Databricks CEO Ali Ghodsi wanted to raise $1 billion. He ended up with $15 billion in investor demand chasing the round, and closed at $5 billion anyway, pushing Databricks to a $190 billion valuation, up from $134 billion just six months ago. The company crossed $7 billion in annualized revenue, growing 80% year over year, and Ghodsi says the real driver is enterprises spending on AI agents that "remember context, deliver accurate answers, and execute work without blowing through their budgets."

Speaking of agents behaving badly: Anthropic published new research showing that when its own Claude agents were given the same software project with conflicting instructions, and no idea other agents were involved, they didn't collaborate, they went to war.

Agents disabled each other's accounts, wrote self-replicating malware disguised as belonging to someone else, and in a separate pricing experiment, secretly colluded to fix prices even after their communication channel was removed. And separately, since a few of you asked, we're breaking down exactly what's true (and not true) about Claude's new text watermarking, which has generated real backlash this week.

• Argon's debut use case is autonomous cybersecurity, finding and patching bugs on its own.

• ElevenLabs hit a $22B valuation through an employee stock sale, not a funding round.

• A Berlin startup raised $20M building the plumbing that keeps AI agents from losing their place mid-task.

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

Google released Gemini 4 Argon on Tuesday, built to handle coding, research, and writing, with cybersecurity as its specific edge. The model is rolling out only to a select group of Google's cyber partners through its Fairwind Program, and Google says it's trained for defensive work, able to "autonomously find, validate, and patch critical software vulnerabilities." Google says Argon is also strong at coding and engineering, with its own staff already using it for daily debugging and codebase migrations, and that it can parse long videos and charts. "Built to sustain deep reasoning across complex, long-horizon workflows, Argon is fundamentally changing the way we work and build at Google," the company said in its announcement. To back the "most powerful yet" claim, Google points to Vals, an AI benchmarking startup, saying Argon currently leads Vals' model index, ahead of OpenAI's GPT-6 Astra and Anthropic's Fable and Opus models. It's worth remembering every major lab makes a version of this same claim at launch, OpenAI said much the same about Astra in September, and Anthropic said it about Fable in June. Google was widely seen as behind in the AI race for most of last year; its Gemini app passed a billion monthly users in August, putting it on par with ChatGPT.

Voice AI startup ElevenLabs let employees sell a portion of their vested equity this week at a $22 billion valuation, double the $11 billion mark it hit in February. The $300 million tender offer was co-led by Wellington and T. Rowe Price, large institutional investors that typically back private companies planning to hold the stock through an eventual IPO. This is the second time the four-year-old company has run this kind of sale for staff; the first was a $100 million tender at a $6.6 billion valuation back in September 2025. The move fits a broader pattern among fast-growing AI startups, using employee liquidity to keep people from jumping to a competitor without forcing a full funding round. Founded in 2022 and known for ultra-realistic AI voices and sound effects, New York- and London-based ElevenLabs now ranks among Europe's most valuable startups.

Senior Software Engineer | Curri
Remote (US) (Full-time, Full stack)
Build the construction and industrial logistics platform.

Software Engineer, Product | Numeral
New York, NY / Remote (US) (Full-time, Full stack)
Build the AI-native sales tax solution.

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

Product Engineer | Lamar Health
San Francisco, CA / San Mateo, CA / Remote (US) (Full-time, Backend) Take the paperwork burden off patients accessing expensive medications.

Staff Full Stack Engineer | Sully
Mountain View, CA (Full-time, Full stack)
Build the autonomous AI agents that run entire hospital operations.

Fullstack Engineer (NYC) | BoldVoice
New York, NY (Full-time, Full stack)
Build the speech and accent coaching app for non-native English speakers.

Principal Engineer, Backend | Encord
London, UK (Full-time, Backend)
Build the data layer for physical AI.

Found the right person, but they’re in a country where you don’t have an entity? Setting one up can take months and significant cost.

Remote removes that barrier by becoming the legal employer through our own entities — handling compliant contracts, local benefits, tax setup, and onboarding for you. In fact, an employee is onboarded to Remote every 7 minutes.

Once they’re hired, the same in-house teams that support employment locally also run payroll — so you’re not bouncing between disconnected providers. Less setup, less complexity, and less time between finding the right person and getting them started.

See how it works

Restate Raises $20M as AI Agents Make Durable Infrastructure Harder to Avoid

The Berlin startup, which builds infrastructure to keep multistep workflows reliable through crashes and network interruptions, closed a $20 million Series A led by Singular with Redpoint Ventures and Capital One Ventures. Founder Stephan Ewen said the technology "wasn't built for agents in the beginning, but it just happened to be a perfect match" for the problems agent workflows create. Customers already include Replit. The round is a direct shot at Temporal, the category's biggest player, which closed a $550 million round at a $12.55 billion valuation earlier this month.

Flow Engineering Lands at a $750M Valuation

Valor, Atreides, and Sequoia backed the AI startup at a $750 million valuation, the latest sign investors are still writing large checks for companies building at the application layer of AI.

Today's Growth Recommendation

Stop guessing whether to build or buy your customer support agents. Front built a four-part conversation series on agentic AI in customer ops, covering the true costs of maintenance versus flexibility, starting October 13, free to join

Register for the Webinar Below ↓

Buying or building your customer service agent both carry a cost the pitch skips: maintenance if you build, flexibility if you buy. Running Agents in Customer Work is four conversations on agentic AI in customer ops, this one on build vs. buy. Register now, four Tuesdays, 10 a.m. PT.

Register today

Leverage Long Decode Continuation for Massive Context Workflows

With Google DeepMind releasing Gemini 4 Argon and its breakthrough Long Decode Continuation API feature, pushing output token limits up to 1M by pausing and seamlessly resuming responses across API calls, handling ultra-long outputs has fundamentally shifted. Instead of breaking down massive reports, code migrations, or multi-repository analysis into fragmented chunks, you can now architect prompts that expect deep, uninterrupted generation cycles.

To take advantage of this, structure your instructions to let the model run continuous token generation for deep analytical or architectural output, rather than forcing short summaries. Define clear milestones within a single long-form execution so the model maintains context across extended decodes without drifting off-task.

The Long-Decode Deep Architecture Brief

Invite friends & get instant rewards

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