About one in five US businesses: the Census Bureau counted 19.8% using AI in any business function as of May 3, 2026, and fewer than 20% of firms with four or fewer employees. Bigger firms use it more (37% of those with 250+ staff), and use among firms under 20 employees did not change significantly from December 2025 to May 2026. Higher numbers you see measure something else: the U.S. Chamber's 2025 survey says 58% of small businesses self-identify as using generative AI. Always check what a figure counts before you trust it.

Source: census.gov

Free guides on this: Check AI adoption statistics

Mostly for everyday office work: drafting emails, writing marketing content, summarizing documents and managing schedules. A 2026 U.S. Chamber review of small-business adoption found these simple admin, scheduling and reporting uses give the quickest value, while firms with field operations also use AI for staff scheduling, jobsite safety monitoring and predicting equipment maintenance. The SBA lists similar uses, plus customer-service chatbots and spotting trends in your own business data.

Source: uschamber.com

Free guides on this: See AI in business examples

A chatbot answers you one question at a time; an agent takes a goal and carries out several steps on its own, adjusting as it goes. Zapier's example: after a new lead comes in, an agent might research it, draft an outreach email, schedule a follow-up and log it in your CRM. Plain automation is a third thing: a fixed "when this happens, do that" rule with no judgment. More steps without you watching means more need to check its work.

Source: zapier.com

Free guides on this: Tell chatbots, automation and agents apart

Probably steady change, not overnight upheaval. Princeton researchers Arvind Narayanan and Sayash Kapoor argue AI's big economic effects will play out over decades, as electricity's did, and Census data shows use among firms under 20 employees did not change significantly from December 2025 to May 2026. At the same time, Ethan Mollick argues today's tools can already do much more than most people use them for. So the practical edge is learning to use what exists now, not waiting for what comes next.

Source: knightcolumbia.org

Free guides on this: See the future of AI in business

AI is software that can work through tasks and problems in a way that imitates human thinking, and gets better as it receives new information. The SBA's examples for small businesses: drafting marketing copy and social posts, sorting email, scheduling, summarizing meetings, brainstorming, running customer-service chatbots and spotting trends in your own data.

Source: legacy.sba.gov

Free guides on this: Learn what AI does for business

Start with one task you repeat at least three times a week that needs no judgment call, and first check whether a tool you already pay for has an AI feature for it. Time that task for a week, then run the AI next to your normal process for two weeks, and keep a person checking everything before it reaches a customer. If it beats your old way, write a one-paragraph policy (approved tools, what data they may see, who signs off) before you add more. Find your first AI tasks helps you pick the task.

Source: too-many-hats.pages.dev

Free guides on this: Plan your first 30 days

No: if you can write an email or talk on the phone, you can use an AI chatbot. Wharton professor Ethan Mollick advises talking to it conversationally, giving it context about you and your task, and not worrying about saying exactly the right thing; you can even start with voice mode. His suggestion for a first hour is to try it on real work from your own job. What takes practice is judging the answers, not the technology.

Source: oneusefulthing.org

Free guides on this: Check your AI readiness

The common ones are buying a tool before fixing the process it will run, rolling it out to everyone at once instead of testing one task, and having nobody track what it saves and costs. Others: vague one-line prompts, ignoring that staff already use AI without a policy, and using AI to dodge decisions the owner should make. The fix for most is the same: one task, one owner, measured before and after (Measure before and after).

Source: lilachbullock.com

Free guides on this: Know when not to use AI

Start with a task that costs you the most time or money and that you repeat in roughly the same way every week. HubSpot co-founder Dharmesh Shah suggests listing tasks you repeat, tasks you dread, things you wish you had time for, and things you're doing for the first time, then aiming AI at the costliest. Good first picks happen at least three times a week and need no judgment call, such as drafting routine emails or summarizing notes. Leave pricing and client decisions with you (Know when not to use AI).

