For most of the past year, Anthropic has been quietly winning the enterprise AI fight. New spending data from Ramp shows Anthropic's share of business AI spend climbing from 41% to nearly 44% between May and July. But the same data shows something shifting underneath: OpenAI is now growing faster quarter over quarter, driven largely by developers adopting GPT-5.6 Sol, while Anthropic's pricier Fable 5 tier has struggled with adoption, partly due to a 30-day data retention requirement that frustrated some business customers.

Meanwhile, the unglamorous business of feeding AI models training data keeps minting fast-growing startups. Micro1, a four-year-old company that pays doctors, lawyers, and engineers to label and evaluate AI outputs, grew its gross annual run rate from $100 million to $500 million in just eight months. Here's what else is worth knowing, plus real roles across engineering, product, and marketing you can apply to right now.

• New spending data shows OpenAI growing faster than Anthropic in Q3, even as Anthropic holds the bigger share.

• Micro1 grew its run rate 5x in eight months feeding AI labs training data.

• A short, separate update on where the Claude watermark story stands.

• What Claude's new text watermark actually proves, and what it doesn't.

…and a whole lot more that you can read about below.

Fintech company Ramp, which tracks AI spending across more than 70,000 US businesses using its corporate cards, found Anthropic's share rose from 41% in May to nearly 44% in July, while OpenAI held around 40%. But the more recent trend tells a different story: OpenAI is now growing faster quarter over quarter, driven by developers adopting GPT-5.6 Sol, while Anthropic's premium Fable 5 has seen weaker adoption, partly tied to its pricing and a 30-day data retention policy. The overall market keeps expanding too, from just over 50% of Ramp's business customers paying for AI in March to nearly 56% by July.

Anthropic's lead widened through July, but OpenAI's growth rate is now the one accelerating.

Micro1, which connects AI labs with contract doctors, lawyers, and engineers to label and evaluate model outputs, grew its gross annual run rate from $100 million to $500 million over the past eight months. Net of contractor payouts, that's $150-200 million, still behind rivals like Mercor ($2 billion) and Handshake ($1 billion), but proof the training-data market has room for more than one winner. Why it matters: as AI labs run out of cheap training data, demand for expert-labeled data keeps climbing, and some researchers now expect future data spending to eventually rival spending on compute itself.

AI-native finance platform Rillet officially locked in a $100 million Series B funding round today at a flat $1 billion valuation, led by ICONIQ Growth. The platform marks a major shift from legacy software by using autonomous AI agent networks to close corporate books, reconcile accounts, and manage multi-jurisdictional tax compliance with no manual human oversight required. Industry leaders point to the milestone as proof that enterprise automation is moving away from basic assistive "copilots" and towards entirely self-contained software ecosystems.

Senior/Staff Software Engineer, Product | Craniometrix
New York, NY (Full-time)
Build AI tools that help dementia patients age safely at home.

Product Manager | Clipboard
Hybrid (San Francisco, CA) / New York City / Remote (Full-time)
Shape the product for a platform that helps businesses cover every shift, reliably.

Full-Stack Software Development Lead | SiPhox Health
United States (Full-time)
Lead full-stack development for wearable and at-home health monitoring built on silicon photonic chips.

Founding AI Engineer | Voiceops
United States (Full-time)
Help B2C companies learn from every customer interaction using AI.

Software Engineer, Fullstack | Aviator
Remote (Full-time)
Build a Google-level engineering productivity suite for fast-moving software teams.

Senior/Staff Fullstack Engineer | WarpBuild
Remote (Full-time)
Build the CI infrastructure that makes software development up to 10x faster.

Marketing Operations Specialist / GTM Engineer | Signadot
Remote (US and Latin America) (Full-time)
Own the marketing systems and GTM engineering behind Signadot's testing infrastructure for AI agents.

Head of Business Development | Malama Health
San Mateo, California (Full-time)
Grow partnerships for a startup providing doula-led, holistic pregnancy and postpartum support.

Mechanical Engineer Intern | GRU Space
San Mateo, California (Internship)
Help build the first hotel on the Moon.

Founders deploying AI agents across sales, marketing, and customer service are closing more with smaller teams. 65% more sales leads. Less headcount.

Download the free Practical Guide to Agentic GTM for Startups and start building the stack your competitors dream of.

Get the Free Playbook

DMicro1 Reaches a $500M Gross Run Rate

Micro1 grew its gross annual run rate from $100 million to $500 million over the past eight months by connecting AI labs with contract domain experts for data labeling and evaluation, according to a person familiar with the company. The startup raised its Series A at a $500 million valuation last September and may have since raised another round at a higher valuation.

Autonomous Financial Architecture Challenger Rillet Clears Unicorn Valuation

AI-native accounting platform Rillet officially locked in a $100 million Series B funding round at a flat $1 billion valuation, backed heavily by ICONIQ Growth. Moving far past standard text-based copilots, Rillet deploys independent, self-contained multi-agent networks that execute end-to-end accounting actions. The software completely automates complex tasks like closing corporate ledgers and handling cross-border tax compliance with zero manual oversight

Alibaba Backing Dexmal in Huge $3B Physical AI Mega-Round

Embodied robotics pioneer Dexmal is finalizing a massive new funding round aiming for a $3 billion (20 billion yuan) valuation. The investment is heavily anchored by e-commerce giant Alibaba Group as part of its broader push into physical AI logistics. Dexmal’s technology focuses on developing multi-modal neural network brains that allow humanoid machinery to adaptively navigate messy, real-world warehouse environments rather than relying on the rigid, pre-programmed paths of traditional factory automation.

Today's Growth Recommendation

Every country has different employment laws, payroll rules, and notice periods. Oyster's Global Hiring Guides cover HR best practices in 50+ countries, so you avoid surprises before you hire.

Get the Country Guides Below ↓

Every country has different employment laws, payroll requirements, notice periods, and benefits expectations.

Oyster's Global Hiring Guides help you navigate the details, avoid surprises, and hire with confidence—wherever you're growing next.

Get the country guides

Give Your AI Coding Tool More Room Before It Forgets

Every AI coding tool has a memory limit, called a context window, and most default to a smaller one than the underlying model can actually handle. That default is tuned for speed and cost, not for the moment you're three hours into debugging a large codebase and the tool starts "forgetting" earlier context, missing a function you defined an hour ago, or losing track of a bug you already ruled out.

If you're using Codex, OpenAI's coding tool inside ChatGPT Desktop, there's a config setting most people don't know exists. GPT-5.6 Sol actually supports a 1 million token context window, more than triple what Codex uses by default, you just have to turn it on yourself.

Only reach for this on genuinely large codebases or long debugging sessions. For everyday small tasks, the default window is faster and cheaper, and a bigger context window doesn't make the model smarter, just less forgetful over a longer stretch of work.

Here's how to turn it on:

The Vague Ask Decoder

Use this the moment someone hands you an ambiguous task, before you start working on it, so you find out what they actually need instead of guessing.

Invite friends & get instant rewards

If you enjoy The Intelligent Edge, share it to unlock free resources when they subscribe.