# Capital, Compute, and the Rules Catching Up

> Today's drop follows the money and the rails it buys: record funding pooling into a handful of mega-rounds while startups like Together, Venice, Taktile, and Oxmiq carve out room, and hyperscalers steering toward a 600 billion dollar capex war. Governments are moving in step, with the EU's AI Act going enforceable, the White House negotiating pre-release reviews, and California standardizing on Claude across its agencies. Underneath the headlines, applied AI keeps quietly delivering, from earlier cancer detection to editors that now run a fleet of coding agents at once.

_Wortins AI briefing · Monday, July 13, 2026 · Updated 2026-07-13_

## Daily AI Updates

### [Google Search Completely Replaces Results with AI-Generated Summaries](https://www.wortins.com/story/google-search-completely-replaces-results-with-ai-generated--96f32015)

_Source: Build Fast with AI · Monday, July 13, 2026_

Google has flipped the switch on the biggest change to search in a generation. AI Mode, powered by Gemini 3.5 Flash, now answers queries directly with generated summaries rather than the familiar list of blue links, and the company says more than a billion people a month are already using it across 98 languages and over 200 countries and territories. For readers this feels convenient, but for the web underneath it the shift is seismic. Search has quietly functioned as a deal for two decades: publishers give Google their pages, Google sends back traffic. When the answer lives inside the results page, that traffic can evaporate. Google is now telling sites to practice answer-engine optimization, tuning content to be quoted by a model rather than clicked by a human. The significance is less about any single feature and more about who controls discovery. If the front door to the internet becomes a summary, the businesses, blogs and newsrooms that depend on referral clicks have to rethink how they reach anyone at all.

[Read the full story at Build Fast with AI](https://www.buildfastwithai.com/blogs/ai-news-today-july-12-2026)

### [U.S. Lifts Advanced AI Chip Export Restrictions to UAE](https://www.wortins.com/story/u-s-lifts-advanced-ai-chip-export-restrictions-to-uae-142b8b77)

_Source: Build Fast with AI · Monday, July 13, 2026_

The Commerce Department has moved the United Arab Emirates into its A:5 tier, the category reserved for trusted destinations, which means advanced AI chips can now flow there without case-by-case export licenses. In practice that removes a major bottleneck for companies like Amazon, Apple and xAI that want to run routine hardware deals with Gulf partners. The change matters because compute has become geopolitics. For the past two years Washington has treated cutting-edge accelerators as strategic goods, rationing them to keep the most powerful chips away from China. Opening the taps to the UAE signals a bet that a friendly Gulf compute corridor is worth more than the caution of tight controls. It is also a gamble on trust and enforcement. Critics worry that chips landing in Abu Dhabi could be harder to track, while backers argue that anchoring the region's data centers to American silicon keeps it inside the U.S. technology orbit rather than pushing it toward Chinese alternatives.

[Read the full story at Build Fast with AI](https://www.buildfastwithai.com/blogs/ai-news-today-july-12-2026)

### [Anthropic Launches Claude Science: Workbench for Drug Discovery Research](https://www.wortins.com/story/anthropic-launches-claude-science-workbench-for-drug-discove-bd45c75d)

_Source: Anthropic · Monday, July 13, 2026_

Anthropic has released Claude Science, a dedicated workbench aimed at researchers rather than general users. It ships with more than 60 curated skills spanning genomics, proteomics, structural biology and cheminformatics, and it is bundled into every paid plan rather than sold as a pricey add-on. To seed real projects, the company is opening applications through July 15 for up to $30,000 in research credits. The interesting design choice is auditability. Scientific work lives or dies on reproducibility, so Claude Science produces artifacts with a full execution history and supports proper paper citations, letting a lab trace exactly how a result was reached instead of trusting a black box. That framing is the story. Plenty of tools now promise to accelerate discovery, but few are built around the boring, essential requirement that a peer reviewer can check the work. By targeting drug discovery and wet-lab adjacent fields with a reproducible, skills-based environment, Anthropic is trying to make its model a lab instrument, not just a chatbot.

[Read the full story at Anthropic](https://www.anthropic.com/news/claude-science-ai-workbench)

### [Mistral Releases Robostral Navigate: Single-Camera Robotics Navigation Model](https://www.wortins.com/story/mistral-releases-robostral-navigate-single-camera-robotics-n-9e1caede)

_Source: Mistral AI · Monday, July 13, 2026_

Mistral has released Robostral Navigate, an 8-billion-parameter model that lets a robot find its way using nothing but a single ordinary camera. No LiDAR, no depth sensors, just an RGB feed, which is the kind of hardware that costs a few dollars rather than a few thousand. On the R2R-CE benchmark of unseen environments it reaches a 76.6% success rate, about 9.7 points ahead of the best previous single-camera approaches. What makes it notable is how it was trained: entirely in simulation, across more than 400,000 generated trajectories, with no real-world data collection. The model is also hardware agnostic, meant to drive wheeled, legged or flying robots alike. The significance is about cost and access. Cheap, camera-only navigation lowers the price of building capable robots and pushes autonomy toward the long tail of makers and startups that cannot afford sensor-heavy stacks. It is a reminder that a lot of robotics progress now comes from better models, not more expensive sensors.

[Read the full story at Mistral AI](https://mistral.ai/news/robostral-navigate/)

### [SK Hynix Completes $26.5B IPO, Largest Foreign Listing in U.S. History](https://www.wortins.com/story/sk-hynix-completes-26-5b-ipo-largest-foreign-listing-in-u-s--ff97c203)

_Source: TechCrunch · Monday, July 13, 2026_

SK Hynix just pulled off the largest foreign IPO in U.S. history, raising $26.5 billion at $149 per American depositary receipt and edging past the record Alibaba set back in 2014. The listing is a milestone for a company most people outside the industry have never heard of, yet whose chips sit at the center of the AI boom. The reason is high-bandwidth memory, or HBM, the specialized stacks of DRAM that feed data to AI accelerators fast enough to keep them busy. SK Hynix controls roughly 60% of global HBM supply and reports a remarkable 72% operating margin on those chips, the highest profitability in the semiconductor sector. The huh factor here is where the money in AI actually pools. While attention fixates on model labs burning cash, a memory maker quietly became one of the most profitable and strategically vital firms in the chain. Proceeds will fund a new Korean fab and more advanced EUV capacity, deepening its grip on a scarce resource.

[Read the full story at TechCrunch](https://techcrunch.com/2026/07/10/sk-hynix-raises-26-5b-in-the-biggest-foreign-ipo-in-us-history-is-urged-to-build-new-us-fabs/)

### [FTC Releases AI Accuracy Policy Statement Addressing State Model Alteration Laws](https://www.wortins.com/story/ftc-releases-ai-accuracy-policy-statement-addressing-state-m-e7f1e31c)

_Source: Federal Trade Commission · Monday, July 13, 2026_

The Federal Trade Commission has put out a proposed policy statement wading into a genuinely thorny question: what happens when a state law tells an AI company to change the truthful output of its model. Directed by a December 2025 executive order, the statement is the agency's first major attempt to map the legal terrain around mandated output modification, and it is open for public comment through July 31. The tension is between two impulses. States have started passing rules that would require models to alter or suppress certain accurate answers, often on political or safety grounds, while the FTC has long treated deceptive or manipulated claims as a consumer-protection issue. Why it matters: this is the kind of quiet regulatory fight that will shape what chatbots are allowed to say. If forcing a model to distort accurate information collides with federal consumer-protection law, the outcome could set the boundaries for AI speech for years, well before Congress writes anything comprehensive.