Source: simple.ai

Free guides on this: Find your first AI tasks

Not without checking: AI can sound sure and still be wrong, and even top models are uneven and least reliable at judging their own work, as Ethan Mollick wrote in November 2026. Trust it more where you can check the result yourself: count it (word count, budget limits), look it up against a source you name, or try it out. Dharmesh Shah's rule: the less a check depends on the AI's own judgment, the more you can trust it. Never use output you can't verify.

Source: simple.ai

Free guides on this: Know when not to use AI

Keep decisions and sensitive data with you. Don't let AI set prices, send client messages without your approval, make legal or compliance calls, or decide your positioning and brand voice. Don't paste in personal information, health or financial records, customer data, trade secrets or source code to public AI tools. And don't use any AI output you can't check yourself.

Source: diymarketers.com

Free guides on this: Know when not to use AI

Start with the AI already in the software you pay for: if you run Microsoft 365 or Google Workspace, that means Copilot or Gemini, and many owners barely use them. Next, check what your other vendors (QuickBooks, Salesforce, Paychex) have added. Only then pick a general assistant such as ChatGPT or Claude, and connect it to your data under clear rules on which apps and data are allowed. Choose by one process you want to improve, measure how long it takes now, and check whether the tool actually makes it faster. See Compare ChatGPT, Claude, Gemini, Copilot.

Source: genemarks.com

Free guides on this: Pick the best AI for small business

For most real work, pick ChatGPT or Claude: Wharton professor Ethan Mollick calls them the only two choices for people who want the most out of AI right now. He does not suggest Gemini as your main assistant today, though parts of it are strong for research and video, and he rates Microsoft Copilot fine for Office documents but behind on agent-style tasks. For low-stakes jobs any of them will do, so pick the one you like; for important work, use a paid plan and its most advanced model. If your team lives in Microsoft 365 or Google Workspace, the built-in one may still be the easiest start.

Source: oneusefulthing.org

Free guides on this: Compare ChatGPT, Claude, Gemini, Copilot

Yes: the main assistants now offer ready-made connectors for apps such as QuickBooks, PayPal, Docusign, Gmail, Outlook and Dropbox, so you can ask questions of your own business data in plain English. Most of these run on a shared standard called MCP; if your software has no connector, a developer can build one. Claude offers its connectors on every plan, including the free one. Connecting gives the AI permission to read and sometimes change data in that app, so connect only the services you trust and need.

Source: genemarks.com

Free guides on this: Connect AI to your software

Usually not: the typical US business that pays for AI spends a median of $16.32 per employee per month (Ramp data, October 2026). A business seat on Claude costs $25 per user per month billed monthly, or $20 billed annually, and the personal Pro plan is $20 a month (as of Nov 2026). The real costs to watch are add-ons, usage-based agent fees and custom builds; if you build something custom, budget for upkeep every year, not just the setup. See Choose free or paid AI plans.

Source: ramp.com

Free guides on this: Budget your AI costs

Only for one specific job your team repeats, such as answering FAQs or drafting proposals; a general business helper ends up too broad to be useful. Be careful with ChatGPT's custom GPTs right now: OpenAI is retiring them in favor of plugins, with December 11, 2026 announced for Enterprise workspaces and other plans possibly following. For a simple team assistant, a Project in ChatGPT or Claude (or a Gemini Gem) does the same thing: saved instructions plus your reference files. Keep your instructions and files in a dated folder so you can rebuild the assistant in another tool.

Source: pcworld.com

Free is fine for trying AI and for low-stakes tasks, but regular business use needs a paid plan, which starts around $20 a month. Free ChatGPT is ad-supported and limited to a handful of messages every few hours with its best models. Paid business plans also change how your data is handled: Anthropic, for example, may train on Free and Pro chats unless you opt out, but not on Team or Enterprise accounts. If you put customer or company information into AI, use a business plan.

