The voice assistant era is closing, Gemini takes its place
According to The Decoder's report of August 5, 2026, Google starts removing Google Assistant from September 4, 2026. The change covers phones, tablets, Wear OS watches, headphones and cars running Android Auto. The rollout can take a few weeks, and once it completes on a device there is no way back.
A decade old product retires
Per The Decoder's report of August 5, 2026, the shutdown begins on September 4, 2026. Search Engine Roundtable covered it the same day. The process does not finish in a day; the rollout could take a few weeks to reach all users.
The scope is broad. Affected devices include smartphones, tablets, Wear OS watches, headphones and cars running Android Auto. Cars with Google built-in continue to support Assistant for now. Google TV, Google Home speakers and smart displays will receive Gemini.
On reversibility the report is clear: after the rollout completes, Assistant no longer works and there is no option to switch back.
The same button, a different system
From the user's side the change looks small on the surface: the same voice button, a different system producing the answer. But the architecture underneath differs. Assistant was built on matching defined commands; Gemini runs on a language model.
The practical consequence is answers that are more flexible but less predictable. The assumption that a given command returns the same result every time weakens. The Decoder also notes that this timeline was originally planned for 2025 and postponed, with 9to5Google as the source.
One boundary is worth stating: the sources we reviewed do not detail the order of the transition by country, which devices are affected first, or what happens on older hardware. We make no claim about those here either.
A language model now answers the voice query
For businesses the point is not the assistant's name but the layer producing the answer changing. When a user says "best hairdresser near me" to their phone, a defined search flow used to run; now a language model composes the answer and picks the sources.
That carries the search visibility conversation into the voice channel. Being chosen as a source by a language model works differently from classic ranking: structured data, consistent business information and machine readable content become decisive. That assessment is ours, the sources offer no visibility analysis.
The second point is on the device side. On screenless or small screen devices such as Wear OS watches and headphones, the answer narrows to a single source. With no ten blue links on screen, there is no middle ground between being the one chosen source and not appearing at all.
Third is the loss of predictability. In the Assistant era the behaviour of specific commands was fixed; with a language model the same question can reach different sources at different times. That makes measurement harder too.
What it means for businesses in Türkiye
The assessment below is our reading, not something stated in the sources. Neither contains a timeline or scope note specific to Türkiye.
Voice search is widespread in Türkiye, particularly in cars and for local service queries. In categories like hairdressers, repair services, pharmacies and restaurants, a search spoken into a phone carries a closer purchase intent than one typed on a desktop.
Four practical notes. First, business information: address, opening hours and phone number should be consistent and current on your Google Business Profile, because you get no chance to correct a spoken answer. Second, structured data: LocalBusiness schema and correct fields on your site improve machine readability. Third, natural language: service pages carrying headings phrased the way people actually ask makes matching easier for a language model. Fourth, expectations: measuring traffic from the voice channel separately is practically difficult today, so treat this as basic hygiene rather than a campaign line item.
In short the work is not new, but it matters more: if your information is consistent, whichever assistant arrives will describe you correctly. That last sentence is our own comment.
The UNALSOFT take
In our web design work, business information and structured data are taken as seriously as the page's visible design. The reason is simple: scenarios where the reader is not a human keep growing. The Assistant to Gemini switch is the most concrete example; the same question will be answered by a different system. The preparation is not a campaign but information hygiene.
Is your business information the same everywhere?
Let's review your business profile and structured data together.