[Read the full story at Federal Trade Commission](https://www.ftc.gov/news-events/news/press-releases/2026/07/ftc-seeks-public-comment-policy-statement-addressing-ai-accuracy)

### [Canary Speech Launches Ambient Behavioral Health AI for Telehealth Sessions](https://www.wortins.com/story/canary-speech-launches-ambient-behavioral-health-ai-for-tele-d6f3e18f)

_Source: MobiHealthNews · Monday, July 13, 2026_

Canary Speech has launched Canary Ambient, a platform that listens in on telehealth appointments and screens for signs of depression, anxiety and cognitive decline in real time. Rather than asking patients to fill out yet another questionnaire, it analyzes acoustic and linguistic patterns in how someone actually speaks during the visit, and it plugs directly into Zoom through the app marketplace so clinicians do not have to install anything new. The pitch is zero friction. The tool runs quietly alongside a HIPAA-compliant session, surfacing indicators a busy clinician might miss, while leaving diagnosis to the human on the call. It is a good example of applied AI creeping into everyday care, and it comes with real questions. Voice-based mental-health screening could catch problems earlier for people who would never seek help, but it also means an algorithm is quietly judging your tone during a doctor's visit. How accurate it is, and how patients are told it is running, will decide whether this feels helpful or intrusive.

[Read the full story at MobiHealthNews](https://www.mobihealthnews.com/news/microsoft-canary-speech-partner-advance-ai-enabled-speech-analysis)

### [University of Chicago Develops Cardiac Patch With 99.6% Arrhythmia Detection](https://www.wortins.com/story/university-of-chicago-develops-cardiac-patch-with-99-6-arrhy-e09013ee)

_Source: MarketScale · Monday, July 13, 2026_

Researchers at the University of Chicago have built a flexible cardiac patch that does something unusual: it runs the AI directly on the body. Instead of streaming heart data to the cloud, the patch performs inference on the device itself, detecting arrhythmias with a reported 99.6% accuracy and returning results in milliseconds. That on-device design is the whole point. For a cardiac emergency, the seconds lost sending data to a server and waiting for an answer can matter, and a patch that never depends on connectivity keeps working in an ambulance, a rural home or anywhere the signal drops. Flexible computing hardware makes it comfortable enough to actually wear. The significance is where clinical-grade monitoring is heading. Detection this accurate has traditionally lived inside hospital equipment, tethered to infrastructure. Pushing it into a thin, always-on wearable moves serious diagnostics to the point of care, and hints at a future where the early warning for a heart problem comes from a sticker on your chest rather than a visit to a clinic.

[Read the full story at MarketScale](https://www.marketscale.com/industries/healthcare/digital-healths-july-2026-signal-ai-wearables-a-new-cms-office-and-the-telehealth-billing-fight)

### [HHS Deploys ChatGPT for $2.1 Trillion Medicare/Medicaid Audit Program](https://www.wortins.com/story/hhs-deploys-chatgpt-for-2-1-trillion-medicare-medicaid-audit-dbf61c5d)

_Source: TechTimes · Monday, July 13, 2026_

The Department of Health and Human Services has quietly turned ChatGPT loose on one of the largest pools of money in government. Its new AERO program uses the model to audit roughly $2.1 trillion in Medicare and Medicaid spending, scanning five years of Single Audit Act compliance records across all 50 states and targeting any entity that receives more than $1 million a year in federal funds. The system flags chronic noncompliance and material weaknesses in near real time, and the stakes are steep: a flag can lead to frozen federal funding for hospitals, clinics and nonprofits that depend on it. This is exactly where applied AI gets uncomfortable. Using a language model to triage mountains of paperwork could genuinely catch fraud and waste that humans would never wade through. But an automated system deciding which providers face funding risk raises hard questions about error rates, appeals and oversight, and the reporting around AERO suggests those guardrails are far from settled. It is a preview of AI moving from advice to enforcement.

[Read the full story at TechTimes](https://www.techtimes.com/articles/319108/20260625/hhs-deploys-chatgpt-on-medicaid-audits-no-error-rate-no-appeals-no-deadline.htm)

### [Agility Robotics Files $2.5B SPAC for Digit Humanoid With $620M Proceeds](https://www.wortins.com/story/agility-robotics-files-2-5b-spac-for-digit-humanoid-with-620-dcd07bb2)

_Source: Robotics and Automation News · Monday, July 13, 2026_

Agility Robotics is heading to the public markets through a SPAC merger with Churchill Capital Corp XI that values the company at $2.5 billion. The deal is expected to bring in about $620 million in gross proceeds, including a $200 million private placement led by Foxconn, and to close before the end of 2026. Unlike a lot of humanoid hype, Agility has something to point to. Its Digit robot is already deployed in nine customer facilities and has logged more than 65,000 hours of operation, with over $300 million in multi-year orders for the next version tied to performance milestones and a pipeline of more than 30 companies. The significance is that warehouse humanoids are inching from demo videos toward real, financed businesses. Going public is a bet that legged robots doing repetitive logistics work can scale like a product rather than a science project, and the involvement of a manufacturing giant like Foxconn suggests the supply chain is starting to take that bet seriously too.

[Read the full story at Robotics and Automation News](https://roboticsandautomationnews.com/2026/07/07/agility-robotics-to-go-public-through-2-5-billion-spac-merger/)

### [Labcorp Launches MyLabcorp AI Health App for Lab Result Interpretation](https://www.wortins.com/story/labcorp-launches-mylabcorp-ai-health-app-for-lab-result-inte-c0bb029a)

_Source: PR Newswire · Monday, July 13, 2026_

Labcorp, one of the largest lab-testing companies in the country, has launched a consumer app called MyLabcorp that uses OpenAI reasoning models to translate cryptic blood-test results into plain language. Instead of staring at a table of numbers and reference ranges, users can ask conversational questions about cardiometabolic, kidney, mental-health and digestion markers, and watch how those readings trend over time. The company frames it as meeting demand: it cites surveys showing that more than half of consumers already see AI as important for making sense of their health data. The app is HIPAA compliant, available on iOS and Android, and its explanations are clinically reviewed rather than free-form generated. Why it matters: this is AI showing up at a very ordinary, anxious moment, the minutes after you get lab results and start googling. Done well, it could reduce needless panic and prompt useful follow-ups. Done badly, it risks overreassuring or alarming people about numbers that need a doctor's context, which is why the clinical review layer is the part to watch.

