Google on April 29, 2026 introduced two distinct personalisation features for Gemini users in the United Kingdom: a persistent memory setting that draws on past conversations to inform future responses, and a set of import tools allowing users to carry their memories and full chat history from competing AI assistants directly into Gemini. The announcement, posted on Google's The Keyword blog and authored by Maryam Sanglaji and Animish Sivaramakrishnan, both Group Product Managers on the Gemini App team, marks a notable expansion of Gemini's personalisation infrastructure into a market where data privacy regulation shapes product design at every level.

Memory is not new for Gemini globally. Google first added cross-chat memory capabilities to Gemini Advanced for paying Google One AI Premium Plan subscribers in February 2025. That initial rollout was limited to premium accounts. Then, in August 2025, Google introduced a broader Personal Context feature and Temporary Chat functionality, extending memory to a wider set of users while simultaneously giving privacy-conscious users an ephemeral conversation mode. The April 29 announcement represents the formalisation of this capability for UK consumer accounts - a market where the interplay between AI utility and regulatory scrutiny is particularly acute.

How the Memory setting works

According to the announcement, the feature is called "Memories" and operates through a setting labelled "Personal context" inside the Gemini app. When activated, Gemini retains key details and preferences shared across past conversations, then draws on that accumulated context when answering new queries. Google describes the intended effect as conversations that feel like collaborating with a partner who is already up to speed.

The document provides three concrete usage scenarios to illustrate how stored context translates into personalised outputs. In the first, a user who has discussed superhero comic book characters with Gemini could ask for a unique birthday party theme and receive suggestions tied to a specific character, including themed food and a custom photo booth with props. In the second, a user who has previously asked for BookTok non-fiction summaries might ask for a book recommendation and receive titles with similar thematic profiles, along with quotes suitable for social posts. In the third - which will be familiar to anyone who has followed the Gemini personalisation roadmap - a user who has brainstormed YouTube content focused on Japanese culture could ask for "new content ideas based on my interests" and receive specific format suggestions, such as a "My First Time Trying..." video structure centred on Japanese food, traditional crafts like origami or a tea ceremony, or Japanese games.

The memory setting is on by default in the UK rollout. Users who want to disable it navigate to Settings in the Gemini app, select "Personal context," and then "Memory." Past conversations remain manageable through Gemini Apps Activity, where individual exchanges can be reviewed or deleted at any time.

According to the announcement, the feature began rolling out on April 29, 2026, with availability extending to all UK users over the coming weeks. The post does not specify a precise end date for the full rollout.

Import tools: memories and full chat history

The second and technically more novel component of the announcement is a pair of import mechanisms designed to lower the friction of switching from another AI assistant to Gemini. According to the document, these tools cover two distinct types of data: a structured memory import and a full conversation history import.

The memory import feature works through a manual copy-paste process. A user navigates to Settings in the Gemini app and selects the new import option. Gemini then provides a suggested prompt. The user copies that prompt, pastes it into their existing AI app - the announcement does not name specific competitors but refers generically to "other AI apps" - and that app generates a summary of the user's stored preferences. The user copies the resulting summary and pastes it back into Gemini. Gemini then analyses the text and saves the extracted details to the user's Gemini context. The outcome, according to the announcement, is that Gemini will understand key facts previously stored elsewhere, such as interests, a sibling's name, or a hometown, without the user having to re-enter that information manually.

The chat history import is structurally different. Rather than a prompt-and-paste workflow, it involves uploading a ZIP file. Users export their conversation archive from another AI provider, upload the ZIP into Gemini, and can then search past threads and continue building on them within the Gemini interface. According to the announcement, the process is designed so that a user can "pick up right where they left off."

Both import mechanisms are described as rolling out over the next couple of weeks from the April 29 announcement date, accessible through the Settings page. Notably, the announcement includes a footnote specifying that memory and chat history imports are not available for Business, Enterprise, or under-18 accounts.

The competitive context

The timing of these features is not coincidental. The AI assistant market has shifted considerably since early 2025. Gemini's traffic share climbed from 5.3% to 22% of global AI website visits across calendar year 2025, while ChatGPT's share declined from 86.6% to 64.6% over the same period, according to Similarweb data published in January 2026. Gemini's monthly active user base grew from 450 million in July 2025 to 650 million by October 2025. That growth, while substantial, has been built largely on Google's distribution advantages - Gemini is embedded across Search, Gmail, Google Workspace, and Android devices. Attracting users who are already invested in a competing assistant requires something different: a credible mechanism for preserving what they have already built.

