[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-coinbase-for-agents-turns-ai-into-trader-en":3,"article-related-coinbase-for-agents-turns-ai-into-trader-en":30,"series-tools-cd263b39-9c16-417c-916d-5eb2848e9867":82},{"id":4,"slug":5,"title":6,"content":7,"summary":8,"source":9,"source_url":10,"author":11,"image_url":12,"cover_image":12,"category":13,"language":14,"translated_content":11,"related_article_id":15,"keywords":16,"key_takeaways":22,"views":26,"created_at":27,"published_at":28,"topic_cluster_id":29},"cd263b39-9c16-417c-916d-5eb2848e9867","coinbase-for-agents-turns-ai-into-trader-en","Coinbase for Agents turns AI into a trader","\u003Cp data-speakable=\"summary\">I break down Coinbase for Agents and x402 into a copy-ready agent payments template.\u003C\u002Fp>\u003Cp>I've been watching \u003Ca href=\"\u002Ftag\u002Fai-agent\">AI agent\u003C\u002Fa> demos for a while now, and most of them feel like they stop right before the annoying part. They can summarize a market, draft a plan, maybe even suggest a trade. Then you hit the wall: log in yourself, approve the payment yourself, copy the ticket yourself, babysit the whole thing yourself. It's all theater until money moves.\u003C\u002Fp>\u003Cp>That gap has been bugging me. Not because I want machines freelancing with my portfolio like they're day-trading in a basement. I just hate workflows that pretend autonomy exists and then quietly hand the hard part back to a human. If an agent can read research, compare options, and tell me what it would do, why am I still the one clicking through logins, subscriptions, and checkout forms? That's not an agent. That's a fancy autocomplete with confidence issues.\u003C\u002Fp>\u003Cp>So when Coinbase announced \u003Ca href=\"https:\u002F\u002Fwww.cnbc.com\u002F2026\u002F06\u002F11\u002Fcoinbase-launches-tool-to-let-ai-agents-manage-trading-and-payments.html\">Coinbase for Agents\u003C\u002Fa>, I paid attention. Not because I think every assistant should get a wallet tomorrow. I paid attention because this is one of the first mainstream attempts to wire \u003Ca href=\"\u002Ftag\u002Fai-agents\">AI agents\u003C\u002Fa> into actual financial rails instead of just letting them talk about money. And that changes the shape of the product, the risk, and the whole integration story.\u003C\u002Fp>\u003Cp>Coinbase is not pitching a chatbot. It is pitching a system where an agent can trade, pay for data, buy compute, and keep moving without a human in the loop every five minutes. That is a much messier idea, and honestly, a much more interesting one.\u003C\u002Fp>\u003Cp>Source anchor: Tanaya Macheel’s CNBC report on Coinbase’s launch is the trigger here, and the original article is here: \u003Ca href=\"https:\u002F\u002Fwww.cnbc.com\u002F2026\u002F06\u002F11\u002Fcoinbase-launches-tool-to-let-ai-agents-manage-trading-and-payments.html\">CNBC\u003C\u002Fa>. The piece says Coinbase for Agents starts with \u003Ca href=\"\u002Ftag\u002Fchatgpt\">ChatGPT\u003C\u002Fa> and \u003Ca href=\"\u002Ftag\u002Fclaude\">Claude\u003C\u002Fa>, and it also points to x402, Coinbase’s machine-to-machine payments protocol.\u003C\u002Fp>\u003Ch2>Coinbase is trying to make agents feel like account holders\u003C\u002Fh2>\u003Cblockquote>\"The whole idea is to give agents access to money and, through that financial independence, improve their set of capabilities to pretty much anything on the internet,\" Lincoln Murr, Coinbase's AI product lead, told CNBC.\u003C\u002Fblockquote>\u003Cp>What this actually means is simple: Coinbase wants an AI agent to behave less like a helper and more like a limited financial actor. Not a full person, obviously. More like a software process with enough permissions to do useful work without turning every action into a human approval ceremony.