Companies are giving AI specific jobs: checking security cameras, helping farmers understand their records, searching satellite images, and preparing sales quotes. This edition explains how those tools work, who can use them, and what companies have reported so far.
Sponsor
Benchmarks should test behaviour, not just answers.
Spec27 helps teams validate whether agents follow policy, handle pressure, avoid unsafe actions, and behave reliably across realistic scenarios.Finding security cameras that have stopped working
A security camera cannot protect a site if it is offline or its view is blocked. AITX, a company that develops AI security systems, announced a product on September 10 designed to find those problems.
Called SARA Assess, it connects to the software an organization already uses to manage its cameras. When it finds a fault, it passes the information to Circadian Risk, a platform that helps security teams assess risks across their sites. The system can flag an impaired camera and initiate a task or work order.
That connects detection to a useful next step: getting the problem fixed. Customers can use it without buying AITX's own security hardware.
AITX says the product is available for trials and full deployment. The announcement does not yet show how widely it has been adopted or whether it reduces security failures.
Letting farmers ask questions about their own farm data
John Deere, the farm-equipment maker, introduced an AI assistant called JD on September 1. It sits inside Operations Center, Deere's platform for managing information about fields, machines, and farm operations.
JD lets farmers ask questions about their own operational data. The idea is to help them get answers from records they already hold, using ordinary language instead of having to work through the data themselves.
For a farmer, the potential benefit is easier access to information when deciding how to run the farm. The assistant is initially available to selected agricultural customers in the United States, with a wider rollout planned for later in 2026. Deere has not yet established in this announcement that using it improves harvests or profits.
Making satellite images easier to search
Planet is a company that operates satellites and sells images and data about changes on Earth. Its customers include businesses and public agencies working in areas such as agriculture, forestry, and land management.
In its September 3 earnings announcement, Planet said its AI application for searching its data archive had entered open beta—a stage in which users can try a product that is still being tested. The application lets people search in ordinary language.
This could make satellite data useful to teams that do not have specialists trained to work with it. The change is in how people find information in Planet's existing archive. The announcement does not yet quantify how much time the application saves.
Preparing sales quotes
Atira's AI helps manufacturers analyze customer requirements and prepare sales quotes. Railway-equipment maker Robel reports saving 95 hours per quote request.
AI coding tools appear in an acquisition agreement
Circle, a digital-payments company, has agreed to buy Tazapay, which helps businesses make payments across borders. An AI-related detail appears in the acquisition agreement dated September 4.
The agreement's definition of an AI tool includes software that assists with writing, generating, or completing computer code. These are tools developers use while building software.
The definition belongs to this particular deal; it is not a new rule for all businesses. The broader lesson is that a company can be using AI internally even when it does not sell an AI product. Looking only at the features customers see can miss how employees build and maintain the software.
Wait, What?
Buy dumplings, get AI credits
Some Chinese businesses are giving customers credit to use AI services, much as a credit card might award airline miles.
Rest of World reported on September 4 that AI usage was being sold in telecom plans and offered as shopping rewards. One example is a credit card launched in July by Moonshot AI, the company behind the Kimi AI model, with Agricultural Bank of China and American Express. Spending earns allowances for using Kimi. A Beijing dumpling restaurant also gives diners computing credits.
These allowances are often measured in tokens: small pieces of text that an AI model processes when it reads a request and generates an answer. For the customer, a token allowance amounts to a budget for using the service.
The interesting change is how AI reaches people. Access can now come bundled with an ordinary purchase, without the customer first choosing a separate AI subscription.
Explore more examples
Our AI Use-Case Library collects reports of organizations putting AI to work. Our Edgar Radar tracks AI-related disclosures in documents companies file with the U.S. Securities and Exchange Commission. Those filings include annual reports, announcements, and deal agreements.
This edition draws on recent additions to both, from tools still being tested to reported customer results.
This week's poll
Which AI result would you most like companies to report next?
Which AI result would you most like companies to report next?
Until next time,
Alexis