ACQ is AI-first. This is your IT team's field guide to the tools you have, when to use them, and the habits worth building. Bookmark it. Come back to it.
#ai-lab in Slack or submit a tech ticket.Fast, light, and more capable than most people expect.
Great for
Quick questions, short summaries, simple edits, drafting outlines, anything you'd ask a knowledgeable colleague in passing.
Your everyday default. Handles the vast majority of real work.
Great for
Writing, research, analysis, coding, emails, reports, anything requiring real depth.
Deep reasoning. For when the stakes are genuinely high.
Great for
Complex analysis, architectural decisions, high-stakes work where being wrong has real consequences.
The heavyweight of the Claude 5 family: the most capable model available, and the most costly to run. Reserve it only when you truly need it — for complex, multi-stage, or long-running work like heavy software engineering, deep data analytics, and autonomous coding builds. Reach for it when Opus isn't enough, not as a daily driver.
All available to you. Each one has its sweet spot. The fastest way to maximize your time, tokens, and output is to mix them — use the right tool for each job, and don't be afraid to bounce between them.
Your primary workspace at ACQ. Best for writing, research, analysis, coding, and most professional work. Three models, covered above. Beyond the chat interface, four modes worth knowing:
Projects
The most underused Claude feature. Upload your files and instructions once, and every chat inside the Project starts with full context. Stop re-explaining yourself every session. Shareable with teammates.
Cowork
Describe an outcome, Claude works through your local files and delivers finished output. Scheduled tasks too. Use it when you want to delegate, not chat. Setup guide.
Design
Describe a deck, one-pager, or prototype. Claude builds it on a canvas you can edit. Exports to PDF, PPTX, Canva, HTML. Runs on Opus 4.7, burns tokens fast — precision tool only.
Code
Terminal-based. Reads your codebase, writes and runs code, handles git. Don't blindly accept changes — review what it's doing before you approve, especially anything touching auth, infra, or production.
A strong general-purpose tool and a great second opinion. ChatGPT Business runs on GPT-5.3 and 5.5 with effectively unlimited base model usage. The advanced reasoning models (Thinking, Pro) have their own weekly caps, so save those for the work that needs them.
Deep Research
GPT browses multiple sources, synthesizes them, and produces a structured report. Good for competitive research or market overviews when you need breadth fast.
Canvas
A document editor alongside the chat. Click into a paragraph and ask GPT to rewrite just that section. Good for longer writing and editing work.
ACQ Custom GPTs
Your workspace has purpose-built GPTs made by your teammates. Find them under "Popular at Acquisition.com" when you open ChatGPT:
A custom AI advisor built on GPT and trained on Alex's content — $100M Offers, $100M Leads, Models, and workshop material. Has its own POV and voice, so it's a different feel from a general assistant.
Best for: offer building, pricing strategy, sales positioning, lead generation frameworks, anything where you want an answer grounded in Alex's frameworks. You can use it for general work too — but Claude and ChatGPT are usually stronger picks for those.
| Claude | ChatGPT | ACQ AI | |
|---|---|---|---|
| Best for | Deep work, writing, analysis, coding | General tasks, research, image gen, second opinion | Alex's voice and frameworks |
| Login | SSO via Google | SSO via Google | Google OAuth — ai.acquisition.com |
| Usage limits | Managed by IT at the org level (Enterprise) | Effectively unlimited base; Thinking/Pro models capped weekly | N/A |
| Install via Iru | Yes | Yes | No, browser only |
Scroll to see more →
Voice dictation across your entire Mac. Press your hotkey, speak, and Flow handles the rest — punctuation, filler words, tone. It reads context: a Slack reply comes out casual, an email comes out professional. Same thought, zero reformatting.
Best for: brain dumping before an idea disappears, replying to Slack without breaking your concentration, drafting emails hands-free, capturing notes mid-thought. Any time your brain is moving faster than your fingers, this is the right tool.
AI can draft, search, and take action across your tools. The more access you give it, the more it does for you — and the more a mistake can cost. Here's how to think about how much rope to give it, and how to connect it to your systems without getting burned.
Words worth knowing.Hallucination
When AI states something plausible-sounding but flat wrong — and does it confidently. Always verify anything legal, financial, or client-facing against a real source.
Connector
An integration that gives AI access to a system — your email, Slack, Drive — so it can read data or act there on your behalf.
Context window
How much AI can hold in one conversation. Go past it and earlier parts get dropped. Feed it only what it needs.
System prompt
Background instructions we configure that set a tool's role and behavior. You won't see it, but it's shaping every response.
A connector lets AI act inside your systems. That's where the real time savings are — and the real risk. A few rules before you wire anything up:
The biggest difference between fast and slow AI users isn't which model they pick — it's the prompt they wrote. A sharp prompt saves time, cuts tokens, and often beats reaching for a heavier model. Build the habit: be specific, give context, state what "done" looks like. Anthropic's prompting guide goes deeper if you want to learn more.
Upload your context and instructions once. Every chat in that Project starts fully loaded. Stop re-explaining yourself every session.
When you're refining the same output — "make it shorter, cut the second paragraph" — stay in the chat. Claude works best with feedback.
Long conversations burn more tokens and slow Claude down. When you move to something different, start fresh.
When Claude builds something substantial — a doc, a script, a table — ask for it as an artifact. It opens in a side panel you can edit and download directly.
AI hallucinates — it can state wrong facts, made-up numbers, and outdated information with complete confidence and zero warning. Treat every output as a fast first draft. Verify anything factual before you act on it, especially anything you'll send externally or use to make a decision.
Got a repeatable task — summarizing calls, polishing emails, automating small workflows? Teach Claude how to do it once. Think of it like onboarding an intern: a skill encodes the how of one focused job, not a whole workflow. In Claude, set this up with a Project. In Claude Code, save it as a focused SKILL.md file Claude loads on demand. Anthropic's complete guide to building skills goes deeper.
Want a specific format or tone? Paste an example. One concrete sample beats three sentences of abstract instruction — Claude matches patterns faster than it follows rules.
This guide covers the tools. The bootcamp covers the foundations. The #ai-lab channel covers the tactics.
Use all three. The fastest people at ACQ aren't waiting for the next training to learn something new. They're watching, posting, and shipping.
The 12-week bootcamp ran April through July — six sessions on AI literacy, prompting, responsible AI use, research, workflow redesign, and agents, capped by the July 9 hackathon. Missed one or want a refresher? Every recording lives on the AI-First Notion home.
The bootcamp's prompting session ran Tuesday, May 5 — recording available on the AI-First Notion home. #ai-lab also has practical examples from your team.
Use cases, AI Corner videos, prompt libraries, and the weekly AI kickoff post.
Going deep on a tool? Skip the third-party blogs and read the source: Anthropic's prompting guide and OpenAI's ChatGPT prompting guide.
Our July 9 hackathon is a wrap — cross-functional teams built real tools in a single day. Round two will come back around. Watch #ai-lab for what's next.
#ai-lab Slack channel is the hub for the AI-first initiative — use cases, resources, training notes, and a weekly AI kickoff post. All recordings and materials live on the AI-First Notion home.
/tech — that opens the Slack bot menu to create a tech ticket. Or visit orca.acq.com/tickets directly.