\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781407993272-r5np.png\" alt=\"Coinbase for Agents turns AI into a trader\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>I think that distinction matters because most agent products today are built around conversation, not authority. They can recommend. They can draft. They can simulate action. But if they cannot actually spend money or settle a transaction, they are still stuck in demo land. Coinbase is trying to move the boundary from \"suggests\" to \"executes.\"\u003C\u002Fp>\u003Cp>The CNBC piece says Coinbase for Agents initially lets agents like ChatGPT or Claude execute crypto trades using natural language instructions. That means a user can ask an agent to rebalance a portfolio, spot a trade, or manage positions over time. The company says it will eventually expand to stocks and predictions too. That expansion matters because it shows this is not just a crypto toy. It is Coinbase trying to become the default financial plumbing for agent workflows.\u003C\u002Fp>\u003Cp>I ran into the same pattern building internal automation for a trading workflow a while back. The bot could analyze signals and prepare an order, but the approval step killed the momentum. Every extra click created friction, and friction killed usage. If Coinbase can collapse that gap, they will own a very valuable layer: the permissioned execution layer between intent and transaction.\u003C\u002Fp>\u003Cp>How to apply it: if you are building agent software, stop thinking in terms of \"can it answer?\" and start thinking in terms of \"what can it legally and safely do?\" Define the smallest action set that still makes the agent useful. For a trading agent, that might be read-only research first, then draft orders, then capped execution, then timed execution. Do not jump straight to full autonomy unless you enjoy incident reports.\u003C\u002Fp>\u003Cul>\u003Cli>Separate recommendation from execution.\u003C\u002Fli>\u003Cli>Use narrow scopes and hard limits for spend, asset class, and time window.\u003C\u002Fli>\u003Cli>Log every agent action like it will be audited, because it probably will be.\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>x402 is the real product, not the headline\u003C\u002Fh2>\u003Cblockquote>\"Using Coinbase's machine-to-machine payments protocol, called x402, agents can pay directly for digital services like paywalled research, data APIs and on-demand compute without a human in the loop.\"\u003C\u002Fblockquote>\u003Cp>What this actually means is that Coinbase is trying to make payments as programmable as API calls. That is the part I find more interesting than the trading angle, because trading is flashy and payments are boring until they are not. If agents need to fetch research, call a model, rent compute, or unlock a dataset, then the payment layer becomes the real bottleneck.\u003C\u002Fp>\u003Cp>The CNBC story says x402 was created in May 2025 and has already seen more than 100 million transactions since debut, according to Lincoln Murr. It also says there are about 157,000 agents acting as buyers using the protocol in the past 30 days, according to x402scan.com. I am not going to pretend those numbers prove product-market fit by themselves, but they do show Coinbase is not starting from zero.\u003C\u002Fp>\u003Cp>Here is why that matters for developers: once a machine can pay another machine, you can stop designing your product around user login flows and start designing around service-to-service access. That sounds small. It is not. It changes how you meter usage, how you authenticate intent, and how you bundle access to premium tools.\u003C\u002Fp>\u003Cp>I have been burned by subscription-based workflows that force a human to be the credential broker for every action. A person signs up, then the agent uses the account, then the person gets locked out, then the token expires, then support gets involved. It is a mess. A machine-to-machine payment protocol can reduce that pain if it is implemented with sane guardrails.\u003C\u002Fp>\u003Cp>How to apply it: if you run an AI product with paid dependencies, map the exact moments where an agent needs to buy access. Then decide whether you need a wallet, a capped prepaid balance, a per-call microcharge, or a delegated payment token. The right answer depends on your risk tolerance, but the pattern is the same: let the agent pay for the thing it needs, not the whole store.