Enphase Energy Launches AI Assistant in Home Energy App

Coinbase Global reported that its agentic trading stack powered most crypto transaction flows over the past week. Management highlighted Grok as a leading AI agent on the platform, driving a large share of AI agent-driven trading volumes, while developer-built custom interfaces using Coinbase for Agents and the x402 payment protocol accounted for the majority of notional market activity. The company said the shift toward agentic trading moves it closer to a model leaning on higher margin services rather than pure retail volumes, with Grok and custom agents plugging into a payment stack built around USDC, Base and APIs. Coinbase added that heavier machine-driven flows do not remove existing risks around fee pressure, cybersecurity costs or competition from decentralized venues. The key proof point ahead is whether upcoming disclosures break out AI agent related metrics, such as the share of notional trading or fee revenue routed through Coinbase for Agents and x402, in the next couple of quarterly updates and product briefings.

Atlassian Corporation rolled out a set of tools on September 10 built around managing AI agents working on codebases, spanning Jira and its DX platform. The new Code Context feature, built on Atlassian's Teamwork Graph, gives Rovo and other coding agents visibility into architecture, requirements, and multi-repository codebases, while Agent Context Controls let platform teams decide which Jira and Confluence spaces those agents can see. Agent loops in Jira now scan backlogs for well-defined, unassigned tickets, hand them to Jira's coding agent for execution and testing, and open pull requests ready for human review, with a new Standards feature letting platform teams set coding rules once and apply them everywhere agents work. Atlassian's 2026 AI SDLC study found that 94% of engineering leaders already use AI, but only 6% have systems in place to scale that work across the full development lifecycle. The company reported a GAAP operating margin of just 0.2% and a GAAP net loss of $54 million for fiscal 2026, even as non-GAAP operating margin came in at 30% and non-GAAP net income reached $1,526 million.

CrowdStrike CEO George Kurtz said on CNBC's "Mad Money" on September 14 that "the genie's out of the bottle" on artificial intelligence, arguing that slowing frontier model development does nothing to secure the models already deployed. The comment helped send CrowdStrike stock up almost 14% on September 14 to a record close above $245 a share, with rival Palo Alto Networks also gaining. Kurtz was responding to a weekend essay by Anthropic CEO Dario Amodei urging AI labs to slow development and adopt outside evaluators and shared safety standards; President Donald Trump criticized Amodei, while OpenAI CEO Sam Altman and Elon Musk backed the idea. Kurtz said the security industry must protect existing models and supervise autonomous AI agents at runtime, adding that "the agents are dangerous" and "need to be controlled." CrowdStrike reported $333 million in net new annual recurring revenue and 31% year-over-year growth last quarter, with its Flex licensing model at nearly 40% of ARR mix, up from 34% a year earlier, and gross margin at 81% against a long-term target of 82% to 85%.