Anthropic quietly shipped its most capable model into your existing plan. Here's what changed and why it matters.

Anthropic just made Claude a lot more useful without raising your bill. Claude Opus 5 is now the default model inside Claude Code for all paid subscribers, and the jump in what it can actually do is significant. The previous version scored 18.7% on a coding benchmark called Frontier-Bench. Opus 5 scores 43.3%. Think of it like swapping a capable intern for a senior engineer, same desk, same salary, better output.

What makes this practical for owners is the autonomous work angle. Opus 5 can run on complex, multi-step tasks for hours, catch its own mistakes along the way, and adjust course without you babysitting it. If you've been running Claude in any kind of workflow, whether that's drafting, research, or code, you're now getting materially more capable results. You can also type `/effort` to tell it how hard to work on a given task, which is useful when you want a quick answer versus a deep analysis. No extra cost, no new subscription. Read the Anthropic announcement.

One related note worth knowing: Claude Fable 5 (the previous top model) is moving to pay-per-use credits, and subscribers get a one-time $100 credit to claim before August 2. Log in at claude.ai and grab it before the deadline if you're a paid subscriber.

Voice control for AI agents is becoming a real workday tool. OpenAI rolled out voice control for its desktop app on both macOS and Windows, letting paid users direct ChatGPT's Work agents and Codex with spoken commands through something called GPT-Live. At the same time, Anthropic upgraded Claude's voice mode so it now runs on Opus and Sonnet, meaning the voice version actually reasons better. Claude's voice mode also gained mid-conversation tool connections like email and calendar. For an owner who's reviewing contracts on screen or doing physical work, being able to say "summarize this and draft a reply" without typing is a genuinely useful shift. OpenAI's announcement here.

Tools Worth a Look

Pushary: Sends AI agent approval requests to your phone lock screen so you can approve or reject without sitting at your desk. For solo founders running agents that need human sign-off.

Fedica: Social scheduling and analytics across 12-plus networks, with data on actual people engaging, not just aggregate follower counts. For owners managing presence across multiple platforms.

YC Has It: Describe your business problem in plain English and it surfaces the YC-backed startups that solve it, with pricing and integrations. Free, and faster than a Google spiral.

MinkNote: Private macOS notes app built on plain Markdown files you own. No account, no cloud, no vendor holding your data hostage. For Mac-based owners who want a fast, clean workspace.

Firecrawl: Web data API that lets AI agents scrape, search, and interact with any website at scale. For owners or developers building AI workflows that need live competitor data or content extraction.

More Headlines

ChatGPT, Claude, or Copilot for your finance team? A practical side-by-side: ChatGPT for flexible analysis, Claude for long documents and careful reasoning, Copilot if you're already in Microsoft 365. Pick based on what you're already running, not what you heard about first.

Miivo launches a self-serve AI customer insights tool for SMBs, designed for owners who want to analyze customer data without a data science team or enterprise pricing.

FAQ schema makes pages 3.2 times more likely to appear in AI Overviews, Semrush published a local AI search playbook worth reading if you care about being found in AI-generated results.

Higher Images launches a national AI Visibility Readiness Initiative to help small businesses audit how accurately AI platforms like ChatGPT and Perplexity actually describe them.

Microsoft 365 available for $70/year for entrepreneurs, a QuickBooks survey in the same piece found 74% of AI-using entrepreneurs say it boosts their productivity.

Plaud AI updates: Ask Plaud now builds persistent memory of your role and preferences, and Auto Polish cleans transcripts automatically on arrival. Useful if you record a lot of meetings.

FLUX 3 by Black Forest Labs: A unified video, image, and audio model in early access, already being used in Audi factory robots. Early days, but worth watching.

Tech founders are attempting to appoint an AI as CEO of an acquired company, a stunt or a real test of autonomous business operations, depending on who you ask.

-Scott

SmartOwner is published (almost) daily by the team at DigitalTreehouse. Want AI consulting or automations for your business? Reply to this email.