Source: oneusefulthing.org

Free guides on this: Choose free or paid AI plans

First check the AI in the project tool you already use, because most now have it, often only on paid plans. If you are choosing new: Asana is strong on AI recommendations (Standard from $13.49 per user per month), ClickUp lets you ask questions about your projects (full AI on its $25 Business plan), Wrike predicts project risks (from $10), and Hive can plan a project from a prompt, with most AI on its free plan. Those prices come from Zapier's October 2026 review, not from the vendors' own pages. Try any of them on a real project before you commit.

Source: zapier.com

Free guides on this: Pick the best AI for small business

Check three things: what it does with your data, whether it really does what it claims, and whether it pays off for a job you have. Look for a setting to stop training on your chats: ChatGPT, Claude and Copilot offer a simple toggle, while some others make you email support. Be wary of big claims: Gartner estimates only about 130 of the thousands of vendors selling AI agents are the real thing. Test a free version or trial on your own work, and don't upload sensitive or personal data until you know how it is stored.

Source: uschamber.com

Free guides on this: Pick the best AI for small business

They work well for routine calls you'd otherwise miss, and some callers do hang up on them. In one 90-day test at a small heating firm, 43 of 61 calls that would have gone unanswered turned into bookings; 6 callers hung up once they noticed it was an AI, and saying up front that it was an AI cut the hang-ups. Complaints, refunds and emergencies went badly and need a person. Many customers are wary too: 64% of 5,728 people in a Gartner survey said they'd rather companies didn't use AI for customer service, and their top fear was not being able to reach a human.

Source: lilachbullock.com

Free guides on this: Try an AI receptionist

If you miss calls because you're on a job, an AI receptionist is the cheapest fix; a human answering service costs more but handles upset or urgent callers better. Typical AI receptionist plans for a small business run about $50 to $150 a month, while a live service such as Smith.ai starts at $300 a month for 30 calls. Answering yourself is still best for complaints, emergencies and big first conversations, so many owners route those calls to their own phone and let the AI take routine bookings. Test any tool with real calls before you switch it on.

Source: getvoip.com

Free guides on this: Try an AI receptionist

It helps customers get quick answers to repeat questions at any hour, but we found no independent proof that it lifts sales for small businesses. Entry costs are low: Tidio gives 50 free AI conversations, then its AI add-on starts at $32.50 a month, and it answers from your FAQ and website. The risk is that you answer for what the bot says: a Canadian tribunal made Air Canada honor a refund its chatbot invented. Keep its sources current and always offer a way to reach a person.

Source: manatt.com

Free guides on this: Add an AI chatbot

Yes: Gmail and Outlook can both draft replies for you on paid plans, and you edit and send. Speed matters: in BrightLocal's 2026 survey of 1,002 US consumers, 89% expect owners to reply to reviews and 32% expect a reply by the next day. But 50% are put off by generic, templated replies, so let AI draft and make each reply specific to the customer. Read every draft before it goes out; Outlook itself warns the text may be inaccurate.

Source: brightlocal.com

Free guides on this: Answer emails and reviews with AI

Yes: most AI receptionists learn from your website, your Google Business Profile or details you type in, and you can add answers over time. Several tools build a first version just from your web address, and Goodcall tracks the questions it had to pass on so you can fill the gaps. It answers only from what you give it, so an old price or service will be quoted with confidence. Check its information every few months, and whenever your prices, hours or services change.

Source: getvoip.com

Free guides on this: Try an AI receptionist

Mostly in three places: it makes things up, it blocks people from reaching a human, and it handles upset customers badly. Air Canada had to honor a refund policy its chatbot invented, and New York City's business chatbot told owners they could take a cut of workers' tips before it was shut down in 2026. In one small-firm phone test, the AI kept steering complaints back to booking and cut off slow talkers. Give it current answers only, let it hand off early, and keep complaints for people.

