AI Company
Google DeepMind
Also known as: DeepMind
Google DeepMind is Google's artificial intelligence research and development unit, formed by combining DeepMind with the Google Brain team in 2023. DeepMind was founded in London in 2010 and acquired by Google in 2014. The unit develops Google's Gemini models alongside longer-running research projects such as AlphaFold.
What it is
Google DeepMind is the part of Alphabet responsible for frontier AI research and for building the models that sit behind Google's AI products. Its work ranges from scientific research, including protein structure prediction with AlphaFold and game-playing systems such as AlphaGo, through to the Gemini family of general-purpose models. Demis Hassabis, a DeepMind co-founder, leads the unit.
Why it matters
The models developed here power Google's AI search experiences, the Gemini app and Gemini features across Workspace and Android, so they influence how a large share of answers are generated and which sources get surfaced. When a new Gemini version ships, summary style, citation behaviour and answer length can change across those surfaces. For search and content teams that makes model releases a monitoring event, not just industry news.
How it works
Practitioners track model releases and documentation to understand what each version can do, then test their own priority prompts in Gemini and Google's AI search surfaces to see how their brand is represented. Log analysis and robots controls are used to manage Google's crawlers and the separate Google-Extended control that governs use of content for Gemini model training and grounding. Findings feed into content structure, entity clarity and reporting.
When it applies
It is relevant whenever you are auditing visibility in Google's AI surfaces, deciding how your content may be used for AI training, or explaining a shift in AI answer behaviour to stakeholders.
Examples
- A B2B team reruns a fixed set of buyer questions in Gemini after a new model version ships to check whether it is still cited.
- A publisher reviews its Google-Extended directive to set a deliberate position on AI training use of its archive.
- An in-house SEO briefs leadership that changes to AI Overviews stem from the underlying Gemini model rather than a ranking update.
How it is measured
- Brand mention and citation rate in Gemini and Google AI search answers for a tracked prompt set
- Referral sessions and assisted conversions attributed to Google AI surfaces
- Crawl requests by Google user agents, including Google-Extended coverage decisions
- Change in answer accuracy or sentiment about your brand before and after a model release
Insights on Google DeepMind
Related terms in AI Company
- Amazon Web ServicesAmazon Web Services is Amazon's cloud computing division, providing compute, storage, databases, networking and machine learning services on a pay as you go basis. It is one of the main cloud platforms used to host websites, data pipelines and AI applications. For AI work it offers services such as Amazon Bedrock, SageMaker, OpenSearch and various vector and database options.
- AnthropicAnthropic is an AI company that builds the Claude family of large language models and positions safety research at the centre of its work. Its products include the Claude apps, a developer API and Claude integrations used inside other tools. For marketers, Anthropic matters because Claude is one of the assistants where buyers now ask questions that used to start in a search engine.
- BroadcomBroadcom is a US semiconductor and infrastructure software company, listed as AVGO, that supplies networking, storage, wireless and custom silicon alongside a large enterprise software portfolio. In AI it is known for data centre networking components and for co-designing custom accelerators with large cloud operators. Its software side expanded significantly with the acquisition of VMware.
- CerebrasCerebras Systems is a US computing company that builds very large wafer-scale processors and systems for training and running AI models. It sells hardware and also offers a hosted inference service that runs open models at high token generation speeds. It matters to search and marketing teams mainly as one of the infrastructure options behind fast AI assistants and agents.
- GoogleGoogle is the search engine and technology company owned by Alphabet, and the operator of Search, YouTube, Chrome, Android, Google Ads and Google Cloud. It develops the Gemini family of AI models and surfaces generated answers in products such as AI Overviews. For most businesses it remains the largest single source of organic and paid search demand.
- MetaMeta is the US technology company behind Facebook, Instagram, WhatsApp, Messenger and the Quest headsets, formerly known as Facebook, Inc. It runs one of the largest digital advertising businesses in the world. It also develops the Llama family of open weight large language models and the Meta AI assistant built into its apps.
Primary research · August 2026
How ChatGPT Shortlists Software Brands
An audit across 10 categories and 60 buying questions. I recorded what ChatGPT reads, throws away and links to when a buyer asks it which software to buy, and what that decides.
- 60
- Questions asked
- 10
- Software markets
- 2,680
- Results read
- 367
- Links shown
Free download
Get the full report
35 pages · PDF · 536 KB. Enter your details and it downloads straight away.