OpenClaw 5.16: What You Need To Know…

OpenCore 5.16 just showed up on GitHub as a pre-release beta, and there are two things in this update that make your AI agents way more useful. First of all, your agent can now use Grok without an API key. If you have a super Grok subscription, you just log in, no extra cost, no developer portal. Your agent gets full access to Grok through your existing subscription.

This is a big thing that's just been released inside Homeless Agent, and it's a bit of a game-changer. Second, your Telegram agent can now sit inside a group chat, listen to every single message, and stay completely quiet until you tell it to speak. It reads the room, it builds context, and when someone finally asks a question, the answer is 10 times better because it already knows what everyone else has been talking about. I'm going to walk you through everything in this update, why it saves you time, and the one setup that turns all of this into a full AI operating system you control from one screen.

Let's get into it. So let's start with Grok because this one helps you save costs straight away. Right now, if you want your open-world agent to use XAI's Grok, you need an API key. That means going to XAI's developer site, add in a card, buy an API credit, and if you're paying for your super Grok subscription and then paying again for API access, well, that's two bills for the same AI.

OpenClaw 5.16 aims to fix that. If you have a super Grok subscription, you just log in with your account. Your agent uses Grok for your subscription. There's no API key, no extra bills, just log in and go.

Grok has been getting really good, especially for finding up-to-date information, creating videos, creating images, and now your agent will be able to use all of that for your existing subscription when this drops one login and you're done, right? So this is in the pre-releases coming from 5.16 inside OpenClaw. Now, if you've been paying for super Grok but only using it on the website, well, this is a big deal, right? Your OpenClaw agent now has access to everything Grok can do for free because you're already paying for it.

Now, the Telegram feature. This one is going to save you a lot of time if you run group chats. Right now, your agent in the Telegram group has two options, reply to everything, which is kind of annoying, or only reply when someone tags it, which means it has no idea what anyone's even been talking about, and both are pretty bad options. OpenClaw 5.16 adds a third option.

Your agent listens to everything in the group, every message, every conversation, every question, but it stays quiet, completely silent. It doesn't say a word until you tell it to. Here's why that's powerful. Imagine you run a customer support group.

A hundred messages go by during the day, people asking questions, sharing feedback, talking about your product. Your agent reads all of it. It knows what everyone said. It understands the topics, the concerns, the requests.

So when someone finally asks your agent a direct question, the answer is much better because it already understands the full conversation. It's like having an employee who's been sitting in the room all day, paying attention to every single word, waiting to help the moment someone needs them. Or, for example, think about a team chat. Your agent listens to project updates, meeting notes, task discussions.

It never interrupts. But when you say, give me a summary of today or create a task list from what the team has already discussed or already has everything inside, most agents jump in too much or miss too much context. This gives you the best of both worlds. Full context, but zero noise.

Now, if you want to get the most out of OpenClaw, check out our new Agentic OS inside the AI Profit Boarding. Link in the comments description or go to the AIProfitBoarding.com. It's a full operating system. I built the Let's You Manage OpenClaw Claude and Hermes for One dashboard.

So your agents share one memory. They know your business, they know your goals. They work together instead of separately, which makes every single one of them way more powerful. And when features like Groko off and silent group mode go stable, they plug right into the same system.

Your agents get smarter, you save more time, and everything compounds. You get the full zip file that I've built for creating this. Every prompt I used to build it, and I'll walk you through the whole setup on our coaching course. 3,000 business owners in their building with AI agents right now.

Link in the comments description or go to the AIProfitBoarding.com to get access. Scheduled tasks got easier to chain together. Before, if you started a scheduled task, it would run in the background. You had no easy way for it to finish before starting the next one.

Now, you can tell your agent to wait with a timeout and it blocks until the task is done. Way simpler if you're building workflows where one step depends on the last. Skills load faster. Every time your agent gets a message, it used to rebuild its entire skill list from scratch before, right?

That slows things down. Now it saves that list and reuses it until your settings change. Your agent responds faster, less waiting. The setup wizard now works in Chinese too.

Simplified and traditional. And if you're helping a client set up OpenCore and they speak Chinese, the whole first time setup is now in their language. Discord reconnections got more reliable. There was a problem where your agent would get disconnected and the reconnection would fail because it tried to connect on the wrong connection.

Your agent would drop out of Discord until you restarted it. Fixed. It reconnects properly now with this new update. Matrix encrypted rooms should work better.

Some new encrypted rooms were failing silently before. Your agent couldn't send messages, but you'd have no idea why. That is fixed. The dashboard now shows you how much of your context window you've used even when your AR model doesn't report that number.

Before if you used a local model, the usage would show blank. Now it estimates from your conversation history. So you should always know when you're running low on context window. And there was a bug when your agent would keep compacting.

Summarizing old messages to save space over and over after every single message. That wasted tokens and slowed everything down. It compacts now once when it needs to and then stops. Now here's what we need to take away from all of this.

Features like Grokker off and silent group mode are useful on their own, but they're 10 times more useful when your agents work together. Most people have Claude open in one tab, maybe open Claude running in somewhere else, maybe Hermes in another window or separate. None of them share information. None of them know what the other one is doing.

It's like having three employees who never talk to each other. When your agents share memory, when they all know your goals, your business, your customers, every single feature becomes more powerful. When your Telegram agent listens silently in a group and feeds what it hears into a shared memory system, well now Claude knows what your customers are saying too. And now Hermes can research based on real feedback.

Everything connects, everything compounds. This is a pre-release beta. It's on GitHub, but it's not a stable release yet. So don't install this on your live setup.

If your agents are working right now, leave them on their own. And if you want to test it, backup first, always run. Open Claude backup, create that, save your settings, your conversations, your memory, everything, right? Then run Open Claude update channel beta to try it, test your channels, test your messages.

If something breaks, you can always roll back to the stable version that you created a backup of. Know your current version, write it down if things go wrong as well. You know exactly where to go back and give it a few days after the stable release drops. You know, let other people test it too.

If things look clean, then you could update from there. And the people who learn AI agents now, whilst things are still early, are the ones who will be ahead when everything settles. Every feature you set up, every workflow you build, every problem you solve, it adds up. If you want your Open Claude setup to actually save you time every day, not just be another tool you check sometimes, go grab the Agentic OS inside the AI Profit Bottom.

It turns Open Claude and Hermes into one system with shared memory, shared context, and one dashboard you control. Your agents understand your business. They remember everything. And every new Open Claude update, like grok, login, silent group chat, makes the whole system just more powerful automatically.

I built it in one session. You get the zip file, the prompts, the Obsidian memory setup, and the coaching calls where you set up together, step-by-step as well. 3,000 members in there, daily tutorials, a 30-day roadmap, a map to find people near you. Link in the comments description or go to the AIprofitbottom.com.

Before you update anything, as I said, run Open Claude Backup Create. Backup first, decide second. And I'll see you in the next one. Cheers, bye.