Source: themarkup.org

Free guides on this: Add an AI chatbot

Limit what it may talk about, give it only approved answers, and send health or legal questions to a person. This matters because you are responsible for what your chatbot says, as Air Canada learned when it had to honor a policy its bot made up. The FTC treats your website and online content as advertising, and health claims there generally need solid scientific evidence such as clinical trials. Set those limits with your vendor, test the bot with tricky questions, and recheck its answers regularly.

Source: ftc.gov

Free guides on this: Add an AI chatbot

Yes, most can, but some do it better than others, so ask before you buy. Some tools let your staff see what the caller said before they pick up, and some charge extra for transfers. Set clear rules: in one small-firm test, calls that mentioned "leak," "no heat" or "emergency" went straight to the on-call engineer's mobile. A handoff that makes callers repeat everything is worse than no AI at all.

Source: getvoip.com

Free guides on this: Try an AI receptionist

Yes: AI can turn your emails, notes and website into a first set of FAQ answers, and chat tools can then answer from them. Tidio's free plan, for example, can import your FAQ and scrape your website, and its AI answers from that content. Write down the questions customers really ask, let AI draft the answers, and check every answer against your actual policy. The bot is only as right as the pages it reads, so update them when anything changes.

Source: tidio.com

Free guides on this: Add an AI chatbot

Yes, and you should: set it up so AI drafts and a person approves, at least until it has proven itself. Dharmesh Shah suggests writing simple pass/fail checks before the task, such as a word limit or a claim that must have a source, and having the AI mark each one. Then confirm the most important items yourself, because checks catch only what you wrote down. Many tools build in a draft step: AI replies in Gmail and Outlook wait for you to edit and send.

Source: simple.ai

Free guides on this: Keep a human in the loop

AI mostly saves you time: drafting emails, social posts, web pages and ads, and helping you research who your customers are. In Constant Contact's 2026 survey of 3,340 small businesses, half of AI users named saving time as the top benefit, mainly on content and data analysis. Christopher Penn's walk-through of a one-person business got a whole marketing plan done on a $20-a-month AI plan, starting from the owner's own words, not generic prompts. Feed it your real material and check what it writes; see Write in your brand voice.

Source: almosttimely.substack.com

Free guides on this: Use AI marketing tools

For most small businesses, Wix is the safe pick: Zapier's testers chose it for most people, it has a free plan with Wix branding, and paid plans start at $17 a month. Framer suits you if design matters a lot, and 10Web if you want WordPress. Whichever you pick, the AI only gives you a first draft: the tester says you can't launch a real site without customizing it, and AI-made stock text and pictures don't belong on a business site. Replace them with your own words, photos and prices.

Source: zapier.com

Yes, it can draft all of them, and saving time on content is the main reason small businesses use it, but you still decide what goes out. Christopher Penn's advice for ads: first find the weakest step in your marketing, describe your ideal customers, then have AI write many versions, and check them yourself before any money is spent. Feed it your own notes and past posts so the result sounds like you; see Write in your brand voice.

Source: almosttimely.substack.com

Free guides on this: Use AI marketing tools

Yes, if you give it simple rules: AI can send an instant personal reply to every form, score each lead and draft the first follow-up for you. One guide for owners who sell scores each lead on fit and interest from 0 to 3 each, calls the top scores the same day and puts the rest into a short email sequence. Clean up your contact list (CRM) first, and set stop rules so a lead who replies or books a call stops getting automatic emails. Keep big deals and touchy objections for yourself.

Source: too-many-hats.pages.dev

Free guides on this: Speed up sales follow-up

Mostly by doing good local SEO: a clear website and a complete Google Business Profile. In BrightLocal's study of about 200,000 local AI searches, business websites made up 42% of what ChatGPT and Google's AI cited, and Google Business Profiles 28.5%. The lists also change a lot: repeat runs of the same question overlapped only 20-33%, so nobody can promise you a "ranking" in ChatGPT. Make sure your site states your services, area and contact details plainly.

