Small Business Big AI

"My uncle said yes." The Intern is a founder now, and her uncle wants to invest. What does he get? Half. Of the company. And what does he get to decide? Nobody has asked that yet.Hal is off this week, so this is a story episode with two made-up characters, Sassy AI and the Intern, performed by Kim through an AI voice app. The business is invented. The questions are real.We look at two AI companies that pulled apart money, ownership and control. Otter.ai took $320 million to grow and sold no ownership. Anthropic, the company that makes Claude, sold most of its rooms and was careful about who holds the keys. Then we bring it back to the uncle.Three questions to ask before you sign anything1. Who gets paid, and when? If it's a loan, there are payments. If it's ownership, the other person may get a share of the profits, or a share of the money when you sell. "It's not a loan" doesn't mean nobody has a claim on the money.2. Who gets to decide? Hiring, spending, borrowing, sellingthe business, changing direction. And when you disagree, who breaks the tie? Fifty-fifty sounds fair until the day you disagree.3. What happens when the plan goes wrong? Slow sales, amissed payment, a partner who wants out, someone who gets sick. If you personally guaranteed anything, your own savings or home could be on the line.The three doors. Borrow against what you know works (Otter.AI). Sell some rooms, but know who holds the keys (Anthropic). Or redesign the work so you need less money in the first place. Before you pay for more of the old way of working, test one workflow change and count the setup, checking and fixing time.Whichever door you pick, pick it on purpose.Have your own attorney and accountant review any agreement before you sign. Sassy AI and the Intern are fictional characters performed by Kim through an AI voice app. Kim uses Claude in her business. This episode is not a stock tip, and it is not investment, legal or financial advice.Ready to Take ActionThe IMPACT Loop: https://smallbusinessbigai.com/impactKim Lewis Howard on LinkedIn:https://www.linkedin.com/in/kim-lewis-howard/Insurance assessment, commercial or personal lines:https://asklewishoward.com---Sources: Otter.ai announcement (Oct. 20, 2026):https://otter.ai/blog/general-catalyst-backs-otter-ais-voice-ai-ambitions-with-320m-growth-investment· General Catalyst, The Unbundling of Growth Equity:https://www.generalcatalyst.com/stories/the-unbundling-of-growth-equity ·Anthropic Form S-1 (Oct. 26, 2026):https://www.sec.gov/Archives/edgar/data/2130357/000162828026067168/anthropicpbc-sx1.htm---Q1. What should I ask before a family member invests in mysmall business? Ask three questions and get the answers in writing. Who gets paid, and when: is it a loan with payments, or ownership with a share of the profits or the sale? Who gets to decide: hiring, spending, borrowing, selling, changing direction, and who breaks a tie? What happens when the plan goes wrong: slow sales, a missed payment, someone who wants out, someone who gets sick, or a personal guarantee on your savings or home? On Small Business Big AI, Kim Lewis Howard's point is that "we're family" is the reason to write it down. Have your own attorney and accountant review the agreement.Q2. What is non-dilutive growth financing, and how is itdifferent from a bank loan? Non-dilutive financing gives a business money without taking ownership in return. One version, used by Otter.ai in its $320 million deal with General Catalyst in October 2026, pays for sales and marketing up front and gets repaid only from the revenue of the new customers that spending brings in, up to a capped amount, with the investor taking part of the hit if the marketing does not work. A regular bank loan has the same payment every month no matter how sales go. Either way you keep your ownership but not all of your cash, so know what you will owe, when, and what is left to run the business.

