• Agent Pulse
  • Posts
  • #33: OpenAI’s Agent Awakens, Amazon’s IDE Surfaces, and a European Unicorn is Born and more

#33: OpenAI’s Agent Awakens, Amazon’s IDE Surfaces, and a European Unicorn is Born and more

AgentPulse: Your Weekly Dose of AI Agents News.

Author

O P
July 18, 2025

In partnership with

Welcome back, AI Agent Enthusiast!

In today’s Agent Pulse:

  • 📢 Top Headlines

  • ⚔️ Agent Arena

  • ✨ Featured Agents

  • 🎓 Free courses

  • 🎓 Must Read Papers

📢 TOP Headlines

OpenAI ChatGPT Agents - The Quiet but Radical Shift

OpenAI’s agent rollout inside ChatGPT may seem subtle, but it’s the most important UI transformation since the original launch.

What changed:

  • You can now create persistent, autonomous agents inside ChatGPT - no external orchestration, no API juggling. Just assign it tasks, provide tools, and it executes.

  • These agents maintain memory, context, and can reason over time. They’re not just chatbots. They’re embedded, task-driven, decision-capable entities.

Why it matters:

  • This is OpenAI quietly converting ChatGPT into an operating system for agentic workflows.

  • The infrastructure is now primed for more than Q&A - it’s moving toward persistent digital workers, deeply integrated with OpenAI’s plugins, file handling, and user-specific goals.

The real shift:

  • It breaks the “prompt/response” mental model. You don’t just talk to it, you deploy it.

  • Developers, startups, and toolmakers will be tempted to build inside the ChatGPT ecosystem instead of launching standalone agents, risking platform dependency.

Takeaway:
If you’re building an AI product, you're no longer just competing with other SaaS startups, you’re competing with OpenAI’s growing internal platform and its ability to collapse full workflows into a single UI surface. Anyone building agent frameworks, orchestration layers, or AI frontends now has to ask: how will this survive if users default to ChatGPT-native agents?

Amazon’s KIRO IDE - Quietly Rewiring How Code Is Written

While the agent world obsesses over orchestration layers and memory systems, Amazon just introduced KIRO, a developer environment designed not for better code suggestions, but for integrated, autonomous code reasoning.

What KIRO brings:

  • A fully integrated AI IDE that observes, reasons, and adapts over time across entire codebases.

  • It's not just generating code it’s tracking intent, context, and developer habits, making it more like a resident AI software engineer than a glorified autocomplete.

What’s different:

  • Unlike Copilot, KIRO is deeply wired into AWS workflows. It’s designed to operate across cloud infrastructure, CI/CD systems, and secure environments OOTB.

  • It doesn’t just sit in your text editor, it becomes part of your DevOps muscle memory.

Why it’s strategic:

  • KIRO gives Amazon a bridge into the developer's day-to-day in a way CodeWhisperer never could.

  • It signals a larger shift from “assistive AI” to “situationally aware AI” agents that operate with continuity, not just reactive suggestions.

Takeaway:
KIRO may quietly become the most embedded AI system in enterprise software engineering because it’s built where code meets cloud, and where tools need context to be useful. While others chase agent frontends, Amazon is playing the backend AI infra game, where stickiness and scale are exponential.

Lovable: From Vibe Coding to Agent-Native App Factories

Lovable is Europe’s breakout AI platform, born in Stockholm, scaling like Silicon Valley. In under 12 months, it hit $75M ARR, 30,000 paying devs, and over 25,000 new AI-built apps per day. Now raising $200M at a $1.8B valuation, it's on track to become the Figma of agent-powered software creation.

What makes Lovable more than a no-code gimmick?

  1. Prompt → Production-Ready Stack
    Users describe an app in plain English. Lovable instantly delivers a full-stack output: React frontend, Supabase backend, authentication, and even Stripe for payments. It's not prototyping it’s deployable code with CI/CD pipelines wired in.

  2. Agent Mode: Code Reasoning on Autopilot
    The new Agent Mode doesn’t just generate it reads the codebase, pulls logs, diagnoses issues, and implements fixes. It's what AI pair programming should have been from the start: not chat, but commit-ready results.

  3. Social Remixability as Growth Flywheel
    Every app built can be browsed, cloned, and remixed publicly. That turns user output into viral acquisition loops. It’s not “community” as a forum, it’s GitHub + TikTok.

Lovable’s real edge isn’t UI polish, it’s the way it operationalizes agent autonomy without requiring users to understand agents. Agent Mode quietly bundles search, context gathering, doc scraping, and implementation steps into one clean UI. Users don’t configure workflows, they just describe goals. Behind the scenes, agents orchestrate everything from code diffing to feature delivery.