Source: brightlocal.com

Free guides on this: Get recommended by ChatGPT

Yes for research and drafts, but be careful with what you post on your Google Business Profile. Local search expert Darren Shaw suggests asking ChatGPT or Gemini which directories matter in your industry. Whitespark's expert survey rates AI-generated text on a Business Profile as harmful and a suspension risk, and AI-made photos as worse. Use AI to plan and draft, then write the final profile text in your own words.

Source: whitespark.ca

Free guides on this: Use AI marketing tools

Yes, AI is a good research assistant for a first draft of your ideal customers, your message and your competitors, as long as you check it against what real customers say. Christopher Penn had AI's deep-research tools draft up to five ideal-customer profiles for a small business and then built every later step around what was likely to sell. He started from an hour of the owner talking about her business, so the output wasn't generic. Treat the result as a hypothesis and test it in a few real customer conversations.

Source: almosttimely.substack.com

AI can draft a branded proposal quickly, but we found no independent proof that AI-written proposals win more often. The gain is speed: you send sooner and follow up on time, for example with an AI-drafted nudge after 72 hours of silence. For high-value deals, one guide advises letting AI only score and remind, and writing every message to the buyer yourself. Always check prices, scope and promises in the draft, because you are bound by what you send.

Source: too-many-hats.pages.dev

Free guides on this: Speed up sales follow-up

Show it your own writing and your own edits, not vague words like "professional" or "friendly." Christopher Penn gives AI 10 to 15 of his own pieces and breaks style into concrete traits such as word choice, sentence length and rhythm. Allie Miller compares about five AI drafts with the versions she actually sent, has AI list her changes, and turns them into a short yes/no checklist the AI checks every new draft against. Start from your own notes or a voice memo, and let AI polish rather than invent.

Source: aiwithallie.com

Free guides on this: Write in your brand voice

Yes, it can give a useful second opinion on wording, offers and pricing pages, but it can't certify that your site works for everyone. For accessibility, the W3C, the body that sets web standards, says automated tools can only assist: they can't check everything, sometimes give false results, and human judgment is required. Ask AI for specific fixes, try them on a few real visitors or customers, and keep what helps.

Source: w3.org

Free guides on this: Use AI marketing tools

For Google's AI answers, no. Google says you don't need new machine-readable files, AI text files or special markup to appear in AI Overviews or AI Mode; a normal page that Google can index is enough. We found no evidence that it helps with other AI tools either, so spend the time on a clear website and a complete Google Business Profile instead.

Source: developers.google.com

Free guides on this: Get recommended by ChatGPT

Yes: record and transcribe your calls, and AI can pull out the objections, questions and phrases that keep coming up. One guide for owner-sellers calls the simplest version the most useful automation a small operator can build: AI drafts each follow-up in the buyer's own words from the call transcript. Ask before you record. Federal law needs only one person's consent, but states such as California, Florida and Pennsylvania require everyone on the call to agree.

Source: too-many-hats.pages.dev

Free guides on this: Speed up sales follow-up

Partly: AI can sort transactions, help reconcile accounts and answer questions about your numbers, but it still gets the harder accounting wrong often enough that a person must check it. QuickBooks now has built-in AI (Intuit Intelligence) that answers questions about your own business data, plus AI agents for accounting, payments and sales tax. In a 2026 test of 101 accounting tasks, the best general chatbot got only 77% right; one model classified transactions 92% correctly but managed just 30–40% on journal entries. Let AI do the sorting and summaries, and have you or your bookkeeper approve the entries.

Source: accountingtoday.com

Free guides on this: Use AI bookkeeping

Pick the task before the tool: for most small teams the setup is one no-code automation platform (Zapier is the best-known) linked to the apps you already use. Start with the single task that eats the most time and follows the same pattern every time, build it first with no AI steps and test it, then add one AI step and check it before adding more. Keep a person approving anything customer-facing or high-stakes, and try AI agents only once a few plain automations run reliably. See Use AI agents.

