Welcome back! Whisker’s new $899 AI-powered litter robot promises to identify which cat used the bathroom, then track their habits for possible health insights. Sounds really cool! But things went a little awry in testing …  The bot apparently confused two cats so badly it thought they had swapped bodies. 

This just goes to show that AI can write code, generate movies, and summarize Supreme Court opinions, but put it in front of two cats and a poop drawer and suddenly we’re back to square one.

Matt, Catherine, and the Future Tools team

Could a $10,000 AI Degree Replace a $250,000 One?

You could soon land a job at Google, Microsoft, or McKinsey without going to a top university or taking on crushing student debt. All you'd need is $10,000 and a computer.

Sal Khan (of Khan Academy) is teaming up with TED and ETS (the nonprofit behind the GRE and TOEFL) to build the Khan TED Institute—a new type of college designed specifically for the AI era.

The key details:

  • Corporate partners are helping design the curriculum. Google, Microsoft, McKinsey, Bain, Accenture, and Replit are all involved, so the program actually reflects the skills employers want.

  • The curriculum blends traditional and practical. Math, statistics, computer science, writing, and economics—plus applied AI skills like building apps with AI, creating agents, financial modeling, and team-based projects.

  • The whole undergraduate degree is planned to cost under $10,000. For reference, Stanford tuition alone is heading toward $68,000 per year.

  • The format is nothing like a traditional college. No professors lecturing from an auditorium—students work through virtual lessons on their own, then come together for group projects to practice.

One big caveat worth flagging: the school is not yet accredited. Khan is pursuing it, applications are expected 12-18 months out, and the official launch is planned for 2027. Getting accredited can take years.

My POV: I love the idea behind this. The cost of higher education has become genuinely absurd, and the gap between what universities teach and what employers actually need has been growing for a while now. If Khan can pull this off—land accreditation, get the first cohort into real jobs at the partner companies, and prove out the model—this could genuinely change how a lot of people think about college.

Big if, though. Accreditation is not a footnote, and Khan's last big AI education bet (Khanmigo) landed with less of a splash than he hoped. But between the partners on board, the price point, and the moment we're in, this is worth paying attention to.

Hit reply and let me know, would you send your kid to Khan-TED over a traditional university?

— Matt

Nvidia Backs OpenAI’s Massive Ohio Data Center

Via Investopedia

Nvidia is investing $1.5 billion in SB Energy, the SoftBank-linked data center developer behind one of OpenAI’s biggest infrastructure projects.

Exclusive access: The investment gives Nvidia a major foothold in OpenAI’s Ports-Pike data center near Cincinnati, Ohio. Nvidia will be the exclusive compute infrastructure provider for the project, making sure the facility runs on its AI hardware.

The financing: Nvidia will also provide up to $105 billion in credit and compute support to help build the data center. The first phase is expected to support 4.25 gigawatts of computing capacity, with an option to add another 3.75 gigawatts. Capacity is expected to come online in phases starting in 2028.

OpenAI’s side: OpenAI gets long-term access to the compute it needs to keep scaling its models and products. The company has signed a 20-year lease for the data center, and OpenAI CEO Sam Altman was also an early investor in SB Energy.

Power problem: SB Energy will build and manage the site, which also includes plans for a massive natural gas power plant. The project could become one of the largest AI data center hubs announced so far. Still, it also comes as natural gas power plant costs are rising sharply, and data center demand is putting pressure on energy markets.

The bigger picture: Nvidia is no longer just selling the shovels in the AI gold rush. It is helping finance the mines. By backing one of OpenAI’s largest data center projects, Nvidia is making sure it stays embedded in the infrastructure layer of the AI economy. The models may get the spotlight, but the companies that control chips, financing, energy, and buildout capacity are shaping the race underneath.

Wispr Raises $280M at $2B Valuation

Wispr, the company best known for its AI dictation app called Wispr Flow, raised $280 million in Series B funding. The startup is looking to expand into productivity tools like meetings and new voice-first interfaces.

The round: The funding was led by Menlo Ventures and values Wispr at $2 billion. Existing investors—including Notable Capital, NEA, Neo Ventures, 8VC, and MVP Ventures—also participated, alongside new investors Acrew, Forerunner, Goodwater, Peak XV, Together Fund, and PLUS Capital.

The details: 

  • Wispr recently launched a meeting note-taker and has been exploring new human-computer interfaces through Wispr Interface Labs, led by Ariya Rastrow, who worked on Amazon Alexa in its early days.

  • Alongside the funding, Wispr announced a new speech-understanding model called Canto. The company says the model will reduce error rates from 30% to under 10%, following user complaints about a dip in dictation quality.

Crowded market: Wispr is facing competition on multiple fronts. Dictation apps like Willow, Monologue, Aqua, and Superwhisper are going after similar users, while its new meeting tool puts it closer to companies like Granola, Fireflies, and Read AI.

The bigger picture: Voice is becoming one of the main battlegrounds for AI productivity. Wispr’s challenge is proving it can be more than a very good dictation app. If it can turn speech into meetings, documents, workflows, and actions, it has a shot at becoming a bigger interface layer for work. If not, it risks getting squeezed in a category that is filling up fast.

The AI Design Tool That Actually Lets You Ship

Sick of AI design tools handing you a flattened image you can’t modify? Then you need Genspark Design—it lets you turn your rough ideas into editable websites, app screens, launch visuals, videos, and more.

With Genspark Design, you can:

  • Generate complete brand kits, mobile prototypes, and launch assets from one description.

  • Edit every element: copy, images, colors, spacing, and layouts.

  • Export a self-contained webpage, project package, or developer-ready handoff.

  • Create images, video, designs, research, and more inside one workspace.

Ready to skip the mockup graveyard? Try Genspark design for free through my link.

Run AI Agents Through OpenRouter

Via Ori Harness

Ori Harness is a CLI wrapper that lets developers run popular AI agent tools, including Claude Code, Codex, Hermes, OpenCode, and Pi, through OpenRouter without managing separate API keys. It uses OAuth sign-in to connect to an existing OpenRouter account and applies org-level controls across every agent.

How you can use it

  • Run multiple AI agent tools through OpenRouter

  • Avoid managing separate API keys for each agent

  • Apply model allowlists and spending caps across teams

  • Swap models on any agent with a single flag

  • Consolidate usage and costs into one OpenRouter bill

Pricing: Paid

An AI IDE With Built-In DevOps

Via lyzo

Lyzo is a standalone desktop IDE that combines AI code generation, persistent project memory, and built-in DevOps tooling in one workspace. It indexes files, functions, and dependencies to give AI agents full project context without repeated explanations.

How you can use it

  • Generate, review, test, and refactor code with AI agents

  • Keep persistent memory across a full codebase

  • Index files, functions, and dependencies for project context

  • Deploy with built-in CI/CD and Docker support

  • Work across 40+ programming languages

Pricing: Free and paid

Jobs, announcements, and big ideas

  • Anthropic reveals Claude designed protein binders for 14 of 15 targets, doubling industry hit rates.

  • Google expands Gemini in Chrome to all U.S. Android users, adding an Auto Browse feature.

  • Perplexity brings its Computer agent to email.

  • Stability AI ships Stable Audio 3.0 as a DAW plugin alongside an enhanced web app in beta.

  • OpenAI debuts ChatGPT for Teens with built-in safety protections and a dedicated study mode.

  • Cursor launches Origin, a code hosting platform with repos, pull requests, and GitHub sync.

These benchmarks are getting out of hand… Apparently the latest metric for models is committing cyber attacks? Follow along as I break down the madness!

That’s a wrap! See you Friday for more.