PLUS: An AI civil war erupts in China, and a new ranking of enterprise voice AI agents

It's a new day AI Rockstars!

YC S23 startup Morph is introducing a new API that edits code at a blistering 4,500 tokens per second. This approach moves away from slow, full-file rewrites, offering developers a faster and more precise way to implement AI-generated changes.

This represents a move toward smaller, purpose-built AI tools that plug directly into a developer's existing process. As these hyper-fast, specialized models become more common, how will they change the expectations for AI-assisted coding?

In today’s AI recap:

• Morph’s new API for hyper-fast code editing
• Retell AI’s new ranking of enterprise voice agents
• A public feud between China’s top AI companies
• Groq’s first European data center for low-latency AI

Morph's AI Edits Code at 4,500 Tokens/Second

The Report: YC S23 startup Morph launched an API that applies AI-generated code edits at over 4,500 tokens per second. The tool is designed to replace slow full-file rewrites and make AI coding agents faster and more reliable for developers.

Broaden your horizons:

• Morph uses a “Fast Apply” model that intelligently patches changes into existing code, avoiding the brittle and error-prone process of rewriting an entire file.
• The underlying technique was pioneered by tools like Cursor but wasn’t available as an API, which Morph now makes accessible to developers through a live demo and a generous free tier.
• The company is also working on a “Next Edit Prediction” model with its Morph Tab API, which aims to predict your next code edit with sub-500ms latency, and you can request early access now.

If you remember one thing: This signals a shift toward specialized, hyper-fast AI tools that integrate seamlessly into developer workflows. Instead of waiting for large models to rewrite entire files, developers can get precise, reliable edits instantly.

The Voice AI Landscape

The Report: Lean AI company Retell AI has published a market overview of the top enterprise AI voice agents, ranking itself #1 while outlining the key criteria for high-stakes deployments. The report provides a framework for evaluating solutions beyond just features.

Broaden your horizons:

• The push for enterprise voice AI is accelerating, with 85% of customer service leaders now exploring GenAI solutions to capitalize on a market projected to hit $103.6B by 2032.
• True enterprise-grade solutions are defined by more than just conversational ability; they must provide 99.99% uptime guarantees, deep CRM integrations, and comprehensive security and compliance like SOC 2 and PCI.
• This trend extends beyond customer service, as agentic AI is predicted to autonomously make 15% of daily work decisions by 2028, fundamentally altering business operations.

If you remember one thing:
The best AI voice agents are shifting from being simple cost-saving tools to becoming core brand ambassadors that transform the customer experience. The critical evaluation metric is moving from a list of features to the total cost of ownership and a clear, rapid return on investment.

China's AI Civil War

The Report: A public dispute has erupted between two of China’s tech giants after an anonymous group accused Huawei of cloning an open-source AI model from Alibaba. The accusation has sparked a rare public feud, raising questions about IP and trust in the country's AI ecosystem.

Broaden your horizons:

• An anonymous group called HonestAGI published a technical analysis claiming Huawei's Pangu model showed an extraordinary correlation with Alibaba’s open-source Qwen model.
• Huawei’s Noah Ark Lab swiftly denied the claims, stating its model was built from scratch on its own Ascend chips and included its own key innovations.
• The public clash fractures AI unity in the country, highlighting how intense competition is damaging global trust in open-source collaboration.

If you remember one thing: This feud exposes the immense pressure on tech companies to keep pace in the global AI race, sometimes at the expense of transparency. The incident raises critical questions about intellectual property, potentially shaping how open-source AI development is governed worldwide.

Groq's European Push

The Report: AI chip startup Groq is launching its first European data center in Helsinki, Finland. This expansion aims to deliver its high-speed AI inference services with lower latency to customers across Europe, the Middle East, and Africa.

Broaden your horizons:

• The new data center gives European customers access to infrastructure that provides the lowest latency possible for their AI workloads, boosting performance and response times.
• The move is a collaboration with digital infrastructure company Equinix, allowing customers on the Equinix Fabric network to directly connect their workloads to the GroqCloud platform.
• Backed by investment from major tech players like Samsung and Cisco, this expansion signals Groq's aggressive push to scale its global AI infrastructure.

If you remember one thing: Groq is strategically placing its powerful hardware closer to its international users to reduce speed barriers. This move makes building and deploying ultra-fast AI applications a more practical reality for developers and businesses in Europe.

The Shortlist

Golioth introduced an experimental open-source project, tinymcp, that allows LLMs to control embedded IoT devices using the Model Context Protocol.

AI-Docs launched as a new CLI tool to help developers manage AI-generated "memory" files like CLAUDE.md by isolating them on a dedicated Git branch with a worktree.

Gamma added a new feature that lets users animate any image directly within its AI presentation platform, adding subtle motion to slides with a single click.