[Read the full story at PR Newswire](https://www.prnewswire.com/news-releases/labcorp-launches-mylabcorp-a-new-ai-powered-mobile-app-designed-to-help-consumers-understand-lab-results-and-track-health-trends-over-time-302777006.html)

### [Google Launches Africa Applied AI Lab in Accra for Startup Commercialization](https://www.wortins.com/story/google-launches-africa-applied-ai-lab-in-accra-for-startup-c-9fcb60b1)

_Source: Ecofin Agency · Monday, July 13, 2026_

Google is opening what it calls Africa's first Applied AI Lab, based at the Accra AI Community Centre in Ghana, with the explicit goal of helping African founders turn AI research into shipping products. Applications opened on July 1 and close August 31, with a development period running from mid-September into early December. The concrete perks are what make it interesting. Selected companies get early access to Google's Gemini, Gemma and Veo models, along with mentorship from Google Research and partner venture funds, focused on the future of work, knowledge, software, creativity and entertainment. The significance is about where AI capacity gets built. Most frontier tooling and talent still cluster in a handful of American and Chinese hubs, and programs like this are how that map slowly widens. There is a self-interested angle, since anchoring African startups to Google's models grows its ecosystem, but for founders on the continent, early access to frontier tools and capital is exactly the kind of leg up that has been hard to come by.

[Read the full story at Ecofin Agency](https://www.ecofinagency.com/news/0207-57009-google-to-launch-africas-first-applied-ai-lab-in-accra)

### [Meta Develops Custom Iris Chip Entering Production in September](https://www.wortins.com/story/meta-develops-custom-iris-chip-entering-production-in-septem-71ebbbc7)

_Source: Let's Data Science · Monday, July 13, 2026_

Meta's custom AI chip, code-named Iris, has cleared six weeks of testing and is set to enter production in September. Designed in partnership with Broadcom and fabricated by TSMC, the processor is aimed squarely at inference and ranking workloads, the everyday grunt work of deciding which ads and posts to show billions of users. The strategic thread is independence. Iris is part of a plan to double Meta's compute to 14 gigawatts by 2027, and building its own silicon chips away at the company's reliance on Nvidia and AMD, whose accelerators command steep prices and long waitlists. Controlling the chip means controlling cost and supply. Why it matters: the biggest AI spenders are increasingly deciding that renting compute from a single vendor is a strategic weakness. Meta joins Google and Amazon in designing accelerators tuned to its own workloads, a shift that, if it pays off, gradually loosens Nvidia's grip on the market and reshapes who captures the profits of the AI buildout.

[Read the full story at Let's Data Science](https://letsdatascience.com/blog/metas-iris-chip-enters-production-in-september-broadcom)

### [Microsoft Sales Agent Becomes Generally Available Across Outlook, Teams, Mobile](https://www.wortins.com/story/microsoft-sales-agent-becomes-generally-available-across-out-64deb62b)

_Source: Microsoft Learn · Monday, July 13, 2026_

Microsoft has moved its Sales Agent into general availability, embedding an autonomous Copilot directly inside Outlook, Teams and its mobile apps. Rather than a separate tool, the agent lives where salespeople already work, pulling in customer and deal intelligence, drafting personalized outreach emails and quietly updating CRM records as conversations happen. It works with both Microsoft's own Dynamics 365 Sales and, notably, rival Salesforce. Early adopters include Sandvik Coromant, which rolled it out across its entire sales organization in July, a sign the company is aiming at whole-team deployment rather than pilots. The significance is less about the feature list and more about the shape of it. This is one of the first mainstream examples of an AI agent handling a real revenue workflow end to end, not just answering questions but taking actions inside systems of record. If it sticks, it nudges the debate about AI and jobs from the abstract toward the concrete, since the busywork it automates, logging calls and chasing follow-ups, is a real slice of what sales roles do all day.

[Read the full story at Microsoft Learn](https://learn.microsoft.com/en-us/copilot/release-plan/2026wave1/copilot-sales/)

### [Anthropic Launches Claude Corps Fellowship for Nonprofit AI Professionals](https://www.wortins.com/story/anthropic-launches-claude-corps-fellowship-for-nonprofit-ai--a2f234a0)

_Source: Medium · Monday, July 13, 2026_

Anthropic is launching Claude Corps, a 12-month paid fellowship designed to bring AI skills into the nonprofit and social sector. The program comes with stipends, requires no college degree, and is explicitly aimed at practitioners and career-changers rather than credentialed engineers, with U.S. work authorization the main condition to apply. The framing is about who gets to use these tools well. Frontier AI has mostly been absorbed by well-funded tech companies, while the organizations working on housing, health and social services often lack the staff or budget to adopt it. Claude Corps is a bet that training people already embedded in that world is more useful than dropping software on them. Why it matters: it is a small program, but it points at a real gap. The value of AI depends heavily on people who know how to apply it to messy, real problems, and the nonprofit sector has been starved of that expertise. Whether a single fellowship moves the needle is unclear, but the framing of AI access as a workforce question is worth watching.

[Read the full story at Medium](https://medium.com/@davidakpovi/ai-news-week-of-july-6-to-july-12-2026-f81a26c49c55)

### [Apple Sues OpenAI for Trade Secret Theft](https://www.wortins.com/story/apple-sues-openai-for-trade-secret-theft-8bbd0d66)

_Source: TechCrunch · Monday, July 13, 2026_

Apple has taken OpenAI to court, accusing the ChatGPT maker of running a systematic campaign to lift its trade secrets. According to the suit, OpenAI's hardware chief Tang Tan, who spent 24 years at Apple and rose to vice president before leaving, directed staff to extract confidential information and even folded secret-sharing into the interview process. Apple points to former engineer Chang Liu, who it says walked off with a proprietary laptop, and claims OpenAI coached departing employees on how to slip past its security checks. The material at stake reportedly includes unreleased project code names, unannounced technologies, internal engineering presentations, and even proprietary metal-finishing methods. That last detail hints at what the fight is really about: OpenAI is building its first piece of hardware, and more than 400 former Apple employees now work there. Whether or not the specific claims hold up, the case captures a genuine tension in the industry. Talent moves fast between the biggest players, and when a software company reaches for physical devices, the line between hiring expertise and importing secrets gets tested in ways that courts, not engineers, end up settling.

[Read the full story at TechCrunch](https://techcrunch.com/2026/07/10/apple-sues-openai-over-alleged-trade-secret-theft/)

### [OpenAI Releases GPT-5.6 Three-Tier Model Family](https://www.wortins.com/story/openai-releases-gpt-5-6-three-tier-model-family-24f5a3d7)

_Source: OpenAI · Monday, July 13, 2026_

OpenAI has broadened its lineup with GPT-5.6, a family of three models aimed at different tradeoffs between smarts, speed, and price. Sol is the flagship for hard reasoning at $5 per million input tokens and $30 for output, Terra matches the older GPT-5.5 at roughly half the cost, and Luna is the cheap, fast option at $1 and $6. All three share a one-million-token context window, can return up to 128,000 tokens, and carry a February 2026 knowledge cutoff. The headline number is speed. Sol reportedly generates up to 750 tokens per second thanks to an integration with Cerebras, the wafer-scale chip maker, which is fast enough to make agentic and coding workflows feel far less sluggish. OpenAI ran a limited preview from June 26 with about 20 government-approved organizations before opening access more widely on July 9. The release underscores where the frontier race is heading. Raw capability alone no longer wins, and labs are now competing on how quickly and cheaply they can deliver it. Pairing a model with specialized inference hardware is becoming almost as important as the model itself.