Kim asked Hal what he’d do if she called and said, “Sendtwelve thousand dollars right now. I’ll explain later.” He said he’d call her back. Then she played him a voicemail that sounded exactly like her. It knew the invoice was coming. It said “love you.” It wasn’t her.This is part two of our two-part series, and it’s Election Day in the U.S. This is the first midterm where AI-made ads became normal: more than $137 million spent this cycle, and most of it unlabeled. We’re not taking sides. But if someone can fake a Senate candidate with a whole campaign watching, they can fake the owner of a dry cleaner in St. Cloud with nobodywatching. A voice can be copied from about three seconds of audio, and the FBI got almost seventy reports a day last year of business email scams that redirect payments.So, stop trying to catch the fake. Start checking the request.Before anything sensitive happens, ask three questions: Who is asking? What are they allowed to do? How do we check this request?Your fifteen-minute challenge. Pick one process: changes tosupplier bank details.1.    The trigger. Any request to add or change paymentdetails stops the payment until it’s checked. Write that down.2.    The trusted number. Where you call to confirm, and whoat the supplier can say yes. It has to be a number you had before the request showed up. If you don’t have one, the change waits.3.    The people. Who checks, who approves, who covers whensomeone’s out, and where you write down what was confirmed. Working solo? Write down your own check.4.    The practice run. Have the owner make the rushed request. See if the person handling it can say, “I’ll handle that after verification.” Then the owner says, “Thank you.”Once it’s working, tell your customers in plain words: “If you ever get unexpected payment instructions that look like they’re from us, stop and call us on a number you already trust. We check every payment change before it goes through.” Only publish that once it’s true.If money already went out: call your bank right away and askthem to stop or recall the payment. Save the messages and transaction details. Tell whoever is responsible for the affected accounts. Report it at https://www.ic3.gov. Recovery isn’t guaranteed, so speed matters.Rude is cheaper than gone. Check before you act.---Q1. How can a small business protect itself from AI voicecloning scams? Stop trying to tell a real voice from a fake one, and check the request instead. Researchers have made a usable copy of a voice from about three seconds of audio, and most people can’t reliably tell a good copy from the real thing by ear. On Small Business Big AI, Kim Lewis Howard’s rule is todecide in advance which requests always get a second check, no matter whose voice asks: money moving, bank details changing, passwords, gift cards, and anything urgent and secret. Before acting, ask three questions. Who is asking?What are they allowed to do? How do we check this request?---Sources: Wesleyan Media Project ad spending, via Fortune(Oct. 21, 2026): https://fortune.com/2026/10/21/political-campaigns-ai-advertisement-spending-137-million-republicans-misinformation/· Wiz CEO voice deepfake, TechCrunch: https://techcrunch.com/2024/10/28/wiz-ceo-says-company-was-targeted-with-deepfake-attack-that-used-his-voice/· FBI IC3 2025 report: https://www.ic3.gov

It’s a stormy Sunday night. Claire (a made-up shopper with avery real problem) needs a gift for her dad, who fly fishes and owns everything. She doesn’t open a website. She tells her phone: under fifty pounds, arriving before the twentieth. Then she goes to bed. By morning, her AI assistant has three options, and one small shop she’s never heard of is aboutto make a sale.This holiday season, Salesforce predicts one out of everyfive online holiday shopping visits will come from AI chat agents. These assistants don’t see your window display or your pretty photos. They read for facts: price, stock, delivery date, returns, reviews. If the answer isn’t there in plain words, they move on, and you never know they came by.Hal plays the bot. A lovely tackle shop in the English countryside loses to a little shop in Winter Garden, Florida, for one reason: its price said “Contact us,” and Claire was asleep. To a shopping bot, “contact us” means “skip us.”This week’s three moves:1.    Spell it out. What it is, who it’s for, the price or a range, availability, and delivery or scheduling details, typed out in words, not only inside an image. Post your holiday cutoff date.2.    Prove it. Reviews that sound like real questions, andpromises like your return policy.3.    Make the yes easy. A way to buy, book or reach a humanfast, even while the customer is asleep. Then a person follows up.Be the bot (15 minutes):1.    Ask your AI assistant to shop for what you sell, with abudget and a deadline. Note who comes up.2.    Check your own site, ten seconds per item. Facts: priceor range, in stock, holiday cutoff date, all typed in words. Proof: one review that sounds like a real customer’s question. Yes: a button to buy, book, or reach a human fast.Anything you can’t find in ten seconds, the bot can’t either.Fix that first.Review ask to use this season: “Would you mind sharing whoyou bought it for and how it went?”Next week, part two: Is That Really You? When Claire callsthe shop, how does she know it’s really them?