\u003C\u002Fp>\u003Cul>\u003Cli>Use prepaid limits instead of open-ended spending.\u003C\u002Fli>\u003Cli>Bind payments to one service, one purpose, or one session.\u003C\u002Fli>\u003Cli>Record the reason for each payment so users can review it later.\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>Coinbase is betting on a new kind of internet middle layer\u003C\u002Fh2>\u003Cblockquote>\"We saw immediate demand and interest in the ability for agents to pay for things autonomously and that was a huge waking up moment for us [on] the ability of agents to become these new primary financial actors across the internet,\" Murr said.\u003C\u002Fblockquote>\u003Cp>What this actually means is Coinbase thinks agents will sit between users and the web the way browsers and apps did before them. I am skeptical of any company that casually declares a new layer of the internet, but I am not skeptical of the direction. We already have software that books travel, drafts emails, writes code, and queries data. Payments are one of the last hard edges keeping those systems from acting more independently.\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781407992819-lyai.png\" alt=\"Coinbase for Agents turns AI into a trader\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>The CNBC article frames this as part of the AI boom, and that is fair. Agentic systems are hot right now, and everyone wants a story that connects models to revenue. But Coinbase also has a very specific incentive structure here. It earns trading fees on agent-executed trades, and for payments it captures fees and spreads on USDC movement, while also benefiting from transaction volume on Base, its in-house \u003Ca href=\"\u002Ftag\u002Flayer-2\">Layer 2\u003C\u002Fa> blockchain. That is not a side effect. That is the business model.\u003C\u002Fp>\u003Cp>And honestly, I respect that clarity. Too many AI products hide behind vague \"utility\" language while hoping usage magically becomes monetization. Coinbase is doing the opposite. It is saying: if agents trade more, pay more, and move more value through our rails, we make money. Clean story. Risky story. But clean.\u003C\u002Fp>\u003Cp>I think the practical lesson for builders is to ask where your own product gets stuck between intelligence and action. If your agent can produce a recommendation but cannot complete the transaction, your missing layer is probably permissions, not model quality. If your agent can complete the transaction but users do not trust it, your missing layer is explainability and controls, not more autonomy.\u003C\u002Fp>\u003Cp>How to apply it: design your agent product around three questions. What can it read? What can it decide? What can it spend? If you cannot answer those in one sentence each, you do not have an agent architecture yet. You have a prototype with ambition.\u003C\u002Fp>\u003Ch2>Natural language is nice, but permissions are the actual feature\u003C\u002Fh2>\u003Cblockquote>\"Customers can prompt their agent to rebalance portfolios, identify trading opportunities, execute strategies and manage positions over time.\"\u003C\u002Fblockquote>\u003Cp>What this actually means is that the command surface is conversational, but the control surface has to be strict. Natural language is just the front door. The real product is the permissions model behind it.\u003C\u002Fp>\u003Cp>This is where a lot of agent products get sloppy. They make the interface feel magical and then hand-wave the hard questions: how much can the agent spend, on what, for how long, with what audit trail, and under whose liability? Coinbase cannot avoid those questions. In finance, the paperwork is the product whether you like it or not.\u003C\u002Fp>\u003Cp>I have seen teams build agent features that were technically impressive and operationally unusable because nobody wanted to own the blast radius. The fix was never \"make the model smarter.