[Read the full story at OpenAI](https://openai.com/index/previewing-gpt-5-6-sol/)

### [xAI Launches Grok 4.5 Coding and Agentic AI Model](https://www.wortins.com/story/xai-launches-grok-4-5-coding-and-agentic-ai-model-1f18f1c1)

_Source: xAI · Monday, July 13, 2026_

xAI has released Grok 4.5, Elon Musk's first model built specifically for coding and agentic work rather than general chat. Musk has pitched it as an 'Opus-class model, but faster, more token-efficient and lower cost,' priced at $2 per million input tokens and $6 for output. Under the hood sits a new 1.5 trillion parameter foundation the company calls V9, trained heavily on coding, science, engineering, and math, and drawing on datasets from Cursor, the AI coding startup xAI acquired. The distribution strategy is telling. Grok 4.5 arrives inside Grok Build, across all Cursor plans, and in the SpaceXAI console, with European availability expected in mid-July. That puts it directly in front of the developers most likely to judge a coding model on real work. The launch tightens an already crowded field. With OpenAI, Anthropic, and Google all pushing agentic coding assistants, xAI is betting that lower prices and its acquired coding expertise can carve out room. Positioning Grok as a 'truth-seeking' alternative is familiar branding, but the more concrete claim here is a cheaper, coding-native model built to plug straight into the tools developers already use.

[Read the full story at xAI](https://x.ai/news/grok-4-5)

### [Meta Removes Controversial Instagram AI Image Feature After Backlash](https://www.wortins.com/story/meta-removes-controversial-instagram-ai-image-feature-after--a62c6186)

_Source: TechCrunch · Monday, July 13, 2026_

Meta has pulled a new Instagram feature barely after introducing it, following a wave of pushback from users and creators. The tool let people @-mention public Instagram accounts inside image-generation prompts, effectively pulling someone else's likeness into an AI-made picture. Meta acknowledged the feature 'missed the mark' and removed the capability. The backlash centered on privacy and consent. Being able to summon a specific person's images into a generator raised obvious worries about unauthorized use of creators' and ordinary users' photos, both in the outputs and potentially in training. For a platform whose value rests on people sharing personal images, handing others an easy way to remix those images without permission was always going to be a hard sell. Notably, Meta did not retreat from AI imagery altogether. The feature was part of its broader Muse image model rollout, and the Muse tools stay in place, with only the direct account-referencing piece gone. The episode is a small but useful marker of where the limits sit: users seem open to generative features, but not to ones that let strangers point the camera at them.

[Read the full story at TechCrunch](https://techcrunch.com/2026/07/10/meta-removes-controversial-ai-feature-on-instagram-after-backlash/)

### [Federal Reserve Launches AI Task Force with Marc Andreessen as Co-Lead](https://www.wortins.com/story/federal-reserve-launches-ai-task-force-with-marc-andreessen--be60b6e4)

_Source: BuildFastWithAI · Monday, July 13, 2026_

The Federal Reserve is standing up a new task force to study how artificial intelligence is reshaping the economy, and it has handed a co-lead role to Marc Andreessen, the venture capitalist and co-founder of a16z. The group is charged with examining AI's effects on the financial system, jobs, productivity, and inequality. The appointment is striking. Andreessen is one of the technology industry's most vocal AI boosters, and putting a prominent investor at the head of a central bank study invites obvious questions about perspective and conflicts of interest. Supporters would argue that understanding AI's macroeconomic impact requires people who know the technology from the inside. Either way, the move signals that AI has climbed onto the agenda of the institution that steers interest rates and watches over employment. If the Fed starts factoring AI-driven productivity gains or labor disruption into how it reads the economy, the conclusions of a task force like this could eventually ripple into policy that touches far more than the tech sector. For now it is a study group, but the choice of who leads it already says something.

[Read the full story at BuildFastWithAI](https://www.buildfastwithai.com/blogs/ai-news-today-july-12-2026)

### [Perplexity Captures 7.67% of AI-Chatbot Referrals, Approaching Google's Share](https://www.wortins.com/story/perplexity-captures-7-67-of-ai-chatbot-referrals-approaching-c3089ba1)

_Source: Digital Applied · Monday, July 13, 2026_

Fresh Statcounter numbers for May put a real dent in the assumption that Google owns how people find things. Perplexity now leads the narrow but fast-growing category of AI-chatbot referrals at 7.67 percent, nudging just ahead of Google's own Gemini at 7.03 percent, with Copilot and Claude trailing. The catch is scale: traditional Google search still handles roughly 90 percent of all queries, so this is a fight over a small slice, not a coronation. What makes the figure worth a second look is the trajectory behind it. Perplexity's monthly users have grown about 370 percent year over year, nearly triple ChatGPT's pace, and its valuation has jumped from around 500 million dollars in 2024 to a reported 20 billion plus today. Google is not standing still: its AI Mode has crossed a billion monthly users and AI Overviews reach more than two billion. The interesting question is which scoreboard ends up mattering. Search share counts blue links served; answer share counts synthesized responses people actually read. Perplexity is winning the second game, and that is the one reshaping how the open web gets monetized.

[Read the full story at Digital Applied](https://www.digitalapplied.com/blog/ai-search-engine-statistics-2026-market-share)

### [AI Healthcare Diagnostics Achieve Breakthrough Accuracy Across Multiple Cancers](https://www.wortins.com/story/ai-healthcare-diagnostics-achieve-breakthrough-accuracy-acro-534d785b)

_Source: TechTimes · Monday, July 13, 2026_

A run of peer-reviewed studies this spring pushed medical AI past the demo stage and into genuinely useful diagnostic territory. One system correctly identified 18 distinct cancer types straight from tissue slides, while another flagged pancreatic cancer as much as three years before conventional methods would catch it, the kind of head start that changes survival odds. The FDA, meanwhile, has cleared dozens of AI diagnostic tools for primary care, enabling rapid screening for conditions from sepsis to depression through simple blood draws or voice samples. The appeal is not just accuracy but speed. Analyses that once took days now return in minutes, and the models can sift enormous datasets in seconds to surface patterns a rushed clinic might miss. Kaiser Permanente reported that AI scribes saved its physicians nearly 16,000 hours across 2.5 million patient encounters, time handed back to actual care. None of this replaces a doctor, and early detection still needs confirmation and treatment. But the direction is clear: AI is quietly becoming ordinary infrastructure in diagnosis, and the surprising consequences are showing up in outcomes rather than headlines.

[Read the full story at TechTimes](https://www.techtimes.com/articles/316451/20260509/biggest-ai-breakthroughs-healthcare-that-are-transforming-modern-medicine.htm)

### [EU AI Act Enforcement Rules Take Effect in August 2026, Global Regulation Accelerates](https://www.wortins.com/story/eu-ai-act-enforcement-rules-take-effect-in-august-2026-globa-1f31f4c1)

_Source: European Commission · Monday, July 13, 2026_

The European Union's AI Act stops being mostly theoretical in August, when the bulk of its obligations become enforceable and companies operating in the bloc have to show real compliance rather than good intentions. The rollout has been staged, but this is the phase with teeth, covering governance, transparency, and risk-management duties that touch any firm shipping AI into European markets. It is not happening in isolation. More than 70 countries have now floated over a thousand AI policy initiatives, and the pace is picking up on both sides of the Atlantic. In the United States, Colorado's AI Act took effect on June 30 with its own risk-management demands, California's SB 53 requires published safety frameworks, and a December executive order aims to consolidate federal oversight. Brussels is also rolling out a July action plan tying cybersecurity to advanced AI models. For builders, the takeaway is that 2026 is the year the regulatory patchwork gets real. Where a startup incorporates and ships now carries compliance weight it did not a year ago, and the quiet cost of that complexity tends to favor larger players.