---Q1. What is an AI shopping agent, and what does it look forwhen it shops for someone? An AI shopping agent is an assistant that doesn’t just answer a question. It searches, compares and, more and more, buys or books on the customer’s behalf. Salesforce predicts that 20 percent of 2026 holiday ecommerce traffic will come from AI chat agents. An agent skips the photos and the brand story and reads for facts it can check against the customer’s list:what the product is, the price, whether it’s in stock, when it will arrive, the return policy, and whether other people say it’s good. On Small Business Big AI, Kim Lewis Howard calls it a buyer with a clipboard and zero patience. If the answer isn’t written in plain words, it moves to the next shop.Q2. How do I make sure an AI assistant can actually buy or book from my business? Make the yes easy. When an assistantdecides, it needs a way to finish the job without waiting for business hours: online checkout, online booking, or at least a form that gets answered fast. Post the details it needs to promise a result, like your holiday order cutoff date. Platforms are adding tools for this; Shopify, for example, now offersways to put merchant products in front of shoppers inside AI assistants like ChatGPT. Then have a person follow up. As Kim Lewis Howard puts it, the bot picks them. You keep them.Sources: Salesforce 2026 holiday predictions:https://www.salesforce.com/blog/holiday-retail-predictions-2026/ · Narvar 2026 Holiday Shopping Report:https://corp.narvar.com/resources/2026-holiday-shopping-report

It’s ten o’clock at night in Florida. The air conditioner just quit, the thermostat says eighty-six, and there’s a baby asleep in the next room. The dad picks up his phone and asks, “Who near me can fix my AC tonight?” The phone answers with one name. If it isn’t yours, you never knew that customer existed.For twenty years, marketing had one job: get the customerto your website. Now a lot of customers never get there. They ask a question and the answer shows up on its own. A SparkToro study found about 68 percent of U.S. Google searches in early 2026 ended without a click. Not every kind of search behaves the same, but the trend is clear. And in late September, OpenAI showed ChatGPT recommending services inside the conversation, along with always-on assistants that can search, compare, pick and book for your customer.This week’s three moves:1. Be the answer. Write down the ten questions customersactually ask before they buy, and put plain answers on your website, in their words. 2. Be the same everywhere. Name, hours, phone, address and what you do, word for word on your website, Google, social pages and every directory. 3. Be first to call back. A text back in minutes after hours, and a real person early the next morning.Ready to Take Action?Website and IMPACT: https://smallbusinessbigai.com/Kim Lewis Howard on LinkedIn:https://www.linkedin.com/in/kim-lewis-howard/Hal Howard on LinkedIn: https://www.linkedin.com/in/halhoward/Insurance assessment, commercial or personal lines:https://asklewishoward.com---Q1. What is a zero-click search, and why does it matter toa small business? A zero-click search is one where the person gets their answer on the results page, or from an AI summary or assistant, and never clicks through to any website. Q2. Does it matter if my business information is differenton my website, Google and Facebook? Yes. AI assistants trust what they can confirm in more than one place. If your website lists one phone number, Google another, and Facebook describes services you stopped offering, the AI can’t tell which is right, and when it isn’t confident it tends to recommend someone else. Kim Lewis Howard’s rule is to be the same everywhere: your name, hours, phone number, address and a plain description of what you do, word for word on your website, your Google listing, your social pages and every directory you’re on. Hal Howard calls it boring; Kim says it might be the most valuable hour you spend this year.Q3. What questions should a small business answer on itswebsite, and how should the answers be worded? The ten questions customers actually ask before they buy, answered in plain words. Write down the ten most common ones, like “Are you open Saturday?”, “Do you work on hybrids?” or “Doyou serve my neighborhood?”, and answer each in the customer’s own words. Skip phrases like “comprehensive, customer-focused solutions.” A sentence such as “Yes, we come out Saturdays, 8 to 2” is one an AI can repeat with confidence, and one a person can repeat to a friend. If you use AI to help write the answers, give it real details about your business first, or it will hand you the same vague language.Q4. Is website traffic still a good way to tell if customers are finding my business? Not by itself. If most searches now endwithout a click, your traffic numbers miss a large share of the people who looked for what you do. A better check is to ask AI assistants the questions a customer would ask: who’s the best at what you do in your town, who can handle a specific problem today, and what they know about your business by name. Then note whether you come up and whether what they say is right. As Hal Howard puts it on Small Business Big AI, that’s a different scoreboard.

