Thursday, 16 July 2026
TYLsemi raises $43M for AI chiplets
Archive item — written before sources were shown.
TYLsemi exited stealth with a $43M round led by Matter Venture Partners for a chiplet platform it says cuts custom AI chip development time in half.
Rime raises $24M, names chief scientist
Archive item — written before sources were shown.
Voice AI company Rime raised a $24M Series A led by M13 and hired Rafael Valle, ex-Meta Superintelligence Labs, as chief scientist.
Spectro Cloud raises $100M Series D
Archive item — written before sources were shown.
Spectro Cloud raised a $100M+ Series D led by Goldman Sachs Alternatives to grow Palette, its platform for running AI infrastructure in production.
SnapLogic launches SnapCode and MCP server
Archive item — written before sources were shown.
SnapLogic launched SnapCode and an MCP server giving Claude Code governed access to deploy and manage integrations across 1,000+ connectors.
InstaLILY raises $60M Series B for Lily
Archive item — written before sources were shown.
InstaLILY raised a $60M Series B and launched Lily, an AI forward-deployed engineer that builds agents inside legacy enterprise systems.
Copilot hardens security across surfaces
Archive item — written before sources were shown.
GitHub shipped MCP server trust checks, a plan-mode hard block on file edits, and BYOK support across Copilot's IDE, CLI, and app surfaces this week.
Valarian raises $50M for sovereign AI
Archive item — written before sources were shown.
Valarian raised a $50M Series A led by NEA, its first European defense investment, to grow ACRA, a control layer for sovereign AI infrastructure.
Claude Code fixes Slack approval spoofing
Archive item — written before sources were shown.
Claude Code 2.1.211 patches a bug where hidden Unicode characters could make a dangerous command look safe in chat-relayed approval prompts.
Oak raises $60M for AI identity control
Archive item — written before sources were shown.
Oak left stealth July 15 with a $60M seed round to build an AI-native identity operating system for human, machine, and AI-agent access.
Nvidia ships Cosmos 3 Edge, locks in Japan
Archive item — written before sources were shown.
Nvidia's 4B-parameter Cosmos 3 Edge world model ships for robots, alongside a Japan-wide physical AI push spanning Toyota, Mizuho, Canon, Fujifilm and RIKEN.
Moonshot launches Kimi K3, API-only for now
Archive item — written before sources were shown.
Moonshot AI's flagship Kimi K3 went live on its API July 16 with a 1M-token context window. Unlike its K2 predecessors, it shipped without open weights.
OpenAI, Anthropic split on AI regulation
Archive item — written before sources were shown.
OpenAI is pushing states to copy a common set of AI bills toward federal preemption. Anthropic is funding states to keep passing tougher ones.
Walden Robotics raises $300M at $1.1B
Archive item — written before sources were shown.
Toyota spinout Walden Robotics closed a $300M seed at a $1.1B valuation, co-led by Toyota and Deviation Capital, with Nvidia and Boeing joining.
Ollama pre-release improves Gemma 4 tools
Archive item — written before sources were shown.
Ollama v0.32.1, a pre-release, improves Gemma 4 tool-calling reliability and fixes a recurring MLX model cache memory leak.
Cline CLI and SDK add compaction progress
Archive item — written before sources were shown.
Cline CLI v3.0.41 and SDK v0.0.61 show live compaction progress in the terminal, persist workspace git info across sessions, and add Plan/Act mode guidance.
Codex CLI patch improves rm-command safety
Archive item — written before sources were shown.
OpenAI's Codex CLI 0.144.5 catches more forced-rm command variants and gives clearer rejection reasons when a dangerous command is denied.
AI access erodes willingness to say IDK
Archive item — written before sources were shown.
A 3,132-person study finds just having AI access nearly erased people's willingness to say I don't know, even when accuracy was financially rewarded.
EU AI Office publishes frontier AI findings
Archive item — written before sources were shown.
A 100+ expert forum tells the EU AI Office the next 1-2 years are decisive, flagging compute and energy as the top two-year priorities for frontier AI.
LangChain 1.3.14 adds ToolErrorMiddleware
Archive item — written before sources were shown.
LangChain 1.3.14 ships a new ToolErrorMiddleware and fixes ToolRetryMiddleware to only retry exceptions that are actually retryable.