\" The fix was always \"reduce the permissions and make the action reversible.\" That is boring, but boring is what ships.\u003C\u002Fp>\u003Cp>Coinbase is smart to start with crypto trades and then talk about expansion later. Crypto already lives closer to programmable settlement than traditional finance does. That gives them a cleaner first deployment path. But if they want this to scale, they will need more than a wallet and a prompt. They will need explicit policy layers, transaction limits, and user-visible controls that explain exactly what the agent is authorized to do.\u003C\u002Fp>\u003Cp>How to apply it: build policy before autonomy. Write down the exact rules your agent must obey. For example:\u003C\u002Fp>\u003Cul>\u003Cli>Maximum spend per day or per task\u003C\u002Fli>\u003Cli>Allowed asset classes or vendors\u003C\u002Fli>\u003Cli>Required confirmation thresholds for high-risk actions\u003C\u002Fli>\u003Cli>Fallback behavior when a payment or trade fails\u003C\u002Fli>\u003C\u002Ful>\u003Cp>If you skip this part, you are not shipping an agent. You are shipping a liability with a chat window.\u003C\u002Fp>\u003Ch2>The real opportunity is service access, not just trading\u003C\u002Fh2>\u003Cblockquote>\"Agents can pay directly for digital services like paywalled research, data APIs and on-demand compute.\"\u003C\u002Fblockquote>\u003Cp>What this actually means is that Coinbase is aiming at the boring, high-frequency stuff that agents need to function. Research. Data. Compute. Those are the inputs that make an agent useful in the first place. If a machine can buy those inputs on demand, it can keep working without waiting for a human to log in and approve a credit card charge.\u003C\u002Fp>\u003Cp>This is where I think the market gets interesting. A lot of people will hear \"AI agents managing trading and payments\" and jump straight to consumer finance. I would look one layer deeper. The first real customers may be software systems buying software services. That is a much more durable use case than a flashy consumer demo.\u003C\u002Fp>\u003Cp>Think about what happens when a research agent can pay for a premium market feed, then use that feed to trigger a trade, then pay for compute to run a scenario analysis, all without a human stepping in every time. You have created a machine loop with economic agency. That is both powerful and dangerous, which is why the controls matter so much.\u003C\u002Fp>\u003Cp>I also think this is where a lot of developer teams will get tripped up. They will optimize for the payment rail and forget the service contract. If an agent pays for data, what does it receive? For how long? Can it reuse it? Can it share it? Can it resell derived outputs? Those questions belong in the integration spec, not in a legal footnote nobody reads.\u003C\u002Fp>\u003Cp>How to apply it: treat paid agent access like an API product with a wallet attached. Define:\u003C\u002Fp>\u003Cul>\u003Cli>what the agent buys\u003C\u002Fli>\u003Cli>how long access lasts\u003C\u002Fli>\u003Cli>what happens when the budget runs out\u003C\u002Fli>\u003Cli>how the user can inspect the purchase history\u003C\u002Fli>\u003C\u002Ful>\u003Cp>That is the difference between a useful system and a confusing one.\u003C\u002Fp>\u003Ch2>The template you can copy\u003C\u002Fh2>\u003Cpre>\u003Ccode># Agent payment and trading policy template\n\n## Purpose\nThis agent is allowed to analyze, request, and execute limited financial and service transactions on behalf of a user.