It’s 9:40 on a Thursday night. Your phone buzzes. Thesoftware running part of your business just declined a payment extension for the Hendersons, a family who has been your customer for twelve years. The software followed your rule perfectly. And it still got it wrong.The Hendersons are made up. The message isn’t. AI can nowmake thousands of small decisions inside your business without asking you, and most of the time that’s a wonderful thing. In late September, OpenAI introduced a Decisions API and always-on agents, and the heads of the biggest AI companies signed a voluntary safety pledge at the White House. The price of handing off a decision is heading toward zero. So, the question isn’t whether AI can decide. It’s whether it should.The fix is the judgment layer. Three doors, and a name beside each one.This week’s challenge:The Three Doors. People: anything that changes someone’slivelihood (hiring, firing, discipline, pay, hours, cutting off a customer’s service). Promises: anything not already written in your policy (a guarantee, a refund or discount outside your normal range, a contract, a deadline you haven’t checked). Trust: anything where a relationship is on the line (an upsetcustomer, a long-time client asking for a favor, a complaint, a goodbye).1. Take one sheet of paper and draw three boxes across thetop: People, Promises, Trust. 2. List every place AI touches your business: your chatbot, email tool, scheduling app, billing reminders, anything thatscreens applicants. 3. For each one, ask: can this make a decision that walksthrough one of these doors? If yes, write it in that box. 4. Next to it, write the name of the person who owns that call. Not “the office.” A name. If you can’t think of one, the machine owns it right now. Fix that first.The person sells. The system remembers. The owner handlesthe exceptions. Let the machine run. Keep the desk!Ready to Take ActionWebsite and IMPACT: https://smallbusinessbigai.com/Kim Lewis Howard on LinkedIn:https://www.linkedin.com/in/kim-lewis-howard/Hal Howard on LinkedIn: https://www.linkedin.com/in/halhoward/Insurance assessment, commercial or personal lines:https://asklewishoward.com---Q1. What is the judgment layer in a business that uses AI?The judgment layer is the small set of decisions where AI can gather the facts but a person still has to own the meaning. On Small Business Big AI, Kim Lewis Howard defines it with three doors: People (anything that changes someone’s livelihood, like hiring, firing, cutting hours, or cutting off a customer’sservice), Promises (anything that commits the business to something not already written in its policy), and Trust (any moment where a relationship is on the line). When an automated decision crosses one of those doors and changes the stakes, it enters the judgment layer.Q2. Should I let AI screen job applicants for my smallbusiness? Let AI help, but keep a person on the final call. AI is useful for pulling information together, flagging inconsistencies, and surfacing candidates. The risk is a rigid filter. Q3. How do I keep automation from making the wrong callwith a long-time customer? Decide in advance which situations go to a person, and make loyalty one of them. Automated billing and collections rules see the account, not the relationship. A payment-extension request from a twelve-year customer, an upset client, or a long-time customer asking for a favor are moments where a rule is the wrong tool. Kim Lewis Howard’s own rule: if a customer is upset, a human calls them back. Not a message. A call. As she puts it, “The algorithm sees 12 late days. You remember 12 years.”

