In the first week of September, roughly $861 million flowed into AI startups across four headline rounds, and the pattern is hard to miss. Investors are backing the operational layer around AI, not just the models themselves. One company is building an enterprise operating system for agents, another is pushing autonomous tools into restaurants and local shops, and two more are attacking the security and reliability headaches that agents are beginning to create.
Wonderful Doubles to $5 Billion Building an 'Enterprise AI OS'
Wonderful, an Israeli-Dutch startup founded in early 2025, closed a $550 million Series C that pushed its valuation to $5 billion, roughly doubling its worth in about six months. Insight Partners led the round, and Salesforce joined as a strategic investor for the first time. The company positions its Wonderful AI OS as a shared operating layer for agents, workflows, and AI-native applications, and markets it as model-agnostic so enterprises can plug in whichever models they choose. Without a shared operating system, AI risks recreating the sprawl of traditional SaaS, CEO Bar Winkler warned.
Delivery leans on a forward-deployed engineers model that sends senior technical owners directly into customer environments. Wonderful is applying the approach across banking, telecommunications, healthcare, utilities, insurance, and retail, and has grown to around 650 employees across 35 markets, with plans for a roughly 1,000-strong team.
Owner Raises $240M to Put an 'AI CMO' in Every Local Shop
On August 28, restaurant-focused software startup Owner announced a $240 million Series D led by Growth Equity at Goldman Sachs Alternatives, valuing the company at $2.3 billion. It has passed $100 million in annual recurring revenue, climbing from about $1 million in 2020. Owner describes itself as an autonomous operational layer for restaurants, and eventually for every kind of local business, with agents that handle website management, online ordering, marketing, and customer support.
The company points to a striking set of results:
- Websites it powers convert an average of 146% more visitors into customers
- Businesses see roughly 40% growth in online traffic within 30 days of launch
- Its AI Grader diagnostic feeds more than 90% of new customers into the sales funnel
The competition is formidable: Toast, DoorDash, and Square each own part of the restaurant tech stack. Still, Owner sees a $44 billion market among US independent restaurants, expanding to about $105 billion across the US, Canada, the UK, and Europe, and to $785 billion if every brick-and-mortar local business is counted.
The world is racing to build AI to replace people's jobs, the company argues; Owner is building AI to do the jobs many small business owners have never been able to afford. Risks remain, including a pending lawsuit from Popmenu over its Website Grader tool and the challenge of scaling one agentic system across 15 different verticals.
The Agent Backbone Gets Real: AIR ($50M) and Empirik ($21M)
AIR, a six-month-old security startup founded by two veterans of Israel's Unit 8200 intelligence corps, emerged from stealth with $50 million raised across two seed rounds. Sequoia led the first $10 million tranche, and Greenoaks led the second $40 million. Founders Yair Saban and Niv Hoffman built a platform that discovers agents running inside companies, continuously vets the skills and tools those agents use, and blocks anything that fails security checks. Saban draws a direct comparison to the early 2000s, when unsigned drivers loaded code into the kernel; today, skills, plug-ins, and MCPs get no such oversight. AIR says it filters out roughly 27% of the add-ons and skills it discovers online and counts more than 20 customers, with the strongest demand coming from financial services and pharmaceuticals.
Separately, Empirik has spun out of Sequoia after raising a $21 million seed round from Sequoia, Canapi, and Alumni Ventures. Incubated at Sequoia since 2023 and led by former Quantum Metric and Salesforce observability executive Kartik Chandrayana, the startup builds observability software that tracks system changes and infers their ripple effects, acting as an autonomous traffic cop that flags the most dangerous updates for human review. What agentic AI did for software, Empirik wants to do for infrastructure engineering, Chandrayana said, and customers already include Guardant Health and a large consumer packaged goods company.
Taken together, the four rounds signal a maturing market. Institutions are no longer funding only compute and frontier-model bets; they are paying for the middleware and autonomy layers where value will actually be captured. If the operating systems, local-business agents, and agent security startups all deliver, this week may look less like a funding spike and more like the moment the AI economy started building its own plumbing.
Comments