Deep Currents 06.06.25

Welcome to the latest instalment of Deep Currents, a monthly curated digest of breakthroughs, product updates, and helpful articles that have surfaced in the rapidly-evolving world of generative AI.

These were the things that stood out to me as impactful, as a design director in IT trying to stay on top of this rapidly evolving field. Hopefully this post will help you keep your head above water too.

Let's dive into this month's currents…

Design + Code

The convergence of design and development continues to accelerate, with several major announcements that blur the lines between creative vision and technical implementation.

• Figma went all-in on AI at their Config 2025 event, announcing a comprehensive suite of AI-powered tools including automated coding and website publishing, allowing for seamless Figma-to-working-prototype workflows. They've also integrated better image generation and AI editing tools, including OpenAI's gpt-image-1 and Google's Imagen 3. It's a clear signal that Figma sees AI as central to the future of digital product design.
• In what feels like a direct response, Google launched Stitch, a "vibe design" tool that not only generates frontend code from natural language prompts but also allows you to export directly to Figma. The tool represents Google's bet on conversational design interfaces becoming mainstream.
• Microsoft released NLWeb, an open-source project that transforms any website into an AI-powered app with natural language search capabilities. It also provides an MCP interface for AI agents to interact with your site, potentially revolutionizing how we think about web interfaces.

Frontier Models + LLMs

Another month, another round of leapfrogging between the major AI labs…

• Google briefly claimed the coding crown with the release of Gemini 2.5 Pro, which dominated benchmarks for a hot minute.
• Two weeks later, Anthropic struck back with Claude Opus 4 and Sonnet 4, reclaiming the top spot in coding performance. The rapid back-and-forth illustrates just how competitive this space has become.
• Mistral released Medium 3, positioning it as a high-performance model at significantly lower costs, alongside their new Le Chat Enterprise platform designed specifically for business environments.
• OpenAI rolled out several business-friendly updates to ChatGPT, including connectors for Google Workspace, Dropbox, SharePoint, and HubSpot, plus the ability to create custom connectors which enables direct integration with any MCP server. They also added a recording mode to the Mac desktop app, providing audio transcripts and synthesis directly within ChatGPT.

Coding Agents

The agent revolution is picking up steam, with several major players launching sophisticated coding assistants that can handle complex, multi-step tasks.

• OpenAI launched Codex, a cloud-based software engineering agent that can run multiple tasks in parallel. It's available on all paid accounts and represents a significant step toward autonomous development workflows.
• GitHub Copilot received an async agent upgrade, expanding its capabilities beyond simple code completion to more complex, multi-step tasks.
• Google introduced Jules, their first async coding agent, while Mistral launched their own vibe coding tool, showing that every major AI company now sees coding agents as essential.
• Perplexity Labs meanwhile now offers agentic productivity tools for creating dashboards, spreadsheets, storyboards, and mini-apps (available on the Pro plan).

Video Generation

The AI video space continues its explosive growth, with new models pushing the boundaries of quality and accessibility.

• Lightricks unveiled LTXV-13B, an open-source AI video generation model that creates high-quality content 30x faster than existing models while being efficient enough to run on consumer hardware.
• Google's Veo 3 can now generate incredibly realistic videos with synchronized audio. Early experiments have caused a stir across social media.
• Flow is Google's new AI filmmaking tool that leverages the Veo 3 model. Flow lets users create cinematic clips and stories using natural language, making professional-quality video creation more accessible.
• HeyGen rolled out Avatar IV, capable of creating lifelike animations from a single photo while capturing vocal nuances, natural gestures, and facial movements, along with the AI Studio video creator platform.

Voice + Audio

Voice technology saw significant advances this month, with improvements in both quality and functionality.

• NVIDIA launched Parakeet 2 (officially Parakeet-TDT-0.6b-V2), their second-generation automatic speech recognition model that promises absurdly fast processing speeds.
• Hume introduced EVI 3, their third-generation speech-language model with improved conversational capabilities.
• Resemble.ai released Chatterbox, an open-source voice model that can clone voices with just 5 seconds of audio.
• ElevenLabs made a couple announcements: Conversation AI 2.0 with state-of-the-art turn-taking models and HIPAA compliance, plus the SB1 custom soundboard for creating custom sound effects.

Images

Visual generation tools continue to evolve... in particular advances are happening in style control and reference consistency.

• Recraft launched Advanced Style Control for better brand consistency, and added support for external image models including Flux, Imagen 4, and Imagen 4 Ultra.
• Black Forest Labs launched FLUX.1 Kontext in both pro and max versions, continuing to push the boundaries of image generation quality with advanced character and object consistency, and omni reference editing capabilities.

Learning + Education

The AI education space is maturing, with major AI companies offering structured learning paths.

• Anthropic launched a free AI fundamentals course, providing a solid entry point for those looking to get up to speed or refresh their knowledge.
• OpenAI Academy offers free educational content and live events, though it seems to be flying under the radar, with only one "community" established so far in their Communities section, and it's for India.

Interesting Articles

Several thought-provoking pieces explored how AI is reshaping entire industries and creative practices:

• Product Development is being fundamentally transformed, with three articles in the past month exploring different aspects of this shift: The Emergent Future of Product Development, The Death of Product Development, and The End of Human Product Management.
• AI Art gets a thoughtful examination in this IEEE Spectrum piece about its place in institutions like the Whitney Museum.
• Workplace AI adoption gets a practical framework in this post about making AI work in leadership and organizational contexts.

What This All Means

This month's updates reveal several important trends: the rapid commoditization of AI capabilities across different domains, the emergence of sophisticated agent workflows, and the continued blurring of lines between traditional creative and technical roles. The speed of iteration between major releases suggests we're still in the early acceleration phase of this transformation.

The focus on business integration tools from OpenAI and others indicates the technology is moving beyond experimental use cases toward mission-critical applications. Meanwhile, the explosion of coding agents suggests that software development workflows are about to undergo fundamental changes.

Most importantly, the diversity of tools launching simultaneously—from voice cloning to video generation to design automation—suggests we're approaching a convergence point where AI capabilities become ubiquitous across creative and technical workflows, rapidly transforming entire industries.

Well, that's enough for one month! Let me know what's resonated with you lately, either in the comments, or send me an email. I'd love to hear from you.

Cover image generated with Midjourney 7. Editing assistance provided by Claude Sonnet 4.