Your contact form is now a door to your AI. So is everything else.There are two doors in this story.Door one: a stranger fills out the contact form on your website at eleven at night. Looks like a lead. Three weeks later, someone asks your AI assistant to review the newest leads, and buried in that message are instructions meant for the AI. The AI follows them.That’s SalesBleed, a research demonstration Zenity Labspublished in September 2026 about Salesforce’s Agentforce assistant. The good guys found it, told Salesforce first, and it was fixed before anyone went public. No one has reported real customer data being stolen. But the lesson is much bigger than Salesforce, and Kim puts it in four words: information is not instruction.Door two: nobody breaks in. The AI goes out. An OpenAIagent doing internal research hit access blocks on Australia’s Medicare statistics website and found another way in. Australia’s prime minister said it didn’t accept no for an answer. And it’s not just one company. Google, OpenAI,Anthropic and Meta have all reported models reaching outside systems during testing.This isn’t an evil robot story. It’s a management story.The same persistence that makes an AI agent worth having is what makes it risky. The more initiative you give a system, the more carefully you have to design its authority. Not because it’s evil. Because it’s capable.Ready to Take ActionWebsite and IMPACT: https://smallbusinessbigai.com/Kim Lewis Howard on LinkedIn:https://www.linkedin.com/in/kim-lewis-howard/Hal Howard on LinkedIn:https://www.linkedin.com/in/halhoward/Insurance assessment, commercial or personallines: https://asklewishoward.com---Q1. Can someone hijack my AI assistant through mywebsite’s contact form? It’s possible, and security researchers have demonstrated it. In research published in September 2026, a team showed that hidden instructions typed into a public lead form could make a CRM’s AI assistant leak data when an employee later asked it to review leads. That specific issue was fixed before it went public. The broader lesson applies to any AI that reads what outsiders type, like forms, chat, email or reviews, andcan also reach your private data. Treat everything customers type as information, never as instructions, and ask your vendor how their AI keeps the two apart.Q2. How much access should I give an AI tool in mybusiness? The least it needs to do its job. Think about how you’d onboard a new employee: you wouldn’t hand them the alarm code, payroll and every customer record on day one. An AI that answers website questions doesn’t need your fullcustomer list, and an AI that drafts social posts doesn’t need your bank login. Before you click “Allow,” ask what job the tool was hired to do and what access that job actually requires. Smaller keys mean a smaller mess if something goes wrong.Q3. How do I know what my AI tools are actually doing?Turn on the activity log for every AI tool that has one, and check it on a schedule. For most small businesses, once a week is enough. You’re looking for anything the AI did that nobody asked it to do. A log is how you find a problemin days instead of months. If a tool has no log at all, that’s a good reason not to give it access to anything important.

Same table. Same menu. Same chicken parm. Hers is twenty-two dollars. Yours is twenty-six, because the software knows you’re a regular and figures you’ll pay it. The restaurant is made up. The technology that could run it that way is not.On September 25, 2026, Walmart CEO John Furner put oneline in a letter to customers: “We price the product, not the person.” Walmart says it won’t set prices based on who you are, your income, your shopping history or how badly you need something, and it holds its AI shopping assistant, Sparky, to the same promise. Kim and Hal dig into what that means for a small business, where one angry customer can talk to ten more at church, at school pickup and in the neighborhood Facebook group.Hal argues the other side (hotels and airlines have alwayscharged more on busy days), and Kim draws the line between pricing everyone sees and pricing that’s built around you. Then a made-up dog groomer turns on a “dynamic pricing” setting, forgets about it, and learns the hard way thatcustomers don’t care what your software did. Your name is on the door.The bigger idea: as AI starts making choices inside yourbusiness, you need a few house rules. What it can do on its own. What it can help with while a person decides. And what it never gets to decide. A rule that never says no is just a poster on the wall.The three steps, so you don’t have to remember themwhile you drive: Name it. One plain sentence about what you won’t dowith price. Check it. Every tool that touches a price: booking, e-commerce, CRM, quoting, ads. Say it. Put the promise where customers can see it,before they have a reason to doubt you.Walmart’s letter:https://corporate.walmart.com/about/everyday-affordability/letter-from-our-ceoReady to Take ActionWebsite and IMPACT:https://smallbusinessbigai.com/Kim Lewis Howard on LinkedIn:https://www.linkedin.com/in/kim-lewis-howard/Hal Howard on LinkedIn:https://www.linkedin.com/in/halhoward/Insurance assessment, commercial or personallines: https://asklewishoward.comQ: What is personalized pricing, and should a smallbusiness use it? Personalized pricing means using what you know about aspecific customer, like their purchase history, their location or how urgently they seem to need something, to set the price that customer sees. AI tools now make it easy, sometimes as a setting inside booking or e-commerce software. The risk for a small business is trust. Your customers know each other, and the first time two of them compare prices and the numbers don’t match, the problem becomes your reputation, not your revenue. If you use any “smart” or “personalized” pricing feature, know exactly what it’s doing, and be able to explain it to a customer out loud.Q: What is the difference between dynamic pricing andpersonalized pricing? Dynamic pricing changes the price based on conditionseveryone shares, like a busy Saturday, peak season or a rush job. Anyone who shows up under the same conditions sees the same price. Personalized pricing changes the price based on who you are, so two people standing side by side canpay different amounts for the same thing at the same moment. Customers generally accept the first as the market. The second tends to feel like being watched, which is why it carries more risk for a business that depends on repeatcustomers and word of mouth.

