Updated every week · last updated 26 October 2026

AI use cases for small business

What AI actually did for real firms.

Real AI results from firms of 1 to 200 people. Each entry names the firm, the job and the outcome, links to its source, and gives you a prompt to try on your own work, plus a note on what to check.

151
use cases
7
business functions
59
not vendor-written
56
AI agent entries

Showing 151 of 151

Legal and complianceVendor story

Law Offices of J.Chrisp

Personal injury law firm, Northern California

Paralegals upload collision reports and medical records and Supio builds the medical chronology, works out the recoverable medical specials and drafts the demand letter. Its assistant also explains injuries and lists defences the other side might raise.

Result Building the medical chronology, economics tab and demand letter for a case, which could take 80 hours by hand, now takes moments.

Not statedUS

Customer servicePressAI agent

Echo Janitorial Services

Commercial cleaning company, Abbotsford, British Columbia

An AI receptionist in the Jobber job management platform answers the company phone, fielding job inquiries and vendor pitches and passing on only the calls the owner needs to take. Inquiries from the website intake form flow straight into the same platform.

Result The AI receptionist costs $99 a month against roughly $4,000 a month in wages and payroll taxes for a person doing the same job.

Not statedElsewhere

Sales and marketingPress

MCG Homes

Residential remodeller, Fernandina Beach, Florida

The owner built shared Claude projects for estimating, sales qualification, HR and procedure searches. The ballpark estimating tool takes a salesperson description of the scope and property, compares it with past jobs and market data and drafts a budget in the company own cost codes, asking for missing information rather than guessing.

Result Compiling a ballpark remodel budget fell from eight hours to about 30 minutes.

32 staffUS

FinanceVendor storyAI agent

Felix Bloch Erben

Theatre and film rights publisher, Berlin

About 200 royalty statements arrive each week from theatres and other licensees, on paper and digitally and each in a different layout. AI agents built on Make by the publisher office systems supplier, Horn & Görwitz, pull out the data, match the line items and pass them to the publisher own billing system.

Result Processing the weekly royalty statements fell from about two days of one person time to two hours.

Not statedEU

Admin and back officePress

Eiduk Tax & Wealth

CPA and financial planning practice for healthcare professionals, Palatine, Illinois

The founder moved client bookkeeping from QuickBooks to an AI based accounting platform built around its own ledger. Once the system learns a client transaction patterns it reconciles accounts automatically, and he mostly reviews its work instead of processing every entry.

Result Onboarding a new bookkeeping client fell from two to four hours to about one hour.

1 staffUS

HR and peopleSelf-reported

Mosby Building Arts

Design and build remodeller, St. Louis, Missouri

The president uses AI as an assistant to write his monthly 300 word staff newsletter article. He also built a custom Claude project, the Mosby Business Advisor, by answering guided prompts on values, goals, target market and the management team and uploading the employee manual, financials and employee reviews. Each of the 12 managers uses it for advice grounded in the business. A separate set-up checks scopes of work from the 13 salespeople for grammar problems and conflicting information before they go to clients.

Result The monthly staff newsletter article fell from 45 to 60 minutes to five minutes.

135 staffUS

FinanceVendor story

BKB Accountants

Accountancy practice, Poole

Clients send receipts and invoices straight from the Dext app, and Dext AI Assist automates the repetitive bookkeeping steps and suggests corrections, which the team reviews and approves before anything reaches the ledger. The junior bookkeeper also uses it to query transactions instead of searching email chains and spreadsheets.

Result Bookkeeping heavy jobs take around 25% less time, on the practice own estimate.

About 23 staffUK

Sales and marketingCase study

Bracewell Stirling Consulting

Architecture and masterplanning practice, Stirling

Through the Scottish Government funded National AI Adoption Programme, The Data Lab built a prototype tender analysis tool for the practice. Staff upload tender documents and the tool pulls out requirements, deadlines and scoring criteria, flags risks, scores and ranks opportunities, and drafts bid responses from the firm own case studies, with staff reviewing and adjusting the results.

Result Each user saves around 2 hours a week on tender review and preparation.

About 29 staffUK

Legal and complianceVendor storyAI agent

Phillips Law Group

Personal injury law firm, Phoenix

Instead of referring out small personal injury cases, the firm kept them and runs the processing through Supio. Supio Agent rates every incoming injury case, drafts the plaintiff fact sheet and complaint, and runs a scheduled morning brief for the managing partner, while attorneys spend their time talking to clients.

Result A portfolio of 1,400 small cases that used to be referred out now runs on three people, earning 2.7 times the fee a referral paid.

About 105 staffUS

Admin and back officeVendor story

Harley Street Hearing

Audiology clinic, London

Heidi records consultations, including tinnitus therapy and balance assessments of up to three hours, and drafts notes, referral letters and reports in the clinic own templates, which clinicians review. It is linked to the patient management system and can write up sessions held in Portuguese or French in English.

Result Weekly admin and reporting time fell from two or three hours to about half an hour.

Not statedUK

OperationsVendor storyAI agent

Topa Bathroom Products

Bathroom products importer and wholesaler, Netherlands

An AI agent built with Lleverage watches the order inbox, reads orders that arrive as emails, PDFs, CSVs and spreadsheets, and creates the sales order in Microsoft Dynamics 365 Business Central. It drafts a reply when a commission number is missing and sends unknown article codes to a person.

Result Over 90% of incoming orders now go into Business Central with no manual input.

Not statedNetherlands

FinancePress

Element Design Build

Design and build remodeler, Hudson, Ohio

The owner uses Claude as the link between the firm software, through MCP connections and a third party JobTread integration layer. On the accounting side an AI agent helps input vendor bills, match purchase orders and budget lines and support reconciliation in JobTread. Claude also turns recorded client calls into draft proposals and pulls job decisions together into final scopes and budgets.

Result Commission calculations that took a full day are now done in under an hour, with an audit trail.

8 staffUS

OperationsVendor story

SCS Advisory

Wealth management firm, Bellevue, Washington

A consultant built a compensation splitting application that runs on the operations lead own Windows PC using local Qwen models, with no cloud and no data leaving the machine. She types a rule change in plain English, the app updates the formula and the written rule together and shows her both to confirm, then runs the quarterly calculation in one click.

Result The firm saves about 200 hours a year on quarterly compensation calculations.

Not statedUS

OperationsVendor storyAI agent

Golden International Corporation

Food importer and distributor, Renton, Washington

A consultant built an order entry agent for the firm using Gemini and n8n. It takes orders sent as handwriting, photos, PDFs or typed documents, reads them, matches each line against the product catalogue of over a thousand items, scores its confidence and sends high confidence orders straight into the old ERP system, flagging the rest for a person.

Result Order entry fell from up to two hours per order to about five minutes.

Not statedUS

Sales and marketingVendor storyAI agent

Retail Trust

Retail wellbeing charity working with more than 200 retailers

CGI built a network of AI agents into the charity's existing Happiness Dashboard. The agents gather material from guidelines, mental health guides, demographic data and retailer feedback, generate tailored campaign content, validate it separately, capture feedback and log every output for human review.

Result CGI reports that the campaign work now needs 80% less manual effort.

About 127 staffUK