In March 2016, AlphaGo beat Lee Sedol at Go. In April, Facebook opened Messenger to bots. Through the autumn, new watches, voice devices and virtual reality headsets arrived. Each launch suggested a different future for software.
A technical breakthrough, a new distribution channel and a sustainable business are three different achievements. Looking back at 2016 with that distinction in mind is more useful than treating every announcement as proof of lasting commercial success.
This article revisits six notable technology trends from that year. They are an editorial selection, not a ranked market-share analysis. Historical milestones are linked to primary sources. The product lessons are Addbox’s editorial interpretation. Not every forecast from 2016 came true, and a launch does not establish that a particular application will endure.
AI as public spectacle
In March 2016, AlphaGo defeated Lee Sedol four games to one. DeepMind described a system that combined deep neural networks with advanced search. It was a remarkable result in a game with an enormous search space.
AlphaGo was specialised for Go. It was not a general-purpose assistant that could answer arbitrary questions or complete unrelated business tasks. The public attention it received can make it easy to blur those boundaries when planning a product today.
A year later, AlphaGo Zero learned through self-play without human game data. That development showed how quickly the underlying research could move, but it still addressed a defined problem with clear rules and measurable outcomes.
For a business building with AI, the useful question is narrower than “can we use AI?” Define the task, decide what would count as success and test against real inputs, awkward cases and measurable outcomes rather than a polished demonstration.
Lasting product lesson: define the task and the evidence of success. Test real inputs, awkward cases and measurable outcomes before treating a prototype as production-ready.
Chatbots as distribution
In April 2016, Facebook opened its Messenger platform to bots. The idea was straightforward: meet customers where they already spend time, inside a messaging app they use daily.
That is a distribution decision as much as a technology one. A bot on an existing platform can reach users without a separate app install, but it also inherits the platform’s rules, discovery model and user expectations.
By F8 2018, Facebook reported more than 300,000 active bots on Messenger. That is platform-reported adoption. It does not prove that every bot was useful, profitable or still maintained.
Conversation suits some tasks well: a quick question, a status update or a guided choice between a few options. Lists, forms, calendars and detailed comparisons often work better in an interface designed for them. A bot that forces every interaction into free text can frustrate users who simply need to complete a defined action.
A useful conversational product still needs reliable information, a way to complete the action and a route to human help when automation reaches its limit.
Lasting product lesson: choose the interface around the task. Users need reliable information, a way to complete the action and a route to help when the automated path is not enough.
Voice assistants and context
Google launched Google Home in October 2016, joining a growing category of voice-controlled devices for the home. The product placed a conversational interface in the kitchen or living room rather than on a desk with a keyboard.
That environment changes what works. Long lists of options are hard to remember when you cannot scroll back through them. Shared spaces raise questions about privacy, accidental triggers and who hears a response. Background noise and interruptions are normal, not edge cases.
Assess a voice product by whether the user can complete the task, confirm important actions and understand the result without relying on a screen. A booking that requires checking three dates, two prices and a cancellation policy may need a different channel, even if voice feels novel.
Lasting product lesson: the environment is part of the product. Design for where and how people will actually use the service, including confirmation, privacy and what happens when a spoken answer is missed.
Wearables and fitness
In September 2016, Apple introduced the Apple Watch Series 2 with built-in GPS and water resistance to 50 metres. The same month, Apple and Nike launched the Apple Watch Nike+, aimed at runners.
Those products illustrated a useful direction for wearables: specific activity and context rather than moving every phone feature to a smaller display. GPS on the wrist matters for outdoor exercise. Water resistance matters for swimming. A running-focused variant signals which recurring need the hardware is meant to serve.
Sensor recording on a device is not the same as a third-party app being able to access and use that data reliably. Permissions, syncing, battery life and what happens when the phone is out of range all shape whether a fitness product actually works in practice.
Lasting product lesson: begin with a recurring user need. Build the hardware, software and permissions model around an activity people will repeat, not around feature parity with a phone.
Virtual reality
PlayStation VR launched on 13 October 2016, bringing virtual reality to a console audience with a lower barrier to entry than many PC-based setups of the time.
Continued hardware investment does not, by itself, establish the business case for every application. Sony’s launch of PSVR2 in February 2023 shows that immersive gaming remained part of the company’s product line. A platform launch announcement still leaves open whether a particular experience justifies the cost of adoption.
A headset changes setup, accessibility, comfort, interaction design and the physical environment required to use the product safely. Users need space, time to configure equipment and often guidance when something goes wrong.
The benefit of a VR experience must justify the hardware, preparation and support. That calculation is different for a ten-minute arcade demo and for software someone is expected to use regularly at work.
Lasting product lesson: include the cost of using the product. The benefit must justify the hardware, preparation and support required before and during each session.
Serverless
At re:Invent 2016, AWS announced capabilities including Step Functions and further Lambda features. Serverless computing attracted attention because it promised to run code without managing servers directly.
Serverless did not begin in 2016. The year’s announcements made the model more visible and easier to adopt for teams already using cloud infrastructure. The appeal is real: less operational overhead for some workloads and a pricing model tied to execution rather than idle capacity.
Convenient infrastructure does not remove engineering decisions. Splitting a booking into payment, confirmation and calendar update still needs defined failure behaviour. What happens when a payment succeeds but the confirmation email fails? Who can retry a step, and what does the customer see while the system recovers?
Access controls, observability and cost also need deliberate design. A function that scales automatically can scale its bill just as quickly if every invocation triggers expensive downstream calls.
Lasting product lesson: convenient infrastructure still needs deliberate engineering for retries, failed requests, access controls and cost.
Lessons to carry into a new project
The table below pairs each 2016 development with a question worth asking before you commit to a similar direction today. The milestones are historical; the questions are for product planning now.
| 2016 development | Question worth carrying into a new project |
|---|---|
| AlphaGo | What task are you solving, and what evidence on real inputs would prove success? |
| Messenger bots | Does the interface fit the task, with reliable information, a way to complete the action and a route to help? |
| Google Home | Can users complete, confirm and understand the result in the environment where they will actually use it? |
| Fitness-focused wearables | What recurring need does the product serve beyond shrinking a phone interface? |
| PlayStation VR | Does the benefit justify the hardware, preparation and support required to use the product? |
| Serverless tools | How will retries, failures, access controls and cost behave in production? |
None of these questions replaces market research or user testing. They do help separate a memorable launch from evidence that a product can work reliably for the people you intend to serve.
Use the history, plan the product
Revisiting 2016 is useful when it sharpens decisions about a product you are building now. The trends that endured did so because teams matched the technology to a defined need, tested it in real conditions and invested in the work that surrounds the interface.
If you are scoping software around AI, conversational interfaces, mobile devices or cloud infrastructure, start with the task and the evidence you need before expanding scope. Our software development services help teams turn that scope into a reliable product. For experiences that belong on a phone or watch, see mobile app development. For AI planning and governance, explore AI strategy.
Turn a product idea into a plan
Whether you are exploring a new interface, a connected device or cloud-backed workflows, define the user need and the evidence that would justify building it.