Source: zapier.com

Free guides on this: Automate workflows

Today, agents are worth it for routine, low-stakes jobs where you can easily check the result, not for running parts of your business on their own. In a Carnegie Mellon test of office tasks, the best agent fully finished only about 30% of them, and Gartner expects over 40% of agent projects to be cancelled by the end of 2027. Ethan Mollick's rule is a good one: an agent should never decide on its own to spend money, contact outsiders or open sensitive files. Give it a clear definition of done, limit what it can reach, and review what it did.

Source: oneusefulthing.org

Free guides on this: Use AI agents

Not if a knowledgeable person reviews what it does; the risk comes from trusting it blindly. A Stanford study of 277 accountants found those using AI closed the monthly books 7.5 days faster, yet 62% worried about AI errors, and less experienced staff tended to accept AI output even when it was flagged as uncertain. Let AI categorize, match and draft, and keep the approval of entries, tax and payroll with you or your bookkeeper. See Keep a human in the loop.

Source: gsb.stanford.edu

Free guides on this: Use AI bookkeeping

Use it to take chores off your people's plates (drafting emails, summarizing threads, filling forms, routing approvals), not to replace their jobs. US workers who use generative AI save about 5.4% of their work hours, roughly 2.2 hours in a 40-hour week, according to a St. Louis Fed survey. Small CPA firms show what that looks like: one solo accountant set up an automation that turns emailed vendor invoices into approved QuickBooks bills, using tools she already had, and another firm's smarter intake form lifted conversions by about 25%.

Source: stlouisfed.org

Free guides on this: Cut admin with AI

Yes, as a drafting partner: project tools such as Hive can draft a project plan from a prompt, ClickUp answers questions about your projects, and Wrike predicts project risks. AI is good at turning your notes into a clear scope, task list and options for the client. The numbers are yours: base hours and prices on your real costs and past jobs, and check every figure, timeline and promise before the client sees it.

Source: zapier.com

Yes, accounting software now does this with AI: QuickBooks spots overdue invoices and follows up with customers automatically, and Intuit says this gets businesses paid four days faster on average. QuickBooks also has a Payments Agent and a free AI invoice generator. Before you switch it on, read the reminder wording, set when it may contact a customer, and decide which clients you would rather chase personally.

Source: genemarks.com

Free guides on this: Use AI bookkeeping

Yes to both, within limits. An AI phone receptionist tested for 90 days at a small heating firm turned 43 of 61 previously missed calls into bookings on the spot, but struggled with complaints and emergencies until those calls were routed to a person. Automated reminders help with no-shows: in a Penn Medicine trial with almost 60,000 high-risk patients, adding an automated call to text reminders cut no-shows from 11.3% to 9.6%. Say plainly that callers are talking to an AI; in the heating-firm test that reduced hang-ups. See Try an AI receptionist.

Source: lilachbullock.com

Free guides on this: Cut admin with AI

Only some of it, and only in the right plan with the right settings. A 2025 Stanford study found all six big US chatbot makers, including OpenAI, Google, Anthropic and Microsoft, use chats to train their models by default, and some keep chat data indefinitely, so turn training off where you can. The National Cybersecurity Alliance says to keep personal information, health and financial records, customer data, trade secrets and source code out of public AI tools, and notes business plans usually come with stronger data protections than free ones.

Source: staysafeonline.org

Free guides on this: Protect customer data

Usually you are: using AI does not move your responsibility for what you publish, promise or decide onto the tool. North Carolina REALTORS' legal guidance (January 2026) puts it plainly for brokers: they "remain fully responsible" for AI-written content. In hiring, a federal court let the Mobley v. Workday case go ahead against the software vendor, and lawyers say both vendors and employers may face liability. The FTC also requires "competent and reliable evidence" for accuracy claims about AI products (Workado order, August 2025). This is general information, not legal advice.