[Read the full story at European Commission](https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai)

### [Cursor 3 Launches Parallel AI Agents Window for 30-50% Speed Gain on Coding](https://www.wortins.com/story/cursor-3-launches-parallel-ai-agents-window-for-30-50-speed--74984074)

_Source: Cursor Blog · Monday, July 13, 2026_

Cursor 3 rebuilds the AI coding editor around a simple but consequential idea: stop treating the assistant as one chat you wait on, and let several of them work at once. Its new Agents Window lets a developer hand plain-language tasks to as many as eight agents simultaneously, each running autonomously on its own isolated Git branch so their changes never collide. You describe the work, then review finished branches instead of babysitting a single thread. The payoff, per teams that have used it daily for more than six months, is a measured 30 to 50 percent speed gain on standard implementation tasks. Underpinning it is Cursor's in-house Composer 2 model, which the company says handles most routine work more cheaply than routing everything to frontier models, reserving the expensive calls for when they matter. It is a glimpse of where coding is heading: less pair programming with one bot, more orchestration of a small fleet. Cursor, now reportedly at 2 billion dollars in annual revenue, is betting that managing parallel agents becomes a core developer skill rather than a novelty.

[Read the full story at Cursor Blog](https://cursor.com/blog/cursor-3)

### [White House Nears Voluntary AI Release Standards with OpenAI, Google, Anthropic](https://www.wortins.com/story/white-house-nears-voluntary-ai-release-standards-with-openai-3f75e498)

_Source: CNBC · Monday, July 13, 2026_

The Trump administration is close to finalizing a voluntary framework with OpenAI, Google, and Anthropic that would insert the federal government into the frontier-model release process. Under the emerging deal, officials would get up to 30 days to review a powerful new model's national-security implications before it ships, alongside agreed benchmarks for cyber-capable systems and more predictable release timelines. The sticking point, unsurprisingly, is the threshold that triggers a review. The AI labs want it set high so only the most capable models qualify, while government negotiators push for a lower bar that would catch more releases. That single number decides how often Washington gets a look, and neither side wants to concede it. The context is pointed: the administration had already asked OpenAI to limit GPT-5.6 to vetted groups. Because it is voluntary, the framework leans on cooperation rather than statute, which makes it faster to stand up but easier to walk back. Still, it marks a real shift toward pre-release oversight of frontier AI in the US, a model other governments will study closely.

[Read the full story at CNBC](https://www.cnbc.com/2026/06/26/openai-limits-new-ai-models-to-trusted-partners-request-us-government.html)

### [California Governor Deploys Claude AI Across All State Agencies at 50% Discount](https://www.wortins.com/story/california-governor-deploys-claude-ai-across-all-state-agenc-678f8ae8)

_Source: California Governor · Monday, July 13, 2026_

California is putting its weight behind a single AI vendor at unusual scale. Governor Newsom announced a first-of-its-kind partnership that offers Anthropic's Claude to state agencies, plus cities and counties, at a 50 percent per-seat discount, bundled with free workforce training and hands-on technical help from Anthropic's own developers. It is being billed as the largest US state government deployment of a commercial AI assistant. The tool arrives through the state's technology department via its SITeS procurement portal, making Claude the first AI system approved for statewide use. This is not a pilot searching for a use case: agencies are already running it. The Department of Technology uses it for cybersecurity scanning and patching, the DMV has put it into customer service, and the Department of Health Care Services is applying it to Medicaid workflows. The significance is less about the technology than the precedent. When a state the size of California standardizes on one assistant and trains staff around it, that becomes a template other governments copy, and a very large, sticky customer base for whoever wins the contract.

[Read the full story at California Governor](https://www.gov.ca.gov/2026/06/29/governor-newsom-announces-a-first-of-its-kind-partnership-providing-anthropic-tools-to-state-agencies-and-improving-services-for-californians/)

### [Claude Code Adds Built-in Browser and AI Diagnostics with /doctor Command](https://www.wortins.com/story/claude-code-adds-built-in-browser-and-ai-diagnostics-with-do-79aba7a9)

_Source: Claude Code Changelog · Monday, July 13, 2026_

Anthropic's coding tool picked up two quietly useful additions in its early-July updates. The desktop app now ships with a built-in browser, so you can pull up documentation, designs, or live website content without leaving the editor and let the agent interact with what it sees. It folds real-time web access into the same surface where you write code, cutting the constant tab-switching that breaks flow. The other addition is a /doctor command that runs a health check on your setup. It diagnoses installation problems, flags skills and MCP servers you have configured but never use, and points out hooks that are slowing things down, the kind of accumulated cruft that quietly degrades a workflow. The same release also hardened the auto mode, which now blocks tampering with session transcript files and pauses to ask before running an rm -rf on an unresolved variable. Individually these are small, but together they signal where agentic coding tools are maturing: less about raw model power, more about the guardrails and diagnostics that make running an autonomous agent on your machine feel safe and maintainable.

[Read the full story at Claude Code Changelog](https://www.gradually.ai/en/changelogs/claude-code/)

## New AI Tools

### [Miora](https://www.wortins.com/story/miora-6a617900)

_Source: Product Hunt · Monday, July 13, 2026_

Miora is pitched as a creative studio in a box, built around the idea that one brief should be enough to produce a whole campaign. You describe what you want, and it orchestrates a set of AI specialists that each handle a different craft: script, storyboard, video, UI and UX, illustration, even 3D. Everything lands on a shared multimodal canvas where you can arrange, compare and version assets in one place. Two touches stand out. It exposes an editable memory so you can inspect and correct what the agents are reasoning about, which addresses the usual frustration of not knowing why a tool produced what it did. And it automatically maps brand guidelines across every output, so colors, tone and style stay consistent instead of drifting between generations. The appeal is for small teams and solo creators who need to move from concept to a rough but coherent set of deliverables fast. Whether the results are polished enough to ship or just strong starting points is the thing to test, but the single-brief-to-many-assets workflow is a genuinely useful shape.

[Read the full story at Product Hunt](https://www.producthunt.com/products/miora-2)

### [Mina Meeting Assistant](https://www.wortins.com/story/mina-meeting-assistant-1835bf1e)

_Source: Product Hunt · Monday, July 13, 2026_

Mina Meeting Assistant wants to be more than a note taker. It joins your calls on Zoom, Teams or Google Meet, transcribes them with speaker attribution, and then goes a step further by actually executing tasks and assigning action items while the meeting is still happening. Afterward it produces a summary that tracks the decisions made and the follow-ups owed. The distinction it is reaching for is between recording a meeting and participating in one. Plenty of tools will hand you a transcript, but Mina is positioned as an autonomous participant that turns discussion into logged tasks in your project tools without someone manually typing them up later. For anyone whose week is a wall of calls, the appeal is obvious: fewer dropped action items and less after-meeting admin. The risk is the usual one with autonomous agents, that it confidently captures or acts on the wrong thing, so how well it understands context and how easily you can correct it will matter more than how slick the summaries look.

