongoing suits both exist). This book does not advise you on the legal use of AI-generated music. If a track is going anywhere public or commercial, treat the rights as an open question and get real legal guidance — don’t assume “the tool let me make it” means “I’m cleared to use it.”

B.6 — Document AI (upload, summarize, ask)

What it’s for. Working against your own documents — upload a long PDF, contract, report, or set of files, and ask grounded questions, get a summary, or extract specific facts. The professional win is real; so is the confidentiality risk (know your org’s policy before you upload anything sensitive).

How to choose. Use NotebookLM when you need answers cited back to your own sources (research, study, compliance); use a general assistant’s file upload when you want to act on the document — draft a reply, rewrite a section, pull out action items. Either way: a summary can quietly invent a claim the source never made. Ground every important line back to the document — that’s the Week 7 discipline.

B.7 — Slides & presentations

What it’s for. Turning a brief, an outline, or a document into a themed, laid-out, editable deck — board updates, pitch decks, training slides. This is the Week 5 project’s category, and the reason it exists at all: a general chatbot hands you bullet-point text you’d still have to lay out by hand, while these tools hand you actual slides with theme, layout, and images already in place.

How to choose. For a fast deck straight from a brief, start with Gamma; if your workplace already runs Microsoft 365 or Google Workspace, use the assistant built into the deck tool you already present from; reach for Canva when visual polish and templates matter most. ⚠ This category churns fast — names, features, and free tiers shift constantly; check the vendor’s page before you commit. And whatever builds the deck, Week 5’s rule stands: these tools will confidently invent a statistic you never gave them, so hunt every number and claim before it goes in front of a room. The tool makes the slides; you make sure they tell the truth.

B.8 — Meeting AI (notetakers & transcription)

What it’s for. A bot that joins your video call, transcribes it, summarizes the discussion, and pulls out action items with owners. Great for the meeting you couldn’t fully attend or the notes you never write up.

How to choose. Match the tool to the meeting platform you actually use and to your organization’s rules — not to a feature list. A bot silently recording a call is a consent and policy question first, a convenience second. Tell the room it’s recording, and confirm your employer permits it before you invite one in. See Appendix C on consent and confidential data.

B.9 — Translation

What it’s for. Translating text, documents, and live speech across languages — for email, documents, and increasingly real-time meeting interpretation.

Also seen for live meetings: Wordly, JotMe, Apple Translate, and Papago.

How to choose. For the highest-quality written translation (especially EU languages), DeepL; for the widest set of languages or a quick free lookup, Google Translate; for live spoken translation in a meeting, a voice-capable option like DeepL Voice or Wordly. Have a fluent human check anything high-stakes — a mistranslated clause in a contract is your problem, not the tool’s.

B.10 — Research assistants

What it’s for. Answering a question with sources. At the light end, a search-plus-citations “answer engine.” At the heavy end, a “Deep Research” mode where an agent browses many sources over several minutes and writes you a cited report.

How to choose. For a quick, cited answer, Perplexity; for a long researched brief, whichever general assistant you already pay for offers a Deep Research mode. ⚠ The non-negotiable rule: verify the citations. These tools can cite a source that doesn’t say what they claim — or a source that doesn’t exist at all. Open the links. This is the noble-Berean discipline of Week 7 and the verification core of Week 8.

B.11 — Coding assistants (overview only)

What it’s for. Helping write, explain, and fix software. You are non-technical and this course does not ask you to code — but you should know this category exists, because AI now lets a non-programmer describe a small automation or script and get working code back. The catch: AI-generated code can be subtly wrong or insecure, and it needs review. “AI proposes, a human gates, tests validate.”

How to choose. For a working professional, the honest answer is usually: know it’s possible, and get a technical colleague to review anything you generate before it touches real systems. If you want to go deeper, that’s a different course — the Litman Coding series exists for exactly this. For now, file it under “AI can do more than write prose,” and keep the human in the loop.

B.12 — Agent modes (multi-step, tool-using AI)

What it’s for. Week 13’s category: an AI you hand a goal instead of a prompt, which then plans, acts (searches, reads, compiles), observes, and repeats — with your approval gate in the loop. Start in BoodleBox: BoodleBot (@boodlebot) orchestrating specialized bots in one chat — plus a Box when people belong in the loop — is the accessible, vetted “agentic” experience this course uses by default.

How to choose. The Week 13 rule travels with you: the more autonomously a tool can act, the more you need the approval gate — delegation transfers the work, never the accountability. Log what you delegated (the agent-log from Week 13) and verify before anything ships.

B.13 — Running AI on your own machine (local models)

What it’s for. Week 11’s private path: a model that runs entirely on your computer, so nothing leaves the room. The full walkthrough — LM Studio (lmstudio.ai, the gentle app) and Ollama (ollama.com, the command-line staple), what hardware you need, and the no-install fallback — lives in Appendix A, section A.5. Both are free; you pay only in disk space and electricity.

Sizing up a tool that isn’t on this list

New tools launch every week, and half of them are noise. When a colleague or an ad pushes one at you, you don’t need this directory to have heard of it — you need a way to judge it. This is the Week 5 skill: evaluate the tool before you adopt it. Run it through five questions:

1. What job does it actually do, and do I have that job? If you can’t name the recurring task it saves you, you don’t need it. A tool without a job is a hobby.
2. Who makes it, and what happens to my data? Read the data policy. Does it train on your inputs? Where is the company based? Would I be comfortable with this vendor holding what I’m about to paste? (When in doubt, see Appendix C.)
3. What’s the real cost? Free-to-start often becomes paid-to-use. Find the pricing page, not the landing page, and check what the useful tier actually costs.
4. How do I verify its output? If you can’t check whether it’s right, you can’t be accountable for it — and you own the verdict regardless.
5. Is it better than the tool I already know? A tool you’ve mastered usually beats a shinier one you’d have to learn. Switch for a real gain, not for novelty.

If a tool clears those five, try it on one real task and compare. If it doesn’t, you’ve saved yourself an account, a password, and a distraction.

A closing word

Notice what this whole directory is really teaching. It is not a shopping list to memorize — half of it will be out of date by the time you finish the term. It is a way of seeing: a job to be done, a short list of candidates, an honest read of what each is best at, and a decision you own. That habit outlives every product name on this page.

Coach’s Note — “Test everything; hold fast what is good” is not a slogan; it’s a workflow. Test the tool against your real task, on your real standards. Keep what serves the work and your neighbor well; drop what’s merely shiny. The verse assumes a person doing the testing and holding — a person with judgment and a conscience. That person is you, and no tool on this page can be that for you. Choose the tool. Own the verdict.

Up next: the on-ramp is Appendix A — Setting Up Your AI Toolkit (free accounts, no install); the guardrails are Appendix C — Responsible & Ethical AI Use; the plain-language definitions are Appendix D — Glossary. Then head back to the chapters and put the right tool in your hand. See you on Monday.