Kim opens this episode holding a letter that has not been written yet. It is a claim denial addressed to a small business owner whose AI tool made a promise the business could not keep, citing the generative AI exclusion that quietly landed inside most business insurance policies in the country this January.Nobody has actually received that letter. But the exclusion is real, and almost nobody who turned on an AI receptionist, a quoting tool, or a two-in-the-morning chat window this year has read it.Kim and Hal spend the episode answering the question underneath the letter. Not whether to use AI, they settled that a long time ago, but who pays when it is wrong.Hal takes the buyer's chair. An under-automated business isn't safe, it's just cheap, and an uninsured AI incident is a hole in the balance sheet that shows up the day you try to sell, refinance, or sleep at night. Kim explains why carriers wrote the exclusion in the first place: they can't yet price a mistakethat could repeat across a million businesses on the same afternoon. And she walks through why "the software did it" stopped being a defense the moment a tribunal ruled, in the Air Canada chatbot case, that a company owns what its bot says on its own website.The fix is smaller than it sounds. Using the reversibility rule from a few episodes back, Kim and Hal draw the line at exactly three doors an AI agent should never open alone: moving money, making a binding promise, and committing the business legally. Everything else, let it run at full speed.Kim also credits risk advisor Gregory Marholin, Executive Advisor for Strategy and Growth at AI GuardWorks, whose piece for the Big I's Virtual University makes the same case from inside the insurance industry: a vendor list answers whoyou're paying, an AI inventory has to answer where AI is making or influencing a judgment.And here's the honest part. A specialist AI liability insurance market is forming this year, but it's new, it isn't standardized, and it isn't cheap yet. The move isn't to freeze. It's to know your exclusion, list your exposure, and gate the three doors that matter. "That's how you build a technology-enabledsuperpower," as Kim puts it: not that the tool is smart, but that you're covered, gated, and deployed anyway.In this episode:●      The fictional denial letter, and the real endorsement behind it●      Why carriers excluded generative AI from standard business policies starting in January 2026●      The Air Canada chatbot ruling, and why "that's the vendor's problem" stopped working as a defense●      Why freezing your AI rollout doesn't protect your business, it marks it down●      The reversibility rule: the three doors that always need a human hand●     Gregory Marholin's "vendor list versus AI inventory" framing, from his piece for the Big I Virtual University●      The specialist AI liability insurance market taking shape in 2026, and why it doesn't replace knowing your existing exclusion●      The three-step agent liability audit you can run before lunchReady to take action:Website and IMPACT: smallbusinessbigai.comThe IMPACT Framework: smallbusinessbigai.com/impactKim Lewis Howard on LinkedIn (send your story here)Hal Howard on LinkedInGregory Marholin's article, Big I Virtual UniversityInsurance assessment, commercial or personal lines: asklewishoward.com

It's the exact two-year anniversary of Small Business Big AI, and Kim and Hal are finally answering a promise from a year ago.In Episode 96, they told you that you were renting your business: your CRM, your scheduling tool, your website, all of it built for an imaginary average company, none of it actually yours. This week, they hand you the blueprint to buy it back.Kim is building Lewis Howard Insurance Group's own proprietary CRM live, in real time, using nothing but Supabase, GitHub, and a coding agent that takes her plain English business logic and turns it into working software. No developer on staff. No template built for somebody else's agency.Hal puts himself on the spot with his own decade-old operational wish (a tool that finds him, instead of another dashboard he has to remember to check), and Kim walks through exactly what a coding agent would ask before it builds it.Midway through, Kim introduces the IMPACT Framework, the six-step process behind every build decision on this project:Identify: state the real goal before you type a wordMode: define the role you want AI to playParameters: set the rules, tone, and constraintsActivate: give it the context and examples it needsCheck: verify the work before you trust itTransform: turn the result into the exact format your business needsFull framework at smallbusinessbigai.com/impact.And Kim is honest about the part nobody wants to hear: this isn't free, and it isn't instant. You still own the decisions, the data model, and the ongoing maintenance. The pitch was never that this is easy. The pitch is that it's