[Read the full story at Product Hunt](https://www.producthunt.com/products/mina-meeting-assistant)

### [Cutrix](https://www.wortins.com/story/cutrix-06ff3129)

_Source: Product Hunt · Monday, July 13, 2026_

Cutrix tackles one of the more tedious problems in video: making a clip work in another language without losing the person on screen. Instead of flat subtitles or a generic dubbed voice, it translates the content while preserving the original speaker's voice, tone, and even facial expressions, so a translated video still feels like the same presenter talking. The pitch is mostly about time and money. Traditional localization means hiring voice actors or juggling subtitle files, and Cutrix folds translation, voice synthesis, and video editing into a single workflow aimed at creators, marketers, and anyone trying to reach audiences in more than one language. It sits squarely in a fast-moving category of AI voice and video tools, and the interesting bit is how convincingly the identity of the speaker survives the jump between languages. For a solo creator, that difference is what turns a novelty into something usable for real distribution.

[Read the full story at Product Hunt](https://www.producthunt.com/products/cutrix)

### [LemonLime](https://www.wortins.com/story/lemonlime-9275990a)

_Source: Product Hunt · Monday, July 13, 2026_

LemonLime aims to make workflow automation feel as simple as sending a message. You type a plain-English prompt, and it connects to your apps and carries out multi-step tasks, from data entry to notifications to cross-app integrations, without anyone writing custom code. The target user is clear: non-technical teams and business users who have repetitive processes but no engineer to wire them together. Tools like Zapier opened this door years ago, but they still ask you to build and maintain explicit chains of steps. LemonLime's bet is that a conversational layer can absorb that complexity and let people describe the outcome instead of assembling the plumbing. It landed near the top of Product Hunt on launch day, which says there is appetite for automation that meets people in ordinary language. The real test, as always with these tools, is reliability once the requests get messy, but the direction, describing work rather than scripting it, is where a lot of the category is heading.

[Read the full story at Product Hunt](https://www.producthunt.com/products/lemonlime)

## Interesting AI Articles

### [A Script for Mark Zuckerberg](https://www.wortins.com/story/a-script-for-mark-zuckerberg-0b039b59)

_Source: Stratechery · Monday, July 13, 2026_

This Stratechery piece lays out a strategic script for how Mark Zuckerberg can justify Meta's enormous AI spending, and the argument hinges on a well-timed accident of history. Meta ramped up GPU purchases in 2022, roughly a month before ChatGPT arrived, leaving it with capacity precisely when the world discovered it needed it. The core idea is that Meta already knows how to turn compute into money. When the company shifted to an algorithmic feed in 2018, investors panicked before that machine-learning bet drove years of growth. The essay argues the same playbook, using models to squeeze more value out of every pixel of advertising, can absorb the cost of today's infrastructure, with rented GPU capacity and clawback terms proposed as a fresh enterprise revenue line. It is worth reading because it reframes the capex anxiety hanging over big tech. Instead of asking whether AI spending is reckless, it asks what business model makes it rational, and for Meta the answer is the one it has always had: better targeting, more monetized attention.

[Read the full story at Stratechery](https://stratechery.com/2026/a-script-for-mark-zuckerberg/)

### [Why the Rise of Open Source AI Isn't Hurting Anthropic Yet](https://www.wortins.com/story/why-the-rise-of-open-source-ai-isn-t-hurting-anthropic-yet-d087ab8d)

_Source: TechCrunch · Monday, July 13, 2026_

This TechCrunch analysis tackles a puzzle: open-source models like DeepSeek are dramatically cheaper, so why has that not dented Anthropic's business. The price gap is not subtle. The piece cites DeepSeek at around $0.06 per million tokens against roughly $1.37 for Anthropic's Opus, and DeepSeek processing 5.3 trillion tokens a week to Opus's 2 trillion. The answer it offers is that frontier and open-source models are not really competing in the same moment. Frontier labs establish new capabilities and use cases, and open-source models commoditize them a phase later. That lets Anthropic keep more than half of enterprise AI spending even after raising prices, because companies pay a premium to be at the edge of what is possible. It is a useful frame because it complicates the popular story that cheap open models will inevitably eat the incumbents. The argument is that the two coexist, with the frontier constantly moving, though it also implies Anthropic's lead lasts only as long as it keeps defining what comes next.

[Read the full story at TechCrunch](https://techcrunch.com/2026/07/07/why-the-rise-of-open-source-ai-isnt-hurting-anthropic-yet/)

### [Sam Altman Seeks New World Order for AI as OpenAI Loses Ground to Anthropic](https://www.wortins.com/story/sam-altman-seeks-new-world-order-for-ai-as-openai-loses-grou-17c807c9)

_Source: Fortune · Monday, July 13, 2026_

Fortune examines an awkward moment for OpenAI: after years as the clear leader, it is watching Anthropic pull ahead. The piece points to Anthropic's projected $47 billion in annualized revenue against OpenAI's $25 to $33 billion, ChatGPT's market share slipping below 50% for the first time in May 2026, and Anthropic overtaking it in business subscriptions. Against that backdrop, the story argues, Sam Altman has pivoted toward calling for a new international framework to govern AI. The essay reads that turn with some skepticism, noting how conveniently a push for global rules and coordination arrives just as a competitor gains commercial ground. Why it is worth reading: it captures how quickly the AI hierarchy can shift, and how strategy follows position. Market leaders tend to favor open competition, while those under pressure often reach for governance and coordination. Whether Altman's proposals reflect genuine concern about AI risk or a bid to reset the board, the underlying signal is that OpenAI no longer sets the pace unchallenged.

[Read the full story at Fortune](https://fortune.com/2026/07/02/sam-altman-new-world-order-ai-openai-google-anthropic/)

### [The AI Funding Record Masks Two Very Different Markets](https://www.wortins.com/story/the-ai-funding-record-masks-two-very-different-markets-0345bee7)

_Source: SaaS Intelligence · Monday, July 13, 2026_

On paper, 2026 is a banner year for startups: global funding reached 510 billion dollars in the first half, and in the second quarter more than 70 percent of that went to AI-focused companies, up from under half a year earlier. Read quickly, it looks like a rising tide. Read closely, it describes two very different markets sharing one headline. The capital is intensely concentrated. Mega-rounds like OpenAI's 122 billion dollars and Together AI's 800 million pull the average up while the median startup sees little of it. Frontier labs and the infrastructure feeding them are awash in money, but bootstrapped and non-venture-backed AI companies, especially those building applications and niche tools, describe a funding drought rather than a boom. That split matters because it shapes what gets built. When capital pools at the model and compute layer, the application layer that turns those models into useful products gets starved of oxygen. The record number is real, but treating it as evidence of broad health misreads a market that is bifurcating, not lifting everyone at once.