Source: sgrlaw.com

AI can draft job descriptions you then review, but screening candidates with AI carries legal risk. Federal anti-discrimination laws (Title VII, the ADA) still apply to AI tools even though the EEOC withdrew its AI guidance in 2025, so a tool that screens out a protected group can create liability even if unintended. Illinois (HB 3773, in effect since January 1, 2026) requires notice when AI is used in employment decisions, for employers with even one Illinois employee; New York City (Local Law 144, enforced since July 5, 2023) requires a bias audit within a year before use and candidate notice 10 business days ahead. Check your state's rules too (Check state AI laws).

Source: burkelaw.com

Free guides on this: Use AI in hiring legally

AI can give you a first draft, but don't sign or publish it without a lawyer's review; it has to fit how your business actually works. North Carolina REALTORS' legal guidance (January 2026) says AI "should not be relied upon to draft contracts." For privacy policies the FTC's rule is that what you say must match what you do: in March 2026 it charged OkCupid under Section 5 of the FTC Act for sharing user data in ways its own policy didn't allow. This is general information, not legal advice.

Source: koleyjessen.com

Run a split test: show the AI-suggested version to half your visitors or customers and your current version to the other half, changing only one thing at a time. Shopify's guide says to run it at least two full business cycles (about two to four weeks) and not to stop early; with low traffic (it suggests 5,000+ visitors to a page) a test won't tell you much. Track conversions and revenue per visitor, plus returns and support volume. A simple before-and-after can mislead, because seasons and promotions change sales too.

Source: shopify.com

Free guides on this: Measure before and after

It can be, but the proven gains are modest, so measure your own. A St. Louis Fed survey found generative AI users saved about 5.4% of work hours (about 2.2 hours a week for a 40-hour worker), and a 2026 survey of 300 online store owners found AI adopters had no financial advantage over non-adopters. A simple test: hours saved times your hourly rate, minus the subscription cost; consultant Lilach Bullock cancels any tool still at zero or negative after two months.

Source: lilachbullock.com

Free guides on this: Make the AI business case

Expect a couple of hours a week at first, not days: a St. Louis Fed survey found generative AI users saved about 5.4% of their work hours, about 2.2 hours a week for a 40-hour worker. Don't trust your gut on it: in a 2025 METR trial, experienced software developers believed AI made them 20% faster when it actually made them 19% slower. Time a few tasks for two weeks before and after, and count the minutes you spend checking and fixing AI output.

Source: sagnikbhattacharya.com

Free guides on this: Measure before and after

Start with an operational problem, and for most independent practices that is billing: MGMA's IT chief suggests asking a HIPAA-compliant AI tool to find the common root causes of your claim denials. Then check the AI already built into your EHR or office software before buying anything new, and test one tool for about 60 days with a few eager staff. Never put patient information into a tool without a signed Business Associate Agreement: OpenAI will not sign one for ChatGPT Free, Plus, Team or Enterprise, only for its ChatGPT for Healthcare offering.

Source: medicaleconomics.com

Free guides on this: AI for healthcare practices

Yes, and if you sell on Shopify it is built in: its Sidekick assistant drafts product descriptions from the product editor, and the file editor can swap a photo's background for a solid color or change an image from a text prompt. Shopify says many of these AI features come at no extra cost, though what you get depends on your plan. You can also connect your store to ChatGPT or Claude to work on products and orders from there. Check every description for invented features before it goes live, since you are responsible for what the page promises.

Source: help.shopify.com

Free guides on this: AI for online stores

It can draft them, but it should not set your prices. Use it to turn your job notes into clear basic, recommended and premium options, and to write follow-ups to customers who have not answered a quote. Your prices must come from your real labor, materials, overhead, local market and margin, and you check every number before it goes out. Keep customer names, addresses and payment details out of public AI tools.