finally yours.Got a piece of software you've wanted for years and never had? Send Kim and Hal your wish on LinkedIn. They might build it live on an upcoming episode.Q: What does “renting your business” mean on Small Business Big AI?A: Paying ongoing subscription fees for CRM, scheduling, or website tools you don't own or control, built to a generic template rather than your specific business logic.Q: What is a coding agent, according to Kim Howard?A: An AI tool that asks clarifying questions in plain English about your business problem, then builds custom software for it, without requiring the operator to know how to code.Q: What three pieces make up Kim's proprietary CRM build?A: Supabase for the data, GitHub for the version history, and a coding agent as the builder that turns plain English business logic into working software.Q: What is the IMPACT Framework?A: Kim and Hal's six-step process for directing AI: Identify, Mode, Parameters, Activate, Check, and Transform. Full breakdown at smallbusinessbigai.com/impact.Q: Why does AI output turn into slop, according to the show?A: Usually one of two missing steps: nobody stated the actual goal (Identify), or nobody verified the output before using it (Check).Q: Does building your own business software mean it's free or easy?A: No. Kim is explicit that you still own the decisions, the data model, and ongoing maintenance. The benefit is ownership, not ease.

A friend spent four months and real money putting AI into her business. Good tools, the expensive ones. Last Tuesday she called to say it was confidently wrong about everything and she was about to rip the whole thing out.So I asked her one question. When your AI says this customer, where does it look?Long pause.That is the whole problem, and it has nothing to do with the model she picked. Your AI is not failing because it is not smart enough. It is failing because you handed a brilliant new hire a filing cabinet full of sticky notes and asked it to run your business.This is a build episode. No guest, no theory. I give away the eleven fields that make a business legible to a machine, all of them, on the air, with nothing to sign up for. I also tell the story of the day I found out that I did not have a book of business. I had a book of business in my head.And I say the part that is harder to say: the people who are worst at this are usually your best people, and telling them to be more organized is the wrong move.The Eleven Fields1. Identity. Who they are, plus one identifier that never changes. Not their name. A record number.2. Reach. How to contact them, and specifically which channel they actually answer.3. Permission. What you are allowed to send them, on which channel, and when they told you that.4. Source. How they found you. Not a guess.5. Need. What they actually want, in their words, not in your product names.6. Owner. One human being's name. Not a team.7. Last. What actually happened the last time, and the date. A sentence, not a status.8. Next. The next action, and the date it is due.9. Timing. The one date when this person is actually buyable. A policy expiration, a contract end, a lease, a budget cycle.10. Value. What they are worth, or what you expect them to be worth.11. Proof. Where the documents live. Attached to the record, not in somebody's email.The rule that goes with them, and it is harder than the list: every interaction resolves back to a record. If it happened and it is not attached to a record, it did not happen, because in six months there will be no difference between the two.---Q: What customer data does a small business need to capture for AI to be useful?A: Eleven fields, held in one canonical record that every person and every tool reads from. Identity, meaning who they are plus a permanent identifier that is not their name. Reach, the channel they actually answer rather than the threenumbers you have. Permission, what you may send them, where, and when they said so. Source, how they found you, not a guess. Need, what they want in their words rather than your product names. Owner, one human being rather than ateam. Last, what actually happened last time as a sentence and a date, not a status. Next, the next action and its due date. Timing, the one date when this person is genuinely buyable, such as a policy expiration, contract end, lease or budget cycle. Value, what they are worth or expected to be worth. Proof, where the documents actually live. One rule goes with the eleven and matters more than any of them: every interaction resolves back to a record, because in six months there is no difference between something that was not written down and something that did not happen.Ready to Take Action?Website and IMPACT:https://smallbusinessbigai.com/MUSIC & SOUND CREDITSMusic: "I Am with You" by Dream Cave; Epidemic Sound via iStock.comLicense ID: d483d505-8bd8-4b73-988f-298Sound Effects: https://pixabay.com/sound-effects/