In partnership with

The Latest Agentic AI Development

💳 Google introduces Agent Payments Protocol (AP2)

Google just unveiled AP2 — the Agent Payments Protocol — an open, collaborative standard designed to let AI agents securely initiate and complete payments on behalf of users. It builds on existing protocols like Agent2Agent (A2A) and the Model Context Protocol (MCP). AP2 was developed with more than 60 partners spanning payments providers, merchants, fintechs, and crypto players.

Why it matters

• Building trust & reducing ambiguity: AP2 makes explicit what "authority" means when an agent acts for you — via mandates (Intent Mandates, Cart Mandates), which are cryptographically signed contracts defining what the agent is allowed to do. This helps verify user consent, reduce fraud, and clarify accountability if things go wrong.

• Payment-agnostic & future-ready: AP2 works across credit/debit cards, real-time bank transfers, and is designed to support stablecoins and other digital/crypto payment methods via an extension (x402). That means developers and businesses won’t be locked into one rail or payment provider.

• Enables new commerce flows: With AP2, you can imagine agents doing more than “alerting” you — agents might monitor product availability, execute purchases when thresholds are met (e.g. price drops), bundle items across merchants, or even handle multi-step procurement automatically, with user-defined constraints.

What to watch out for / challenges

• Adoption & ecosystem alignment: For AP2 to work, merchants, payment networks, agents, identity providers, and regulators all have to buy in. Many have committed, but integration work is still early.

• User experience, especially for “human-not-present” flows: Drafting Intent Mandates with enough specificity (price limits, timing, quality, etc.) without burdening the user is tricky. If mandates are too vague, risks increase; too specific, they can be pain to set up.

• Regulatory, privacy, and fraud risk: Cryptographic mandates and automated purchases raise questions around liability, especially across jurisdictions; also managing personal and payment data securely, preventing misuse.

🤖 Notion 3.0 Debuts AI Agents

Notion 3.0 has arrived, and its headline feature is Notion Agent — a built-in AI agent that can do everything a human can do inside Notion. That means building docs, creating databases, automating workflows, and working across hundreds of pages. Agents can operate autonomously for up to 20 minutes at a time. The update also introduces personalization (set your agent’s instructions, tone, memory of how you work) and previews of “Custom Agents” that run on schedules or triggers. New additions: row-level permissions in databases, expanded connectors & MCP integrations.

Why It Matters

• Massive productivity boost: Agents can handle complex, multi-step tasks (e.g. pull feedback from Slack/email, build summary reports, update databases) that used to require manual assembly. That means less busywork and more time for higher-value work.

• Context & memory built in: Your Agent remembers your style, preferences, how you organize materials — via instructions and memory pages. That means subsequent tasks are more aligned and require less correction.

• Scalability & flexibility: With the introduction of Custom Agents, teams will be able to spin up specialized agents (e.g. onboarding, knowledge base upkeep, project management) each with their own roles, workflows, triggers, and schedules. Useful for organizations that want automation across different domains.

What to Watch Out For

• Autonomy duration limits & oversight needed: While 20 minutes of autonomous work is powerful, some workflows require longer or more sustained context. Also, autonomous agents may make errors (mis-attributing context, permissions issues) if oversight is lax.

• Data privacy, permissions & tools integration: The agent draws from connected tools (Slack, email, etc.) and linked data. Ensuring correct permissions and managing what the agent can access will be critical. Also making sure the connectors work well and securely.

• Learning curve & customization effort: To get maximum value, users will need to spend time customizing agent instructions, memory, and maybe refining the agent’s behavior. Some may find initial setup overhead nontrivial. Also, depending on how well the models and integrations perform, there may be friction in edge cases.

❤️ Open Source Agents

• Teable 2.0: The Next Gen Airtable Alternative: No-Code Postgres

  

• Open Lovable: Clone and recreate any website as a modern React app in seconds

⚔️ AArena: All-in-One AI Workspace

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

TOP 5 AI (Sep 18) based on 5mln tokens used

1. Gemini 2.5 Flash-Lite

2. Grok 4

3. DeepSeek V3

4. Llama 3.3 70B

5. Kimi K2

👉 Enter the Arena
👉 Battleboard

✨ AI Agents Leaderboard

👉 Explore the Leaderboard

🗺️ AI Agents Landscape Map

We’re mapping thousands of AI Agents. Find your next AI agent today — explore thousands of options on our interactive Landscape Map.

👉 Explore the Map