The import tooling addresses that specific friction point. Users of ChatGPT, Claude, or other assistants accumulate context over time - preferences, project history, recurring tasks. Starting over with a new assistant means losing that continuity. By enabling structured import of both memories and full conversation archives, Google is attempting to remove one of the more durable barriers to switching.

OpenAI introduced memory features for ChatGPT in December 2024. Google followed with Gemini Advanced memory in February 2025. The sequence matters: each iteration has expanded the scope and broadened the user base eligible to access persistent context. The April 29 announcement extends that trajectory into the UK consumer market while adding the import dimension that earlier rollouts did not include.

Privacy implications in the UK context deserve attention. Google Gemini's automatic activation of Android device permissions for phone, messaging, and WhatsApp in July 2025 drew significant criticism from privacy advocates and legal observers in Germany and across Europe, with concerns about whether opt-out defaults met the requirements of GDPR's lawful basis provisions. The memory feature announced on April 29 is also on by default, which follows the same structural pattern. The announcement provides a straightforward path to disabling the setting, but the default-on design means UK users who do not actively review their settings will have memory activated without taking any affirmative action.

The chat history import feature introduces a separate data consideration. Uploading a ZIP file of conversations from another AI provider into Gemini involves moving a potentially substantial body of personal data between platforms. The announcement does not detail how Gemini processes, stores, or retains the uploaded archive beyond confirming that past threads can be searched and continued within the app.

What this means for the marketing community

For the marketing and advertising technology community, the practical significance of these features operates on two levels. At the product level, persistent memory in AI assistants changes the nature of the tool. A user who briefs Gemini on a client's brand positioning, target audience, or content calendar in one session may find that subsequent sessions require less re-contextualisation. That compounding continuity is particularly relevant for tasks that span days or weeks - campaign planning, content series development, or market research synthesis.

At the competitive intelligence level, the import mechanism provides a direct window into how Google is approaching user acquisition in an increasingly fragmented AI assistant market. Rather than competing purely on model capability - which is difficult to evaluate in isolation - Google is competing on switching cost reduction. The ability to carry memories and history from a competitor into Gemini reframes the decision calculus for users who might otherwise stay with a familiar tool despite finding Gemini's underlying capabilities appealing.

The broader trajectory of Gemini's personalisation investment is visible in the arc of coverage at PPC Land: memory for premium subscribers in February 2025, Personal Context and Temporary Chat in August 2025, and now UK consumer rollout with import tooling in April 2026. Each step has extended reach and added capability. The pattern suggests Google views persistent context as a core differentiator for the Gemini product, not a peripheral feature.

There is also a question of what persistent memory means for the quality and consistency of AI-assisted work. When an assistant retains preferences across sessions, the outputs it generates - copy drafts, research briefs, audience analyses - become progressively more attuned to the user's specific framing and vocabulary. That compounding alignment can accelerate workflows. It can also introduce a kind of epistemic lock-in, where the assistant consistently reflects back the assumptions and terminology the user has already established, rather than challenging them. Neither outcome is inherently good or bad; both are worth understanding as the tools become more integrated into professional routines. The import mechanism adds a further dimension: a user migrating their full conversation history from another platform is not just bringing preferences, they are bringing the full record of how they have framed problems, briefed tasks, and received outputs over potentially months of use. Gemini will inherit that accumulated framing wholesale.

Timeline

Summary

Who: Google, represented by Gemini App Group Product Managers Maryam Sanglaji and Animish Sivaramakrishnan, announced features for UK consumer Gemini users. Business, Enterprise, and under-18 accounts are excluded from the import functionality.

What: Two sets of personalisation features were introduced. The first is a "Memories" setting that enables Gemini to retain preferences and details from past conversations, activated by default. The second is a pair of import tools - a structured memory import using a prompt-and-paste workflow, and a full chat history import via ZIP file upload - allowing users to bring data from competing AI assistants into Gemini.

When: The announcement was made on April 29, 2026. The Memories feature and memory import began rolling out on that date, with full availability for all UK users expected over the following weeks. The chat history import is also rolling out over the couple of weeks following the announcement date.

Where: The features are available in the United Kingdom for consumer Gemini accounts. The rollout is specific to the UK market in this announcement, though earlier iterations of Gemini memory features have been available in other markets since early 2025.

Why: Google is reducing the friction of switching from competing AI assistants to Gemini by enabling import of accumulated memories and conversation history. The move responds to a competitive market in which persistent context has become a defining product characteristic, and where Gemini's traffic share has grown significantly but established users of other platforms face meaningful switching costs associated with losing built-up conversational history.

Share this article
The link has been copied!