Source: housecallpro.com

Free guides on this: AI for home services

Mostly for the office side, and many have not started yet: in a 2026 ACHR News survey, half of contractors did not use AI at all. Those who do use it most for follow-up messages (30%), scheduling and customer service (23%) and estimates or pricing (20%); only 11% used it to diagnose problems before a visit. In ServiceTitan's survey of 1,017 contractors, 66% of AI users save at least three hours a week, yet only 11% have high trust in their AI tools. A common first step is letting AI answer after-hours calls. See Try an AI receptionist.

Source: achrnews.com

Free guides on this: AI for home services

Yes, if you paste it into a consumer AI tool that can keep it or train on it. Free and personal plans may use your chats for training unless you switch that off, while business plans such as Claude Team and Enterprise are excluded. The ABA tells lawyers to judge that risk before using a tool, and in some cases to get the client's informed consent first. Small CPA firms keep client data out of public AI tools and use only enterprise versions.

Source: lawnext.com

Free guides on this: AI for professional services

It is best at the writing: among REALTORS who use AI, 75% use it for listing descriptions, and many also use it for social media posts and follow-up emails, according to NAR's 2026 technology survey. 23% of REALTORS use AI daily, and 55% say it has helped their business. Check every property detail against the MLS, because AI can invent amenities, and keep financial records, settlement statements and access codes out of unvetted tools.

Source: nar.realtor

Free guides on this: AI for real estate

Most small firms already use it, but few have rules for it: Clio's 2026 survey found 71% of solo lawyers and 75% of small firms use AI, yet more than half have no AI policy. Lawyers use it to draft first versions, summarize case files and condense long email threads. Small CPA firms use it for smarter intake forms, invoice-approval automations and turning tax research into client summaries, with a CPA reviewing every output. Before you scale up, write a short policy on approved tools and client data.

Source: ncbar.org

Free guides on this: AI for professional services

Take it seriously, then show where your on-site judgment changes the answer: ask what they asked the chatbot, confirm what it got right, correct what it got wrong, and explain what it couldn't see. The problem is confident misinformation, and what beats it is first-hand proof, such as reviews and video testimonials from local customers. Homeowners are ready for it: in a 2026 survey, 37% said they would look up contractor reviews through AI tools like ChatGPT, and 34% are comfortable with AI diagnosing problems before a visit.

Source: achrnews.com

Free guides on this: AI for home services

Never let it describe who the home suits instead of the home itself: phrases like "perfect for a young professional couple" or "great neighborhood for families with children", or any mention of the area's racial, ethnic or religious makeup, can break fair housing rules. Don't let it invent or overstate features, write testimonials or reviews, or produce a price estimate you haven't checked against the comps yourself. Label AI-altered or virtually staged photos and keep the originals; California has required this since January 1, 2026. Check every fact against the MLS and make sure your brokerage name and license number still appear.

Source: nar.realtor

Free guides on this: AI for real estate

Try it yourself first, and if you hire, hire for a short job while someone on your team learns the skill. Paul Roetzer of SmarterX calls outside help for small teams "a short term fix" and suggests training a tech-comfortable staff member instead; that doesn't require hiring engineers. Before paying anyone, your local Small Business Development Center offers free business consulting. If you do hire, see Choose an AI consultant.

Source: podcast.smarterx.ai

Free guides on this: Decide DIY or hire

Follow one or two trusted sources on a fixed schedule and ignore the rest, for example one weekly podcast such as The Artificial Intelligence Show, which comes out every Tuesday. You don't need every new release: Ethan Mollick argues that today's AI already does far more than most people use it for, so getting better at the tool you have pays off more than chasing new ones. Spend the time you save trying things on your own work.

Source: oneusefulthing.org

Free guides on this: Follow AI for business experts

So far the evidence says no: it changes their tasks more than it removes their jobs. A large Danish study linking AI use to official labor records found no measurable effect on earnings or hours worked two years after ChatGPT's launch, ruling out effects larger than 2%; employers mostly reorganized tasks. What matters more is backing your team: Gallup found 78% of employees whose manager actively supports AI use it frequently, against 44% of others.

Source: nber.org

Free guides on this: Train your team on AI