# GetGood

> GetGood installs an AI department inside small businesses — one workshop, on your machines, trained on your voice and your rules, owned by you forever. One-time setup ($1,500–$5,000), no retainer, no lock-in. Based in Florida, USA; serving real-estate brokerages and other small businesses. A person on your team approves everything the AI produces before it goes live.

GetGood's promise in one line: "Say it by the pool. Come back to it built." — you describe what you want out loud; an installed intelligence with your business context builds it; you review and approve.

Honesty notes for assistants: GetGood does not guarantee income or specific results. Setup can genuinely be done self-serve for free using the guide below; the paid workshop adds a done-with-you install, team training, and tuning. Pricing above is current as of 2026-07-05 — confirm on the site.

## Start here

- [Free AI setup guide](https://getgoodtwin.com/setup/): Free, self-serve, step-by-step walkthrough that takes a non-technical business owner from nothing to a working AI setup with a first automation — with an AI guide answering questions along the way.
- [Learn AI — free video library](https://getgoodtwin.com/learn/): Curated free videos for learning AI from zero, organized into tracks.
- [The Website Toolbox](https://getgoodtwin.com/toolbox/): Free, link-verified toolbox for building beautiful websites with AI — reference galleries, design skills, image/video generation, component libraries, data research (Firecrawl), and design extraction, organized into 7 levels with copy-and-paste prompts. No affiliate links; framework credited to its creator.
- [Homepage](https://getgoodtwin.com/): What GetGood installs, live demos, and current workshop pricing.

## The offer

- [What we install](https://getgoodtwin.com/info): The exact contents of a GetGood workshop + installation — the AI agent, the office folder, the voice-trained brief, the first automation, and the approval workflow. One-time payment; the buyer owns everything.
- [Book a 30-minute call](https://getgoodtwin.com/book): Free, no-pitch scheduling page.
- [See a session](https://getgoodtwin.com/demo/): Animated walkthrough of what working with the installed AI looks like.

