Ask Your CRM Anything. Get Answers Instantly.

Your sales data just became actionable.

No more digging through notes, emails, or dashboards. Just ask, and let AI find what you need in seconds.

Attio connects your customer data, automates your workflow, and gives your team the insights they need to close deals faster.

Today In Ai

1  Anthropic ships Claude Sonnet 5 and a dedicated research app that connects to 60+ scientific databases.

Claude Sonnet 5 is Anthropic's most capable agentic Sonnet yet, matching Opus 4.8 performance at a significantly lower price. It is now available in the Claude app.

Alongside it, Anthropic launched Claude Science in beta, a dedicated research tool that can run analyses, integrate with your existing tech stack, and pull from more than 60 scientific databases in a single session. See how it works (4M+ views).

2  Acti replaces your phone keyboard with a Gemini-powered agent that reads your screen and takes action.

Acti swaps the standard iOS and Android keyboard with a Gemini-powered AI agent that reads what is on your screen, understands what you are trying to do, and acts directly from the keyboard without switching apps. Type what you need, press the Acti Bar, and it drafts replies, finds information, or completes tasks in the background. Explore what it can do or watch it in action.

3  Google opens personalized image generation to everyone and drops two new image and video models.

Gemini's personalized image generation, an opt-in feature that uses Gmail, Photos, and Search history to customize outputs, is now available to all eligible users. Google also released Nano Banana 2 Lite and Gemini Omni Flash, its fastest and cheapest image and video models yet. The official Nano Banana Prompt Guide is worth bookmarking if you plan to use either.

From The Frontier

The headline number. US job openings held flat at 7.6 million last May, unchanged from the previous month, according to the Bureau of Labor Statistics. At the macro level, the labor market looks frozen. But a new study from Ramp suggests the freeze is not evenly distributed.

What AI adoption actually does to headcount. Ramp analyzed roughly 22,000 companies and found that firms spending heavily on AI, defined as $30 or more per employee per month, grew headcount by 10.2% over the two years following adoption. Entry-level roles inside those same firms grew even faster, at 12%. Firms with low AI adoption saw no headcount growth over the same period.

The caveat. The study's authors are upfront that the data skews toward tech-forward, venture-backed companies that were likely already growing and hiring before they adopted AI. The sample is not a clean read on the broader economy. But the authors draw one firm conclusion from it: AI may not be creating jobs yet at scale, but it is not destroying them either.

Reshaping, not eliminating. Amazon Web Services CEO Matt Garman put it this way: "I do think that half of white-collar jobs may change, but wipe out and change are different." His comparison is the Excel analogy. Spreadsheet software wiped out manual calculation roles but created an entirely new demand for analyst work. The pattern may be repeating, just at a speed that makes it harder to see in the moment.

What to watch. The Ramp data and the BLS data are telling two different stories that are both true at the same time. The macro freeze is real. The growth inside AI-adopting firms is also real. The gap between those two groups is the story. Companies that are waiting to adopt are not just missing productivity gains. They are also missing the headcount growth that tends to follow adoption. The question is how long that gap can widen before it becomes irreversible.

What people are actually watching and sharing

Dashboards instead of decks. Disposable HTML dashboards built by AI may start replacing static PowerPoint presentations. If your AI is connected to a live data source, it can spin up a fresh dashboard on demand for any meeting, no slides required. Learn how the workflow operates (4,000 bookmarks).

Doomscrolling, now educational. NotebookLM rolled out a feature that turns complex topics into 60-second explainer clips, complete with images, formatted for short-form browsing. Whether it fixes doomscrolling or just makes it feel more productive is an open question.

The sandwich problem. A viral video captures the specific frustration of trying to walk an AI through the most basic real-world tasks, like cutting a sandwich (2M views). A useful reminder that the gap between AI capability and AI reliability is still wide at the edges.

Nine names on a list. A leaked list of US companies reportedly considering switching to Chinese AI has gone viral because it includes nine major tech names. The list's authenticity is unverified, but the conversation it triggered about vendor loyalty and cost pressure in enterprise AI is very real.

A VC picks up a CEO badge. Venture capitalist Chamath Palihapitiya says AI is moving so fast that the decisions made in the next few years will define the next 20. That conviction led him to step into the CEO role at his new AI coding startup, 8090, which just closed a $135M Series A.

Prompt Station

Stress-test any business idea before you spend a dollar on it

This ChatGPT prompt acts as a startup validation expert, running your idea through six dimensions of scrutiny, surfacing its three biggest risks, scoring it out of 100, and giving you three ways to validate it quickly with minimal cost. It is designed to be harder on your idea than your instincts will be, which is the whole point. Paste it in and follow the idea description at the end.

Act as a startup validation expert. I will provide a business idea, and your task is to rigorously evaluate its viability. Analyze it across the following dimensions: target market size, urgency of the problem, willingness to pay, existing competitors, differentiation potential, and scalability. Then identify the top 3 risks and how to mitigate them. Provide a validation score out of 100 with justification. Finally, suggest 3 ways to validate this idea quickly with minimal cost. Here is the idea: [INSERT IDEA].

Replace [INSERT IDEA] with a plain-language description of what you are building and who it is for. The more specific you are, the more useful the output. Try adding one sentence about your intended price point and one about who your closest existing competitor is, even if it is a manual process rather than a product. The validation score is less important than the three risk flags and the three cheap tests it surfaces, which tend to be the most actionable part of the output.

Introducing The First Agentic CRM

Get revenue agents, workflows, and automations across every stage of your motion. Access customer data in real time through Attio's web app, MCP, API, and SDK.

Then Ask Attio anything about your business and get instant answers.

It's the CRM that runs the work behind every win.

Keep Reading