AI Startups

Wrtn's $72M Round, Skild's Robots & Alice's $140M Bet: AI Startups to Watch for August 26, 2026

Wrtn's $72M raise, Skild's video-trained robots, a DeepMind spinout, and Alice's $140M safety bet—six startup moves worth your time today.

Daniyal

· 3 min read
Editorial image of a rocket silhouette lifting off past a circuit-board patterned skyline at dawn

Six AI startups moved the market in the last 24 hours: a Korean companion platform funding up for a 2028 IPO, a robotics model that learns from a single video, a DeepMind-spinout agent that claims to out-reproduce frontier models, an AI safety firm valued near a billion dollars, an Indian voice AI scaling past the phone call, and an audacious space data center bet. Here is who is disrupting what, and who stands to lose.

Wrtn (South Korea): companion AI with real revenue

Reuters reported Wrtn Technologies raised about 100 billion won ($72.2 million) in a Series C at a valuation above $722 million, bringing total funding to around $166 million. Coreline Ventures, Eugene Asset Management, Goodwater Capital, Antler, KDB and Capstone participated. The company says its North America entertainment platform OOC cleared $7.2 million in monthly revenue within three months of its May launch, and it expects 2026 revenue to top 200 billion won, up from 47.1 billion in 2025, as it prepares a 2028 IPO push.

Who it hurts: text and video companion apps that monetize through subscriptions. Wrtn is proving the category can convert hype into repeatable consumer revenue in under a year.

Skild AI: robots learn multi-minute tasks from one video

Skild AI released S1, a robotics foundation model that executes tasks up to 10 minutes long from a single human video prompt, with no task-specific fine-tuning. The company reports 66% success on unseen tasks versus 9% for language-prompted vision-language-action models at the same 100,000-hour training scale. Demos covered pancake flipping, pour-over coffee, plant potting and kit assembly. Sequoia's Alfred Lin called it a step toward a "ChatGPT moment" for humanoid robots.

Who it hurts: the companies that still teach robots one task at a time. If in-context learning from video holds, the marginal cost of a new robotic capability collapses.

Inherent (United Kingdom): a $50M bet on AI that reproduces science

TechCrunch reported London-based Inherent emerged from stealth with a $50 million seed round led by Index Ventures with Radical Ventures, founded by DeepMind alumni including Louis Kirsch, Kaloyan Aleksiev, Tantum Collins and chief scientist Edward Hughes. Its agent, Faraday, runs on a 27-billion-parameter Qwen base and calls OpenAI's GPT-5.5 Codex for coding subtasks. The company claims it outperforms Claude Opus 4.8 and GPT-5.5 at independently reproducing published scientific findings.

Who it hurts: the "large closed model as universal scientist" narrative. Inherent is arguing a small, purpose-built agent beats bigger general models on reproducible research.

Alice (Israel): $140M to stop AI from going rogue

Bloomberg reported Alice, an Israeli AI trust, safety and security company that works with Google and Anthropic, raised $140 million in a round led by Apax Digital. Globes said the round values the company near $800 million and brings total funding to $280 million. Its Rabbit Hole dataset compiles real-world adversarial and harmful content used to harden models against misuse, a market that has grown sharply after a spate of high-profile AI hacks.

Who it hurts: reactive security vendors. Enterprise trust budgets are moving toward model-native protections rather than perimeter tools.

Ringg (India): voice AI past the phone call

TechCrunch reported Bengaluru-based voice AI startup Ringg raised a $10 million Series A extension from Peak XV Partners. It already processes 20 million call attempts a month and is pushing voice AI beyond traditional phone systems into broader applications. The backing is a signal that investors are segmenting voice into separate markets instead of treating it as one call-center category.

Who it hurts: telephony-first voice vendors that have not expanded past outbound calling.

The pattern worth watching

What ties these six together is capital chasing a narrow, defensible slice rather than a general model. Companion revenue, one-video robot learning, reproducible science, model-native safety, non-phone voice and orbital compute all found willing backers in one news cycle. For founders, the lesson is granular: the funding window favors startups that can name the incumbent they displace and show a numbers-backed wedge.

Catch future funding and launch roundups under the AI startups tag on this blog.

Background worth having before you act on the above:

More on AI Startups

Browse every AI Startups story →