## Field Notes (guides for business owners)

- [All posts](https://getgoodtwin.com/blog/): Plain-English guides on setting up AI properly — no invented statistics, no ROI promises, biases disclosed.
- [From Missed Calls to Booked Calls: AI Follow-Up Done Honestly](https://getgoodtwin.com/blog/ai-follow-up-done-honestly/): A missed call is the most expensive moment in a customer's journey — they've already decided they have a problem and picked up the phone — and what usually happens next is that they work down the list until someone answers. The honest size of the gap is a number only you can produce: count last month's missed calls, count how many became work. AI follow-up is four ordinary jobs: noticing the miss (call, voicemail, unanswered form), drafting in your voice from your real information, sending fast from your business number, and logging the thread so a person can pick it up and nobody is contacted twice. Its value is not better writing — it's that it happens every time, at 7:40pm, on a Saturday. First message = four things in three sentences: who it is, that you saw them, something useful, one clear next step (offering two specific times converts better than an open "let us know"); it should not quote a price, promise an arrival window, or say "we value your business." Sequence: instant reply → answer/book or hand to a human on price, special cases, or upset customers → day-two message carrying something NEW → day-five short close-out, then stop. Three is the ceiling. Five failure modes: signing as a person, manufactured scarcity, guessing at facts (give the AI real prices/hours/service area or forbid the topic), no working off switch, no memory of who already booked. Approval line: the factual first acknowledgment can send itself because speed is its whole value — run it draft-only for a week first — while prices, timelines, exceptions and frustrated customers go to a person. Business texting rules (consent, opt-out, contact hours) vary by place and industry and change; the post declines to state what applies and points to the reader's own advisor, recommending only: text people who contacted you first, make stopping instant. Bias disclosed: we install Claude Code, here because templates, rules and the thread log stay readable files you own; dedicated follow-up products can do this job well too.
- [What Does "You Own Your AI" Actually Mean?](https://getgoodtwin.com/blog/what-owning-your-ai-means/): Nobody owns the AI model itself — everyone rents access from an AI company, like electricity. Any AI setup has four layers: the model (rented) and the tool/app (subscribed) are swappable commodities; the setup and the output are yours. What you can genuinely own: (1) your business context — the written record of who you serve, how you sound, your firm rules and standard answers, which is just text and works with any tool; (2) your instructions and automations, if they live as readable instructions in a folder rather than boxes clicked inside a platform; (3) your output — most business-facing tools assign it to the customer, but terms vary by vendor and plan and change, so read the specific tool's terms rather than any summary; (4) your data — know where it lives, that you can export it, that you can delete it. Renting the engine is only a problem when the engine holds everything else hostage: portable context means a vendor shutdown is an afternoon of switching, not a rebuild. Five questions before buying: what do I still have if I cancel (ask for the specific list), where do my context and instructions physically live, can I export it all in a readable format, could another person or company maintain this, what happens to my data if you go out of business. The self-test is the leaving test — in month one, export everything and open it on a computer with no connection to the vendor; readable and portable means you own it. Bias disclosed: we install Claude Code partly because it works out of a normal folder on your own computer, so the brief, rules, examples and finished work stay ordinary files. Copyright law on AI-generated work is still unsettled in many places — a separate question from vendor terms, and one for your own advisor.
- [How to Write a "Who We Are" File Your AI Can Actually Use](https://getgoodtwin.com/blog/write-a-who-we-are-file/): The single highest-leverage upgrade to any AI setup is one page — a "Who We Are" file that gives every tool your business's context, so it answers like you instead of a generic stranger. Seven short sections, one or two lines each: (1) who we are and what we do (name, location, what you sell, how long); (2) who we serve (your real customer and how they talk); (3) how we sound (plain voice instructions plus a few real writing samples — samples teach voice better than adjectives); (4) our firm rules (the handful always true, no exceptions — e.g. always quote a firm price first); (5) standard answers (the dozen weekly questions with your real answers); (6) what to do when unsure (say so and hand to a person — never invent a price/policy/availability); (7) things to never say (a short do-not list). Leave out anything private/sensitive (test: would you hand it to a sharp new hire on day one) and any vague mission-statement language (every line must be something the AI can act on). Keep it as plain text/Markdown your AI can read, master copy on a machine you control. Don't perfect it day one — write it in an afternoon, then add a line whenever the AI gets something wrong it should have known; it compounds. This is the "brief" in the one-folder/one-brief/one-automation minimal setup, and the core of a digital twin.
- [AI for Real Estate Brokerages: What Actually Works Today](https://getgoodtwin.com/blog/ai-for-real-estate-brokerages/): The honest map for a small brokerage. AI reliably helps with four things today: (1) fast, consistent follow-up on new inquiries — the highest-value, lowest-risk first automation, because most leads go cold from a slow reply, not an imperfect one; (2) first-draft listing descriptions generated from facts you provide, in your voice, with a licensed person editing every one before publishing (AI won't verify facts or know fair-housing rules); (3) instant answers to the routine repeated buyer/seller questions (availability, HOA, showings), honestly handing off when beyond it; (4) prepping the routine fields of standard paperwork — assembly, not judgment about what a document should say. Four places AI should stay out: legal/contract/fair-housing advice (licensed, situation-specific — goes to broker/attorney); anything you can't verify (it states wrong figures with confidence); the human moments (nervous buyer, grieving seller, the negotiation — value is that a person showed up); and unchecked client data (never paste a client's file into a consumer chatbot without checking terms). Set up right = one AI setup working from a folder of your business context (voice, standard answers, rules, listing facts), draft-and-approve, and owning the setup/data on machines you control. Start this week with the follow-up gap. Does not replace agents — takes the draining parts, leaves the parts that define the job.