[Read the full story at SaaS Intelligence](https://saasintelligence.substack.com/p/the-ai-funding-record-masks-two-very)

## AI Funding Tracker

### [Prime Intellect Raises $130M Series A at $1B Valuation](https://www.wortins.com/story/prime-intellect-raises-130m-series-a-at-1b-valuation-7d00ea27)

_Source: TechCrunch · Monday, July 13, 2026_

Prime Intellect has raised a $130 million Series A at a $1 billion valuation, led by Radical Ventures with participation from NVIDIA, Intel Capital and Dell Technologies. The round vaults the company into unicorn territory on the strength of real traction: it reports over $100 million in annualized revenue and around 6,000 customers, including well-known names like Ramp and Zapier. The product is a full-stack platform for building and training agentic AI systems, and the pitch is independence. Prime Intellect lets enterprises develop their own agents without leaning on a frontier lab's closed models, which is increasingly attractive to companies wary of depending on a single provider for something this strategic. The significance is where the money is flowing in this cycle. Investors are backing not just the labs that make models but the picks-and-shovels layer that helps everyone else build on top of them. A strategic chipmaker like NVIDIA joining the round underscores how valuable the tooling that keeps developers productive has become.

[Read the full story at TechCrunch](https://techcrunch.com/2026/07/08/prime-intellect-raises-130m-series-a-to-help-enterprises-build-their-own-ai-agents/)

### [TwelveLabs Raises $100M Series B for Video Intelligence Platform](https://www.wortins.com/story/twelvelabs-raises-100m-series-b-for-video-intelligence-platf-7d60b215)

_Source: GlobeNewswire · Monday, July 13, 2026_

TwelveLabs has raised a $100 million Series B co-led by NEA and NAVER Ventures, with Amazon participating, bringing the company's total funding to roughly $150 million. The startup is chasing what it calls video superintelligence, the ability for software to genuinely understand footage rather than just store and stream it. Its models are the pitch. Marengo 3.0 is built to understand sound, speech, motion and even emotion across the timeline of a video, while Pegasus 1.5 turns raw footage into structured, searchable data. Together they aim to make video as queryable as text, so you can find the exact moment something happens instead of scrubbing through hours of it. The significance is that video has been the hardest major modality to make useful at scale, precisely because it is so information dense. If TwelveLabs can make searching and analyzing it routine, it unlocks a long list of applications from media archives to security and sports, and the backing from NAVER and Amazon signals serious belief that this layer is worth owning.

[Read the full story at GlobeNewswire](https://www.globenewswire.com/news-release/2026/07/01/3320545/0/en/twelvelabs-raises-100-million-in-series-b-funding-to-build-video-superintelligence.html)

### [SambaNova Systems Closes $1B Series F at $11B Valuation](https://www.wortins.com/story/sambanova-systems-closes-1b-series-f-at-11b-valuation-c1a8e5a6)

_Source: Bloomberg · Monday, July 13, 2026_

SambaNova has completed the first close of a $1 billion Series F at an $11 billion post-money valuation, led by General Atlantic with a heavyweight roster that includes BlackRock, Intel Capital, the Qatar Investment Authority and T. Rowe Price. A second close is expected within weeks, so the final round size could climb higher. The company builds AI chips and systems aimed at inference, the ongoing cost of actually running models in production rather than training them once. That focus is timely: as businesses deploy AI at scale, the bill for inference is becoming the dominant expense, and SambaNova is positioning itself as an alternative to Nvidia for that workload. JPMorgan Chase has signed on as an inference deployment partner, a notable enterprise validation. Why it matters: the AI chip market is not a one-horse race, and investors are pouring money into challengers betting that specialized inference hardware can undercut general-purpose GPUs on cost and efficiency. Whether SambaNova can convert marquee partnerships into broad adoption is the open question this round is funding.

[Read the full story at Bloomberg](https://www.bloomberg.com/news/articles/2026-07-08/ai-chip-startup-sambanova-raises-funds-at-11-billion-valuation)

### [Clair Health Raises $11.6M Series A for Hormone-Monitoring Wearable](https://www.wortins.com/story/clair-health-raises-11-6m-series-a-for-hormone-monitoring-we-c5777f34)

_Source: TechCrunch · Monday, July 13, 2026_

Clair Health, founded by two Stanford graduates, has raised an $11.6 million Series A led by Khosla Ventures to build a noninvasive wearable for hormone tracking. The device combines 10 biosensors with AI voice analysis to follow more than 130 proprietary biomarkers, including estrogen, progesterone and other reproductive hormones that today usually require a blood draw to measure. The ambition is to make hormonal health continuously visible the way heart rate and sleep already are. The company says the wearable can read signals like inflammation, energy levels and cycle phase, and it is planning a consumer launch in November with a waitlist already past 25,000 people. The huh factor is the modality: pairing skin sensors with voice as a window into body chemistry is an unusual bet, and hormones have been a stubborn blind spot in the quantified-self boom. If the readings hold up to scrutiny, a category that has been underserved for years suddenly gets a mainstream consumer product, though noninvasive biomarker claims always deserve a healthy dose of clinical caution until validated.

[Read the full story at TechCrunch](https://techcrunch.com/2026/06/17/two-stanford-grads-raise-11m-to-build-a-noninvasive-wearable-for-hormone-tracking/)

### [Ollama Raises $65 Million Series B from Theory Ventures](https://www.wortins.com/story/ollama-raises-65-million-series-b-from-theory-ventures-5958108d)

_Source: TechCrunch · Monday, July 13, 2026_

Ollama, the tool that makes it easy to run large language models locally, has raised a $65 million Series B led by Theory Ventures, pushing its total funding to $88 million. The bigger story is adoption: the company says its developer base has grown to 8.9 million since January, with usage roughly doubling and about a million new installs landing every week. Its niche is running models on your own machine rather than through a cloud API, which matters most where data cannot leave the building. Ollama says it is now used inside 85% of the Fortune 500, including heavily regulated corners like government, healthcare, and finance, and it supports more than 67,000 integrations. That traction is a quiet counterpoint to the assumption that all AI runs through a handful of big cloud providers. Plenty of teams want capable models they control locally, and Ollama has become the default way to get there. The new money is earmarked for product work, community investment, scaling its cloud compute, and key hires.

[Read the full story at TechCrunch](https://techcrunch.com/2026/07/09/popular-open-source-ai-developer-tool-ollama-raises-65m-grows-to-nearly-9m-users/)

### [Norm AI Raises $120 Million Series C, Hits Unicorn Valuation](https://www.wortins.com/story/norm-ai-raises-120-million-series-c-hits-unicorn-valuation-1b87db12)

_Source: TechCrunch · Monday, July 13, 2026_

Norm AI has raised a $120 million Series C at a $1.2 billion valuation, crossing into unicorn territory just three years after launch. Khosla Ventures led the round, with Blackstone, Bain Capital, Coatue, and Vanguard joining, bringing the startup's total funding past $260 million. What sets Norm apart is its structure. Rather than sell software to existing firms, it operates Norm Law, LLP, an AI-native law firm where AI agents draft documents and handle legal tasks under the supervision of senior attorneys. It charges on an outcome basis instead of billable hours, a pointed departure from how the profession usually works. That puts it against rivals like Harvey and Legora, but on a different footing: competitors mostly build tools for lawyers, while Norm is trying to be the firm itself. Keeping human attorneys in the loop is both a safety measure and a regulatory necessity in a field built on professional accountability. Whether clients embrace an AI-first firm is the open question, but investors are clearly betting they will.