This makes Lovable one of the first true AI-native development environments, not just “AI-assisted.”

Learn AI in 5 minutes a day

This is the easiest way for a busy person wanting to learn AI in as little time as possible:

  1. Sign up for The Rundown AI newsletter

  2. They send you 5-minute email updates on the latest AI news and how to use it

  3. You learn how to become 2x more productive by leveraging AI

AArena: All-in-One AI Workspace

Get Sh*t Done with AI.
Bring your task. Get best results.

TOP 5 AI (July 17):

  1. Llama 3.3 70B Instruct

  2. ChatGPT-4o

  3. Llama 3.3 70B Instruct Turbo

  4. Llama 4 Maverick 17B

  5. DeepSeek R1 0528

✨ Featured Agents

Start learning AI in 2025

Keeping up with AI is hard – we get it!

That’s why over 1M professionals read Superhuman AI to stay ahead.

  • Get daily AI news, tools, and tutorials

  • Learn new AI skills you can use at work in 3 mins a day

  • Become 10X more productive

Other Headlines

  • Amazon Web Services announced Amazon Bedrock AgentCore – a new framework of services to securely deploy and operate scalable AI agents – and unveiled an “AI Agents and Tools” category in AWS Marketplace, along with a $100 million additional investment in its Generative AI Innovation Center (Source)

  • Microsoft revealed that “Custom Engine Agents” are now generally available in Microsoft 365 Copilot – enabling organizations to integrate their own AI agents (built with Copilot Studio, Azure AI Foundry, etc.) into Copilot workflows for tailored productivity automation (Source)

  • LangChain announced that its LangSmith (agent monitoring/evaluation) and LangGraph platforms are now listed in AWS Marketplace’s AI Agents and Tools category, allowing enterprises to easily procure and deploy these tools for building and scaling AI agents (Source)

  • Google rolled out an AI-powered business-calling feature in Search (US only) that lets Gemini call local businesses to retrieve pricing/availability on the user’s behalf, and upgraded its Search AI Mode with Gemini 2.5 Pro and a new Deep Search multi-query research tool that automates chains of searches into reports (Source)

  • Anthropic launched a “Connectors Directory” for its Claude AI, adding integrations to popular tools (Notion, Canva, Stripe, etc.) so Claude agents can securely access user data and apps when completing tasks (Source)

  • Anthropic introduced Claude for Financial Services, a packaged solution on AWS Marketplace combining Claude 4 models, Claude Code, and pre-built finance data connectors. It targets market analysis and investment workflows by giving enterprises an AI assistant with finance-specific knowledge and tools (Source)

  • Unify, a startup using AI agents to automate sales and go-to-market tasks, closed a $40 million Series B on July 17 (led by Battery Ventures with OpenAI Startup Fund, etc.). Its platform uses AI agents to surface high-intent leads, personalize outreach at scale, and automate growth workflows (Source)

  • Thinking Machines Lab – the AI startup led by ex-OpenAI CTO Mira Murati – announced a $2 billion funding round (at a $12 billion valuation) on July 15, led by Andreessen Horowitz (with Nvidia, Accel, etc.). The investment will accelerate development of its AI systems (including open-source components) for broader, safer AI applications (Source)

  • Barcelona-based Murphy, which builds AI agents for debt collection, raised $15 million on July 17 (led by Northzone with ElevenLabs, Seedcamp, etc.). Its platform uses OpenAI-powered chatbots that communicate with debtors via SMS, email, WhatsApp and phone to automate the collections process in traditionally analogue financial workflows (Source)

🎓 FREE Courses

  1. BeeAI: Agent Communication Protocol

  2. Anthropic: AI Fluence Course, designed for everyday users of AI.

  3. HuggingFace: Model Context Protocol (MCP)

  4. Microsoft: Generative AI for Beginners

  5. OpenAI: Advanced Prompt Engineering

🎓 Must Read Papers

  1. KPMG: AI Quarterly Pulse Survey: Q2 2025 (Doc)

  2. Stanford University: Future of Work with AI Agents (Doc)

  3. Google: Guide for using AI at work (Doc)

  4. Google: An Introduction to AI Agent Security (Doc)

  5. Thomson Reuters: Agentic AI 101 (Doc)

  6. OpenAI: A Practical Guide to building Agents (Doc)

  7. BCG: AI at Work (Doc)

  8. ServiceNow: Enterprise AI Maturity Index 2025 (Doc)

  9. IBM: Agentic AI in Financial Services (Doc)

THANK YOU

Visit website 

Follow us on Linkedin, X

I appreciate your time.

OP & Team

How'd we do?

Login or Subscribe to participate in polls.

Reach 23,000+ Readers:

Newsletter is read by VCs, founders, engineers, managers and tech professionals.

Reply

or to participate.