IoT has never been about simply connecting things. What matters is the information that is exchanged over that connection, the way that it is processed and understood, and the actions, opportunities, and new value that enterprises, public agencies, and teams can create by exposing the intelligence in things.
Yet while the term “Internet of Things” has been with us for more than 25 years, until recently a key element has been missing. The IoT has certainly had the connections (according to IoT Analytics, now numbering in the billions), and it has had the information (zettabytes worth of data collected annually in recent years, by some estimates). However, even very successful large-scale deployments have existed in the form of highly specialized point solutions, with limited ability to abstract from data to generalize insight beyond the immediate use case. Even when AI and ML have been available, applying these capabilities required rare, specialized technical skills, and application has been limited to the immediate data set.
AI: the missing piece
The advent of large-scale, publicly accessible AI platforms, now offers the missing piece of the puzzle for IoT. Giving entire teams access to natural language interaction brings IoT connections and endpoints alive with intelligence. Current AI tooling also gives IoT deployments the ability to tap into historical data combine inputs from connected sensors, cameras, GPS trackers and other devices to generate deeper understanding and quickly carve out new pathways to revenue generation, cost savings, and end user satisfaction.
Public GenAI platforms thus represent an important step toward harnessing the full potential of the billions of connected devices already deployed worldwide. As we look ahead to the full convergence of IoT and AI, we are already seeing new use cases emerge that might not have been possible without massive investment even a year ago.
Image intelligence leads the way
Some of the earliest examples of this convergence of IoT and AI involve computer vision, a field in which the ability to conduct real-time analysis of image data and take quick action can translate to revenue generated or costs saved. Each of these examples is in use today by real IoT customers.
The first example involves a vast, busy warehouse where personnel and shipments are constantly moving in and out. In this environment, security is very important, so the warehouse has numerous surveillance cameras mounted throughout the interior and exterior of the space, all connected via IoT. While such cameras can record everything that is happening, it often takes trained eyes to notice anything amiss in the footage after the fact.
But, by using smart cameras that can relay image data to the cloud for instant AI analysis, anomalies (like a person wearing the wrong uniform or a fake badge) can be spotted immediately, warehouse management can be alerted quickly, and a potential loss by theft can be avoided. Even today, all of this can be achieved by integrating any of the many commonly general-purpose multimodal LLMs (like ChatGPT 4o or Google Gemini), with no specialized training required.
In another scenario, the same combination of IoT-connected cameras and AI in the cloud can be used in a grocery store, where numerous ready-to-eat meals sit stacked on shelves, aisle after aisle, waiting for customers to pick them up. Monitoring inventory status and frequency of purchases is important because if inventory remains high, it could lead to fresh food going to waste, translating to a financial loss. If the meals are being picked up with greater frequency than expected, there could be an opportunity to prepare more meals to satisfy demand, translating to a revenue gain.
In this case, AI can be used to analyze image data and recommend actions. For example, if inventory is too high late in the day, store staff can be notified with a recommendation to start offering price discounts on the meals in hopes of boosting sales and avoiding waste at the end of the day. If meals are flying off the shelves, the recommended action based on image analysis could be to make more meals. The financial implications of these use cases make AI a transformative force in each case.
AI as a two-way street
An important thing to note from the examples above is that warehouse operators and grocery store managers do not need to understand AI, coding, or other underlying technical aspects of how it all works in order to obtain AI insights and act on them. In the warehouse example, alarms can be automatically triggered if an anomaly is detected or locked containers are compromised. AI analysis and automated alerts can allow warehouse staff to focus on other tasks until something out of the ordinary occurs.
But, usage of AI also can be a two-way street. In the grocery store example, AI can create automated alerts based on how many meals remain on a shelf, but a store manager also could use a simple generative AI user interface to query natural-language requests like “Show me how many meals were left at the end of the day” or “Which hour of the day featured the most pick-ups?” Cloud-based AI processing and generative AI connected to the customer’s IoT platform should be able to do the rest.
The key to enabling these capabilities is to provide organizations with a foundation to deliver all the elements–IoT connectivity, cloud AI processing and generative AI support, data warehouse access, and service and application enablement.
The notion of a two-way street also applies to the sources of data flows that AI can analyze. Much of the value of AI comes from processing, understanding, and acting on the data generated from IoT endpoints. But, vast oceans of historical data and organizational knowledge also exist everywhere in an enterprise, not just at the endpoints. AI can leverage data throughout the organization, in some cases blending historical data and organizational knowledge with data gathered from IoT endpoints to create new instruction sets for AI tools, and help enhance AI insights and better inform the actions it recommends or autonomously takes.
For the next generation
For a company that has had IoT connections in place for years, AI could prove to be not just an important missing piece of the puzzle, but a revelation, the addition of a new dimension to what had been a somewhat static infrastructure. What is even more promising for the global IoT community is the possibility that AI will provide the foundation for the next generation of IoT use cases, built by enterprises and teams who may have not invested in IoT because they could not see a clear path between connecting devices and turning the data into something that could add value.
We may have already connected billions of devices through IoT, and wonder, “Where else can we go from here? Perhaps the job is done.” But AI can also help us create blueprints for tomorrow. It can spark the fires of our imaginations, and open doors to new possibilities in terms of new devices to connect, new markets to enter, new applications to create, and new things to do with our ever-deepening oceans of data.
AI also allows us to leverage insights from diverse industries and data sources. For example, hyperlocal weather information from companies using local cameras and user-generated images can help AI predict demand for food and goods at stores. These forecasts can then alert staff to adjust inventory and launch targeted promotions. By combining IoT and AI, we create new demand for shared data and open opportunities for a broader data-sharing economy.
As a global IoT community, we are not exactly sure where we go from here with IoT. The future may be increasingly automated, but we still control our plans and goals. AI will not decide the future of IoT for us, but it is an important tool we can put to use as we clear the path to the future. Our biggest goal in the short term is to make sure that IoT users are aware of the potential benefits that AI can bring and show them how to integrate AI into their IoT strategies and infrastructures.