[Read the full story at TechCrunch](https://techcrunch.com/2026/07/07/ai-law-startup-norm-raises-120m-hits-unicorn-valuation/)

### [Even Realities Raises $150 Million, Hits $1B Valuation for Smart Glasses](https://www.wortins.com/story/even-realities-raises-150-million-hits-1b-valuation-for-smar-3a79e1a3)

_Source: TechCrunch · Monday, July 13, 2026_

Even Realities has raised $150 million and reached a $1 billion valuation for a deliberately contrarian take on smart glasses: no camera. The pre-Series B round was led by Chinese giant Meituan alongside returning investor Tencent, backing a team of ex-Apple engineers whose CEO worked on the Watch and iPhone. The design bet is the whole story. While most smart glasses race to add cameras and capture the world around the wearer, Even Realities beams information into the wearer's line of sight and leaves the camera out entirely, pitching privacy as the feature rather than an afterthought. In a category dogged by worries about being secretly recorded, glasses that cannot film anyone are a genuinely different proposition. The timing helps. IDC pegs smart glasses shipments at 2.25 million units in the first quarter of 2026, a 167% jump from a year earlier, so the market is expanding quickly. The funding is aimed at a next-generation platform, deeper AI integration, and global scaling for a device that competes as much on restraint as on features.

[Read the full story at TechCrunch](https://techcrunch.com/2026/07/06/smart-glasses-maker-even-realities-hits-1b-valuation-with-150m-funding-led-by-meituan-tencent/)

### [Quaise Energy Raises $134 Million Series B for Superhot Geothermal AI](https://www.wortins.com/story/quaise-energy-raises-134-million-series-b-for-superhot-geoth-c5e1c803)

_Source: Business Wire · Monday, July 13, 2026_

Quaise Energy has raised $134 million in the initial close of a Series B to attempt something audacious: drilling deep enough to tap superhot rock and build what it calls the world's first commercial superhot geothermal power plant. The round was led by Prelude Ventures, with Japanese energy heavyweights JERA and Idemitsu Kosan taking part, lifting the company's total raised to about $230 million. The core technology comes out of MIT and swaps conventional drill bits for millimeter-wave energy that can bore through rock too hot and hard for normal equipment. AI helps guide the system as it reaches temperatures of 300 to 500 degrees Celsius at depths beyond five kilometers, the conditions where geothermal energy becomes far more powerful and economical. The appeal is that superhot geothermal could deliver clean, always-on power almost anywhere with the right geology, not just volcanic hotspots. It is very much a deep-tech bet with hard engineering still ahead, but the funding, and the backing of major power companies, signals real confidence that AI-assisted drilling can crack a problem the industry has chased for decades.

[Read the full story at Business Wire](https://www.businesswire.com/news/home/20260707844187/en/Quaise-Energy-Raises-$134-Million-in-Initial-Close-of-Series-B-to-Build-Worlds-First-Superhot-Geothermal-Power-Plant/)

### [Venice AI Reaches $1B Valuation with $65M Series A](https://www.wortins.com/story/venice-ai-reaches-1b-valuation-with-65m-series-a-1c5ce357)

_Source: TechCrunch · Monday, July 13, 2026_

Venice AI crossed into unicorn territory with a 65 million dollar Series A at a 1 billion dollar valuation, led by Dragonfly with Coinbase Ventures and North Island Ventures participating. The draw is its privacy-first design: the platform offers access to more than 200 AI models while declining to store any user prompts or responses at all. That positioning is pulling real usage, more than 3 million active users and over 70 million dollars in annual recurring revenue, striking for a company founded only in 2024 by crypto entrepreneur Erik Voorhees and Jesse Proudman. The new money is earmarked for engineering hires, GPU compute, and zero-knowledge security work meant to make the no-logging promise verifiable rather than just stated. It is a reminder that a sizable slice of users will pay specifically for AI that forgets them, and that privacy can be a product category rather than a compliance checkbox.

[Read the full story at TechCrunch](https://techcrunch.com/2026/07/01/venice-ai-becomes-a-unicorn-with-65m-series-a-as-its-privacy-first-ai-platform-takes-off/)

### [Taktile Raises $110M Series C for AI Financial Decision Automation](https://www.wortins.com/story/taktile-raises-110m-series-c-for-ai-financial-decision-autom-dc596eaf)

_Source: Fortune · Monday, July 13, 2026_

Taktile raised a 110 million dollar Series C led by Goldman Sachs Growth Equity and Tiger Global, lifting its total funding to 184 million dollars. The startup builds an agentic decision platform aimed at the high-stakes, heavily regulated corners of banking and insurance, where a wrong automated call is expensive. The results it cites are what justify the round. Customers report 95 percent automation in B2B underwriting and 75 percent fewer false positives in anti-money-laundering checks, and one of the world's largest insurers projects more than 90 million dollars in savings on claims processing alone. The fresh capital funds product expansion and a new São Paulo office to reach financial institutions beyond its current markets. Taktile is a clean example of applied AI paying its way: not a chatbot demo, but automated decisions wired into workflows where the savings are measurable and the compliance stakes are real.

[Read the full story at Fortune](https://fortune.com/2026/06/24/exclusive-taktile-goldman-sachs-ai-bank-insurance-funding/)

### [Oxmiq Raises $35M Series A for Unified GPU/CPU/TPU Architecture](https://www.wortins.com/story/oxmiq-raises-35m-series-a-for-unified-gpu-cpu-tpu-architectu-7cf48458)

_Source: BusinessWire · Monday, July 13, 2026_

Oxmiq raised a 35 million dollar Series A co-led by Fundomo and Samsung Catalyst Fund, bringing its total to 60 million, to scale an unusual idea in AI hardware: a single licensable IP block that combines GPU, CPU, and TPU functions. Rather than build finished chips, Oxmiq wants to sell the architecture so semiconductor firms and cloud providers can design custom AI processors without shouldering the full cost of a system-on-chip from scratch. The names attached are the real signal. The company is led by Raja Koduri, former graphics chief at both AMD and Intel, and its board just added Jim Keller, the celebrated chip architect now running Tenstorrent, with an ex-Intel process-technology fellow joining as advisor. In a market where Nvidia's dominance has everyone hunting for alternatives, a licensable, mix-and-match architecture is a contrarian angle. Whether IP licensing can dent a world of vertically integrated giants is the open question, but the talent betting on it is hard to ignore.

[Read the full story at BusinessWire](https://www.businesswire.com/news/home/20260701241910/en/OXMIQ-Raises-$35-Million-to-Scale-OxCore-Architecture)

---

_Curated and written by [Wortins](https://www.wortins.com) — The daily AI briefing. Every story links to its original source; the "Wortins read" on each is our own original analysis. [About Wortins & our editorial approach](https://www.wortins.com/about)._