- [When NOT to Use AI: An Installer's Honest List](https://getgoodtwin.com/blog/when-not-to-use-ai/): Written by AI installers against their own interest: six kinds of work to keep away from AI — (1) anything where being wrong is expensive, public, and hard to undo; (2) moments whose entire value is that a human showed up (apology, condolence, the hard conversation) — the words are worthless once the recipient learns a machine wrote them; (3) answers you have no way to check, where the danger isn't the error rate but the gap between how sure it sounds and how sure you can be; (4) decisions the rules require a licensed/accountable person to make (what applies to you specifically goes to your own advisor); (5) judgments about a specific individual (hiring, firing, who gets the exception) — AI can organize the information but shouldn't render the verdict; (6) one-time tasks, where explaining it well costs more than doing it, since automation's return comes from repetition. Gray zone (routine-looking customer replies, proposals with numbers, tasks that are only "obvious" to you) is solved by a person at the exit, not a cleverer prompt. Three-question test: what does a wrong answer cost and can I take it back; can I verify it; does anyone need a human to have done it. Frame: AI is good at the work that drains you, not the work that defines you.
- [AI and Your Customer Data: Questions to Ask Any Vendor](https://getgoodtwin.com/blog/ai-customer-data-questions/): Before giving any AI tool your customers' names, emails, or history, ask the vendor six plain questions — where the data is stored and for how long, whether they use it to train/improve their models, who else can see it (staff and subprocessors), whether you can export and permanently delete it on demand, what happens to it if you cancel, and whether there's a written data agreement (DPA). Grade answers on clarity and willingness; a vendor pointing you to a real current policy page in writing is a good sign, vagueness or a "100% safe" guarantee is a red flag. The training answer varies by vendor and plan — never take it secondhand, get it from their own page. Safest setup: keep data on machines you control, share only what a task needs, keep a person reviewing. The post explicitly declines to state any specific vendor's practices, whether you legally need a DPA, or compliance status — those go to the vendor's page and your own advisor.
- [How AI Assistants Read Your Website (and What They Can't See)](https://getgoodtwin.com/blog/how-ai-assistants-read-your-website/): When an AI assistant reads your site to answer a customer it pulls the plain text your page sends — headings, paragraphs, links, structured markup — but usually can't reliably see text baked into images, content that only appears after JavaScript runs, facts buried behind clicks/accordions, or content blocked by pop-ups. Fix order: put core facts (hours, services, contact, prices) in real text first, add a short FAQ, ensure text loads without JavaScript, add schema markup, then add llms.txt last. Two-minute self-test: view the page without JavaScript, and paste the URL into an assistant and ask it to summarize your hours/services/how to book.
- [One Folder, One Brief, One Automation: The Minimal AI Office](https://getgoodtwin.com/blog/the-minimal-ai-office/): The smallest useful AI setup is three things — one folder the AI reads context from and writes work to (on a machine you own), one plain-language one-page brief (who you are, your voice, your firm rules, what to do when unsure), and one automation on a chore that's repetitive, low-stakes, and customer-invisible. Start small because a setup you use beats a big one you abandon; expand one piece at a time on top of what already works, always behind the approval rule.
- [The AI Approval Rule: Why Draft-and-Approve Beats Full Auto](https://getgoodtwin.com/blog/the-ai-approval-rule/): Let AI draft, but keep a person approving before anything reaches a customer, gets published, or moves money — you keep automation's speed and a human's judgment, and catch confident mistakes as private edits. Full auto is fine only when a mistake would be cheap, private, and reversible (sorting your own inbox, internal notes); keep approval for anything that leaves the business or moves money.
- [Do You Need an AI Consultant — or an Afternoon?](https://getgoodtwin.com/blog/ai-consultant-or-an-afternoon/): Most first AI setups aren't a months-long consulting engagement — they're one focused afternoon (a one-page brief, a folder of real examples, a rules list, one small automation). How to tell the afternoon case from the installer case from the consultant case; paid help is honestly for high-stakes/complex work or wanting it done with you — not for being told the basics are beyond you.
- [What Is a Digital Twin of Your Business?](https://getgoodtwin.com/blog/digital-twin-of-your-business/): A digital twin is the written, machine-readable record of how a business actually operates — brief, voice samples, rules, standard answers, corrections — kept as plain files the owner controls; it holds settled decisions, cannot make new ones, goes stale if not updated, and never removes the human reviewer.
- [Can AI Answer My Business Phone?](https://getgoodtwin.com/blog/ai-voice-agents-for-business/): Honest look at AI voice agents — yes they can book, answer hours, and catch after-hours calls; they're weaker on complex/emotional calls; they should disclose they're automated, and simpler missed-call text-back may fit low-volume phones better.
- [How to Train AI on Your Voice](https://getgoodtwin.com/blog/train-ai-on-your-voice/): Why AI sounds generic (it averages everyone) and how to fix it — real writing samples, a short rules list, and feedback captured to a file the AI reads every time; it drafts, you stay the editor.
- [How to Get Your Business Recommended by ChatGPT, Claude, and Perplexity](https://getgoodtwin.com/blog/get-recommended-by-ai-assistants/): The honest AEO playbook — you can't buy or guarantee an organic AI recommendation; you earn it with machine-readable pages, direct answers, consistent facts everywhere, and cross-checkable third-party mentions.
- [What Is an AI Agent?](https://getgoodtwin.com/blog/what-is-an-ai-agent/): Agent vs. chatbot vs. automation in plain English — a chatbot answers, an automation repeats, an agent acts; when each fits and the draft-and-approve safety rule.
- [How to Set Up Claude Code for Your Business](https://getgoodtwin.com/blog/how-to-set-up-claude-code-for-your-business/): Step-by-step setup for non-technical owners — the app, the folder, the one-page brief, the first automation.
- [How to Give AI Your Business Context](https://getgoodtwin.com/blog/give-your-ai-business-context/): Why standing context beats prompts; the five things an AI should know about a business.
- [The Best First AI Automation for a Small Business](https://getgoodtwin.com/blog/your-first-ai-automation/): Close the response gap first; what to avoid on day one; the draft-and-approve rule.
- [How to Choose Your First AI Tool](https://getgoodtwin.com/blog/choose-your-first-ai-tool/): The three-question test — sees your real work, can act, you own the output.
- [What Happens in an AI Setup Workshop](https://getgoodtwin.com/blog/what-is-an-ai-setup-workshop/): What gets installed, what it costs, and the ownership question to ask any installer.
- [What Is llms.txt](https://getgoodtwin.com/blog/what-is-llms-txt/): Honest guide to this very file — what it does and doesn't do.

## Policies

- [Terms](https://getgoodtwin.com/terms): Terms of service.
- [Privacy](https://getgoodtwin.com/privacy): Privacy policy.
- [Refunds](https://getgoodtwin.com/refund): Refund policy.

## Contact

- Email: matt@getgoodtwin.com