RuntimeWire

Real-time startup intelligence for the AI economy. Funding, model launches, infra shifts, and founder moves — the signals that matter.

Latest stories — page 58

• OpenAI puts GPT-Live into ChatGPT Voice

  The new full-duplex voice models can listen while speaking, delegate harder work to GPT-5.5 and will come to the API later.

• Katalyst's LINK spacecraft is chasing NASA's Swift in a test of satellite rescue as a business

  Ghonhee Lee's space robotics company built and launched LINK in under a year after NASA awarded the mission.

• Bespoke Labs raises $40M for AI agent training environments

  Mahesh Sathiamoorthy and Alex Dimakis are betting that agent reliability will be won in simulated workplaces before live deployment.

• Perception AI raised $11 million and is shipping a 22-DOF robotic hand

  The robotics company is entering a crowded humanoid manipulation race with one verified claim: a many-jointed hand has moved from demo to shipment.

• SonarSource study finds clean code cuts coding-agent token use, not pass rates

  The May 19th arXiv preprint tested Claude Code on paired repositories and found fewer tokens and file revisits on cleaner code.

• Techstars and Emirates NBD make enterprise banking access the pitch for AI founders

  The July 6th partnership gives selected AI and FinTech teams a path into Emirates NBD pilots, with no disclosed capital commitment.

• Kyrall pitches prompt-driven CAD for engineers who still need editable models

  The startup says users can turn text, sketches, meshes and files into parametric CAD, with pricing that starts at 10 free generations.

• Castro's Dustin Bluck says personal support did not buy customer loyalty

  The indie app operator's July 3rd essay turns two years of subscription support into a sharper product lesson for small software teams.

• Herdr turns the terminal into an agent control room

  Can Celik's Rust multiplexer keeps Claude Code, Codex and other coding agents alive across panes, SSH sessions and laptop closes.

• CERT/CC warns Tenda router firmware has an undocumented admin backdoor

  CERT/CC says five firmware builds accept any username if a hidden alternate password matches a device configuration value.

• Agent Gateways Are Turning Into Enterprise AI's New Control Layer

  Arcade, Manufact and Nutanix are converging on the same enterprise problem: who controls the tools, models and spend behind AI agents.

• Andra Keay's June robotics ledger turns funding into physical AI's scorecard

  Robots & Startups counted 136 June rounds worth $6.57 billion, with defense dollars and late-stage checks distorting the readout.

• Runway opens Paris research hub with a $30 million France bet

  The France office starts with 10 employees and puts Runway closer to European AI researchers working on world models and physical AI.

• Nebula Security publishes GhostLock exploit for 15-year Linux kernel flaw

  The AI security lab says the bug powered an IonStack chain and earned a $92,337 kernelCTF reward from Google.

• Sara Wilczynska left Google and turned a Thailand watercolor habit into Swil Arts

  The former software engineer built a San Diego illustration studio after a 2023 travel year showed her a different way to work.

• Deyan Ginev's ar5iv Is Still Chasing arXiv's Paper Firehose

  A June 6th status post puts April 2026 papers in ar5iv, with the harder claims around speed and AI-assisted development still unbenchmarked.

• OpenComputer turns Slack threads into durable agent sessions

  The agent infrastructure product is pushing Slack from notification surface to shared runtime control plane for long-running AI work.

• $150 and an afternoon for a multiplayer shooter is the AI coding story

  A $150 Claude Fable 5 run produced a multiplayer capture-the-flag prototype in an afternoon, turning an AI coding demo into a cost-of-software question.

• DeepSeek is developing its own AI inference chip, Reuters reports

  Liang Wenfeng's AI lab is moving toward hardware as export controls and inference costs reshape China's model race.

• Figma buys Bud team to pull AI agents closer to the design canvas

  Bud, formerly Orchids, will shut down its products by July 18th as Kevin Lu's YC-backed team moves into Figma.

• comma.ai bans Anthropic as an AI vendor after CTO warns of cloud token dependence

  Harald Schaefer told employees to stop sending company spend to Anthropic and invest instead in comma.ai's own inference stack.

• OpenAI backs EU AI content labeling code as August deadline closes in

  The June 11th endorsement ties OpenAI's provenance work to a voluntary path for meeting EU AI Act transparency rules.

• Google adds background tasks and remote MCP to Gemini Managed Agents

  The July 7th update moves Gemini's agent API closer to a hosted worker layer for long-running, tool-heavy jobs.

• Meta's Muse Image gives Wang's superintelligence lab its first image model

  Muse Image plans, searches the web and edits across turns inside Meta AI, Instagram Stories and WhatsApp chats.