Preview
A preview is an early version of a product that a provider deliberately releases even though it isn't finished yet. Users are allowed to try it out, but must expect bugs, changes, and a lack of guarantees.
When a company builds a new program or a new AI feature, the launch is rarely a single day. Often there is an intermediate stage: the product is released even though the developers still consider it unfinished. This intermediate stage is called a preview. With it, the provider is essentially saying: “You may use it, but don’t complain if it breaks.” Preview often stands alongside similar words like Beta or Testing Phase, which mean almost the same thing. A comparable example is a theater preview performance: the audience is already allowed in, but the play will still be changed afterward.
Why providers show unfinished software
The most important reason is feedback. A team can never play through, in the office, all the situations real users will run into. A thousand people will find bugs that no one thought of. With AI products this is especially clear, because users employ the systems in ways the developers never planned for.
The second reason is speed. In the race for AI features, no provider wants to be second. A preview makes it possible to announce a date and get attention without everything having to be perfect. For investors and journalists, this is a signal: something is happening here.
For you as a user, the difference is tangible. A finished version usually comes with commitments: it will stay available, the price is fixed, your data remains accessible. With a preview, these commitments are often missing. The provider is allowed to shut down the feature, rebuild it, or suddenly charge money for it. So anyone building a school project or a company on top of it is taking a real risk.
From the closed round to general availability
A typical sequence involves several stages. First comes a closed preview, often called a Private Preview. Only selected partners or people on a waiting list get access. This keeps the circle small in case something goes wrong.
Next comes the open preview, also called Public Preview. Anyone can sign up, sometimes for a reduced fee or with a limited volume. At the end comes general availability, referred to in industry jargon as General Availability or GA for short. Only from this point on is the product officially considered stable and fully supported.
A common misconception is to equate preview with “poor quality.” Some preview models are already technically very good. The label mainly says something about the level of commitment, not necessarily about performance. Conversely, there are products that remain in preview for years because the provider deliberately doesn’t want to commit.
Preview labels in product names and announcements
The word is most visible in product names. Major AI providers release models with additions like “preview” or “exp” for experimental in the name. Cloud services too — that is, rented computing power over the internet — mark new features this way for months. In app stores, the same stages are sometimes called Beta or Early Access.
In business news, the distinction matters more than it sounds. When a company says a product is “in preview,” it often means it’s barely making any money from it yet. Only general availability brings in paying customers in large numbers. Analysts therefore pay close attention to which stage is meant.
Practical tip: with a preview, always check the terms of use regarding duration and your data. And save important results outside the service. Previews disappear faster than you’d think.