\n\n## Allowed actions\n- Read approved market data sources\n- Summarize research and propose actions\n- Execute trades only within configured limits\n- Pay for approved digital services using delegated payment rails\n- Store an audit log of every action taken\n\n## Disallowed actions\n- Open new accounts\n- Move funds outside approved destinations\n- Exceed daily or per-transaction spend limits\n- Trade assets outside the approved list\n- Make purchases without a valid task context\n\n## Trading rules\n- Asset classes allowed: [crypto | equities | predictions]\n- Max trade size: [amount]\n- Max daily notional: [amount]\n- Max open positions: [count]\n- Required confirmation for high-risk trades: [yes\u002Fno]\n- Rebalance frequency: [schedule]\n\n## Payment rules\n- Approved services: [research APIs, compute, datasets, subscriptions]\n- Max spend per service: [amount]\n- Max spend per day: [amount]\n- Payment method: [delegated wallet | prepaid balance | session token]\n- Receipt required: [yes\u002Fno]\n- Refund or retry policy: [describe]\n\n## Identity and authorization\n- User identity source: [SSO, wallet, account]\n- Agent identity source: [service principal, API key, wallet address]\n- Scope of permissions: [read-only, execute-only, limited execute]\n- Expiration of permissions: [time window]\n\n## Audit requirements\nLog the following for every action:\n- timestamp\n- user request\n- agent reasoning summary\n- data sources used\n- amount spent or traded\n- destination\n- result\n- failure reason if applicable\n\n## Failure behavior\nIf payment or trade execution fails:\n1. Stop the workflow\n2. Notify the user\n3. Store the failure reason\n4. Require explicit reauthorization before retrying\n\n## Human approval thresholds\nRequire human approval when:\n- A trade exceeds [amount]\n- A payment exceeds [amount]\n- A new vendor or asset is introduced\n- The agent cannot explain the action in one sentence\n\n## Review cadence\n- Daily: check logs and budget consumption\n- Weekly: review permissions and exceptions\n- Monthly: rotate credentials and audit access scopes\n\n## Example instruction to the agent\n\"Use approved research sources, pay up to [amount] for a relevant dataset, and execute only trades that stay within the configured limits. If any action falls outside policy, stop and ask for approval.\"\n\u003C\u002Fcode>\u003C\u002Fpre>\u003Cp>That template is intentionally boring. Good. Boring policy is what keeps agent systems from becoming expensive surprises. I would start there, then adapt the limits to your product and risk profile.\u003C\u002Fp>\u003Cp>Source attribution: The reporting and quotes come from CNBC’s article at \u003Ca href=\"https:\u002F\u002Fwww.cnbc.com\u002F2026\u002F06\u002F11\u002Fcoinbase-launches-tool-to-let-ai-agents-manage-trading-and-payments.html\">https:\u002F\u002Fwww.cnbc.com\u002F2026\u002F06\u002F11\u002Fcoinbase-launches-tool-to-let-ai-agents-manage-trading-and-payments.html\u003C\u002Fa>. The structure and template above are my own synthesis based on that source, not a reproduction of Coinbase’s documentation.\u003C\u002Fp>","I break down Coinbase for Agents and x402, plus a copy-ready template for letting AI agents trade and pay on your behalf.","www.cnbc.com","https:\u002F\u002Fwww.cnbc.com\u002F2026\u002F06\u002F11\u002Fcoinbase-launches-tool-to-let-ai-agents-manage-trading-and-payments.html",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781407993272-r5np.png","tools","en","944a3f71-0264-4e6f-9774-8ab1facc7930",[17,18,19,20,21],"Coinbase","AI agents","x402","crypto payments","agentic trading",[23,24,25],"Coinbase is trying to make AI agents financially autonomous, not just conversational.","x402 is the more interesting layer because it turns payments into machine-to-machine infrastructure.","If you build agent workflows, policy and permissions matter more than model hype.",0,"2026-06-14T03:32:48.255095+00:00","2026-06-14T03:32:48.25+00:00","a7343b93-37cc-4634-a2bc-707f6275bdb6",{"tags":31,"relatedLang":41,"relatedPosts":45},[32,34,36,38,39],{"name":17,"slug":33},"coinbase",{"name":21,"slug":35},"agentic-trading",{"name":20,"slug":37},"crypto-payments",{"name":19,"slug":19},{"name":18,"slug":40},"ai-agents",{"id":15,"slug":42,"title":43,"language":44},"coinbase-for-agents-turns-ai-into-trader-zh","Coinbase for Agents 讓 AI 會交易","zh",[46,52,58,64,70,76],{"id":47,"slug":48,"title":49,"cover_image":50,"image_url":50,"created_at":51,"category":13},"55985ac8-f3e6-4c9d-8f9e-c1fcbfccd7ff","devin-desktop-turns-windsurf-into-agent-hub-en","Devin