In partnership with
Hello from The AI Night,
Today in AI:
Gemini Intelligence Turns Android Into AI Agent
Anthropic Launches Goal-Driven Automation for Claude Code
Manus Expands Similarweb Data for SEO Research
Here's the deal: Google announced Gemini Intelligence, a system-level integration bringing Gemini directly into Android. The upgrade automates multi-step tasks, fills forms, and builds custom interfaces across devices.
The Breakdown:
Gemini automates tasks across food, rideshare, and shopping apps using on-screen context and voice commands.
Gemini in Chrome launches late June with browsing assistance for research, summarization, and appointment booking.
Intelligent autofill uses Gemini to complete complex forms across apps with strict opt-in controls.
Rambler converts unstructured speech into polished text via Gboard, supporting multilingual input in a single message.
Create My Widget builds custom home screen widgets from natural language prompts, extending to Wear OS.
Rollout starts this summer on Galaxy S26 and Pixel 10, expanding to watches, cars, glasses, and laptops later in 2026.
The bigger picture: Apple Intelligence asks what you want to know. Gemini Intelligence does what you want done. That is not a feature gap. It is a philosophical split that will define mobile for the next decade. Three billion Android devices becoming autonomous agents inside third-party apps means developers no longer control how users interact with their own products.
Here's the deal: Anthropic released /goal for Claude Code, a command that sets a completion condition and keeps Claude working across turns until the condition is met. A separate evaluator model checks progress after each turn, removing manual prompting between steps.
The Breakdown:
The command accepts natural language conditions up to 4,000 characters and starts working immediately.
After each turn, a fast model (defaulting to Haiku) checks whether the condition holds and returns a reason if work should continue.
One goal can be active per session, clearing automatically once met or when manually stopped.
Works in non-interactive mode, the desktop app, and Remote Control for CI/CD-style automation.
The evaluator judges only what Claude has surfaced in conversation; it cannot run tools or read files independently.
The bigger picture: The hardest part of AI coding agents is not starting them. It is getting them to stop at the right moment. Every developer has watched an agent overwrite working code because it didn't know the job was done. Using a separate model to judge completion sounds simple. It solves the one problem that keeps teams from letting agents run unsupervised.
Here's the deal: Manus deepened its Similarweb integration to let users analyze keywords, referrals, landing pages, and popular pages behind competitor traffic. The update is available to all Manus Pro users through plain-language prompts.
The Breakdown:
New metrics cover Keywords Overview, Website Analysis Keywords, Referrals, Landing Pages, and Popular Pages.
Pro users can access Similarweb data without a separate subscription, charged through Manus credits.
Paid Similarweb customers can connect their own API key via MCP connector.
Each Pro user gets 20,000 Similarweb data credits monthly.
Credit costs vary by metric: one keyword overview costs 17 credits, 100 popular pages costs 300.
Manus can package findings into slide decks, reports, or charts on request.
The bigger picture: Similarweb starts at $125/month and most teams only use it twice a quarter. Manus just made that same data available through credits, so you pay for what you actually pull. This is the playbook for AI platforms eating SaaS. Bundle expensive tools, charge per use, and make the original subscription feel like overpaying.
AI Agents Are Reading Your Docs. Are You Ready?
Last month, 48% of visitors to documentation sites across Mintlify were AI agents, not humans.
Claude Code, Cursor, and other coding agents are becoming the actual customers reading your docs. And they read everything.
This changes what good documentation means. Humans skim and forgive gaps. Agents methodically check every endpoint, read every guide, and compare you against alternatives with zero fatigue.
Your docs aren't just helping users anymore. They're your product's first interview with the machines deciding whether to recommend you.
That means: clear schema markup so agents can parse your content, real benchmarks instead of marketing fluff, open endpoints agents can actually test, and honest comparisons that emphasize strengths without hype.
Mintlify powers documentation for over 20,000 companies, reaching 100M+ people every year. We just raised a $45M Series B led by @a16z and @SalesforceVC to build the knowledge layer for the agent era.
What else you need to know:
Anthropic shared a prompt caching tip for the Claude API: sending the system prompt before the user message pre-warms the cache, cutting time-to-first-token on longer prompts.
Kimi launched Web Bridge, a browser extension that lets AI agents interact with websites through search, scroll, click, and type actions, with support for Claude Code, Cursor, and Codex.
Notion launched ntn, an open source CLI that brings its full API to the terminal and lets developers build and deploy Workers, designed for both humans and coding agents.
Perplexity's Computer tool now connects to Snowflake, letting users run SQL queries, build dashboards, and automate analysis against live warehouse data with admin-controlled access and shared business logic.
Anthropic partnered with the Gates Foundation to commit $200 million in grants, Claude credits, and technical support for global health, education, and economic mobility programs over four years.
That’s it for today’s edition of The AI Night.
Our goal is to cut through the noise, surface what actually changed, and explain why it matters.
2 ways to support us:
Forward this to your AI-curious friend → https://www.theainight.com
Sponsor The AI Night and reach 500+ AI builders daily → passionfroot.me/theainight