Major Launches from Gmail, Apple, and Salesforce

Nov 1

Emerging Tech Roundup — November 1

The Quantious team’s top picks for timely trending news in the tech world.

This week in tech: Meta is reportedly developing a search engine for its chatbot, Gmail will now help you write an email on the web with AI, Apple Intelligence is officially here, Salesforce AI chatbot Agentforce hits general availability, Tabnine launches its code review agent, DeepMind and Hugging Face release SynthID to watermark LLM-generated text, and Apple offers hackers $1 million to hack its AI servers.

Meta is Reportedly Developing a Search Engine For Its Chatbot

(Engadget, October 28)

Meta is reportedly developing a search engine for its chatbot to reduce its reliance on Google and Microsoft—following setbacks like Apple's App Tracking Transparency feature, which costed Meta significant ad revenue. The search engine project aims to integrate web indexing directly into Meta AI, providing an alternative to Google Search and Microsoft Bing. Meta has been working on web indexing for at least eight months and recently partnered with Reuters to improve its AI's responses to news-related queries.

Gmail Will Now Help You Write an Email on The Web with AI

(The Verge, October 28)

Google is bringing its "Help me write" AI feature to Gmail on the web, allowing Google One AI Premium and Workspace Gemini users to generate or refine emails with ease. Users will find options to formalize, elaborate, shorten, or "polish" emails, with a new shortcut (Ctrl + H) for quick refinement on web drafts over 12 words. The rollout began earlier this week, and the "polish" shortcut will now automatically refine emails without needing multiple swipes on mobile.

The Morning After: Apple Intelligence is Officially Here

(Engadget, October 29)

Apple has officially launched "Apple Intelligence," its suite of AI tools across iOS, iPadOS, and macOS—bringing features like live transcription, text summaries, and enhanced proofreading to users. Initially previewed to beta testers, this AI integration enhances productivity by summarizing notifications, emails, and web content—though it doesn't yet include all WWDC-announced features. While some tools are valuable, others may feel less advanced compared to offerings from competitors like Google and Samsung, such as photo editing.

Salesforce AI Chatbot Agentforce Hits General Availability

(Tech Crunch, October 29)

Salesforce has launched Agentforce, an AI agent development platform offering low- or no-code chatbot deployment for employees and customers. CEO Marc Benioff has emphasized Agentforce's distinction from competitors like Microsoft, which he criticized for data inaccuracies in its Copilot AI, calling it "Clippy 2.0." Agentforce, central to Salesforce's vision of a “work operating system,” operates autonomously on data triggers and automations while aiming to support human collaboration for customer success.

Tabnine Launches its Code Review Agent

(Tech Crunch, October 29)

Tabnine has introduced a code review agent designed to help developers adhere to their organization's best practices. Organizations can configure the agent with their own documentation or “golden code repos,” allowing it to flag issues and suggest fixes as developers work. Additionally, Tabnine has partnered with companies like Redis to integrate pre-trained best practices, enabling developers to toggle these guidelines as needed.

DeepMind and Hugging Face Release SynthID to Watermark LLM-Generated Text

(Venture Beat, October 25)

Google DeepMind and Hugging Face have introduced SynthID Text, a watermarking tool that marks and detects text generated by specific large language models (LLMs) without affecting the model’s performance. Released in Hugging Face's Transformers library, SynthID embeds a statistical watermark into AI-generated text, allowing detection of specific LLM-produced text for moderation or verification purposes. Unlike other methods, SynthID doesn’t require model retraining or sensitive data storage. Instead it uses generative modeling to make context-specific modifications in text, creating a subtle, detectable signature.

Apple Offers $1 Million Bug Bounty to Anyone Who Can Hack Its AI Servers

(PC Mag, October 27)

Apple is offering up to $1 million to anyone who can hack its new AI-focused "Private Cloud Compute" servers, designed to handle complex Apple Intelligence tasks beyond the capabilities of on-device processing. The system prioritizes privacy by deleting user requests immediately after processing and employing end-to-end encryption—ensuring Apple cannot access request data. Apple has expanded its security testing program to the public, providing access to key source code, a virtual research environment, and a detailed security guide to facilitate this vetting process.