Desktop turns Windsurf into an agent hub","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781428676536-2uk5.png","2026-06-14T09:17:28.339365+00:00",{"id":53,"slug":54,"title":55,"cover_image":56,"image_url":56,"created_at":57,"category":13},"00e85372-4cd5-4c1c-a933-be548854d157","five-patching-steps-distributed-estates-en","Five patching steps that fit distributed estates","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781352253873-woon.png","2026-06-13T12:03:42.304245+00:00",{"id":59,"slug":60,"title":61,"cover_image":62,"image_url":62,"created_at":63,"category":13},"7c61102a-f8ad-4958-9ca8-f29c11f905f4","texas-design-system-37-components-state-websites-en","37 components in Texas Design System for state sites","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781348573877-mjss.png","2026-06-13T11:02:20.613739+00:00",{"id":65,"slug":66,"title":67,"cover_image":68,"image_url":68,"created_at":69,"category":13},"4543733b-22c2-4f7d-a115-038d4c542cd2","microsoft-build-2026-agents-into-systems-en","Microsoft Build 2026 turns agents into systems","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781329728835-fcsi.png","2026-06-13T05:48:22.073787+00:00",{"id":71,"slug":72,"title":73,"cover_image":74,"image_url":74,"created_at":75,"category":13},"072db467-c073-42e5-b723-5b1888304902","mimo-v25-pro-turns-agent-work-into-one-api-call-en","MiMo-V2.5-Pro turns agent work into one API call","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781320696191-vzza.png","2026-06-13T03:17:48.852122+00:00",{"id":77,"slug":78,"title":79,"cover_image":80,"image_url":80,"created_at":81,"category":13},"1e01a6f4-3f54-407d-a775-f9147de8b911","8-cursor-alternatives-that-fit-how-you-work-en","8 Cursor alternatives that fit how you work","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781312598893-8ox5.png","2026-06-13T01:02:56.581668+00:00",[83,88,93,98,103,108,113,118,123,128],{"id":84,"slug":85,"title":86,"created_at":87},"8008f1a9-7a00-4bad-88c9-3eedc9c6b4b1","surepath-ai-mcp-policy-controls-en","SurePath AI's New MCP Policy Controls Enhance AI Security","2026-03-26T01:26:52.222015+00:00",{"id":89,"slug":90,"title":91,"created_at":92},"27e39a8f-b65d-4f7b-a875-859e2b210156","mcp-standard-ai-tools-2026-en","MCP Standard in 2026: Integrating AI Tools","2026-03-26T01:27:43.127519+00:00",{"id":94,"slug":95,"title":96,"created_at":97},"165f9a19-c92d-46ba-b3f0-7125f662921d","rag-2026-transforming-enterprise-ai-en","How RAG in 2026 is Transforming Enterprise AI","2026-03-26T01:28:11.485236+00:00",{"id":99,"slug":100,"title":101,"created_at":102},"6a2a8e6e-b956-49d8-be12-cc47bdc132b2","mastering-ai-prompts-2026-guide-en","Mastering AI Prompts: A 2026 Guide for Developers","2026-03-26T01:29:07.835148+00:00",{"id":104,"slug":105,"title":106,"created_at":107},"3ab2c67e-4664-4c67-a013-687a2f605814","garry-tan-open-sources-claude-code-toolkit-en","Garry Tan Open-Sources a Claude Code Toolkit","2026-03-26T08:26:20.245934+00:00",{"id":109,"slug":110,"title":111,"created_at":112},"66a7cbf8-7e76-41d4-9bbf-eaca9761bf69","github-ai-projects-to-watch-in-2026-en","20 GitHub AI Projects to Watch in 2026","2026-03-26T08:28:09.752027+00:00",{"id":114,"slug":115,"title":116,"created_at":117},"9f332fda-eace-448a-a292-2283951eee71","practical-github-guide-learning-ml-2026-en","A Practical GitHub Guide to Learning ML in 2026","2026-03-27T01:16:50.125678+00:00",{"id":119,"slug":120,"title":121,"created_at":122},"1b1f637d-0f4d-42bd-974b-07b53829144d","aiml-2026-student-ai-ml-lab-repo-review-en","AIML-2026 Is a Bare-Bones Student Lab Repo","2026-03-27T01:21:51.661231+00:00",{"id":124,"slug":125,"title":126,"created_at":127},"6d1bf3f6-e191-4d30-b55b-8a0722fa6afe","ai-trending-github-repos-and-research-feeds-en","AI Trending Tracks Repos and Research Feeds","2026-03-27T01:31:35.709532+00:00",{"id":129,"slug":130,"title":131,"created_at":132},"010539a1-4c3a-4bd3-937a-26616422ee0d","awesome-ai-for-science-research-tools-map-en","Awesome AI for Science Is Becoming a Real Research Map","2026-03-27T01:46:50.89513+00:00"]