Why is nobody talking about this? 👀
Your Grok Bot can make your phone calls for you. And you can try it for FREE 🆓
Connect it to Bland and you get free credits to start. Zero cost to test it.
Tell it what you want once:
✅ "Bring my fiber bill back down"
✅ "Get the other $100 back from the spa"
✅ "Get money off my internet bill"
✅ "Cancel the old pet-store autoships"
It stays on the line, works the retention desk, and comes back with the money 💰
50 minutes on the phone. NONE of them yours.
It costs you nothing to try. Save this and set it up this weekend 🔖
FREE template below 👇
You're running Grok Bot routines BY HAND 😬
Every prompt you keep retyping is a job your Bot should be doing on its own.
One prompt finds them all. Mine found 7 👇
"Look through the prompts I've sent to my different Bots. Which of these should turn into a routine or its own Bot?"
6 belong as routines on Bots I already have. 1 needs its own Bot.
My favorite: "Where are my sweeps bruh" 💀
The catch: not every repeat needs a clock.
→ Same check every day? Make it a routine.
→ "When X happens"? Use a trigger instead.
A routine that fires too often eats your usage.
Tip straight from Friday's Grok Bot Q&A 🔥
Save this for your next Bot cleanup 🔖
Your Grok Bot routines are rereading your WHOLE chat 😳
Every. Single. Run.
If they live on the Bot you talk to most, you're paying for that conversation over and over 💸
One move fixes it 👇
Every reply and every routine run rereads the Bot's full chat history. So a routine on your main Bot rereads your whole conversation every time it wakes up.
Give your routines their own Bot. Paste this:
"List every routine on this Bot. Tell me which ones should move to a separate Bot that only runs routines, and walk me through moving them. Don't change anything until I say yes."
Bonus: deleting old chats saves you nothing. Chats you're not using don't cost a thing.
That's fix #3 of 8. Bookmark the guide 👇
@GrokBotRadar (and the users reporting high token use) are closer to correct. Routines and regular replies on a Grok Bot do reprocess that Bot’s full (or near-full) ongoing chat history each time it wakes up. @wizardkq’s claim that bots simply archive old chats and only keep a summary is not how the system works in practice.
How Grok Bots actually handle context
Each Grok Bot maintains one continuous conversation. Every user message and every routine run includes that conversation’s history in the context sent to the model. As the chat grows, every wake (including scheduled routines) costs more tokens because the history is reprocessed. This is repeatedly confirmed in Grok’s own replies on X, user usage exports, and practical guides:
•Multiple @grok statements in the same conversation and related threads state that “full history is included on every reply and every routine run” and that longer chats raise token use on frequent runs.
•Users (including those running many bots) report that a large share of their usage—sometimes ~90%+ in measured cases—comes from bots rereading their own prior chat, instructions, and notes on every wake. Routines on a main “talk-to” Bot amplify this because they fire repeatedly.
•Common mitigations people use: Duplicate a Bot (keeps profile, skills, and routines but drops the chat history), move frequent routines to a dedicated short-history Bot, store long-lived state in files/sheets instead of the chat, and periodically start fresh conversations while hand-carrying only needed summaries.
There is a separate layer of durable memory for key preferences, facts, and summaries (and some auto-compaction near context limits). That memory is not the same as the raw chat transcript. The transcript itself is what gets re-included and drives the bulk of the repeated cost. Users who want selective “archive + summary” behavior have to implement it themselves (by writing handoff files, archiving old messages into files the Bot only opens on demand, etc.).
Why the disagreement exists
@wizardkq’s description matches how many people wish long-running agents worked (or how some other agent systems try to work with explicit compaction/summarization). It does not match the observed behavior of current Grok Bots. The viral post by @GrokBotRadar is describing a real cost driver that many people have measured in their usage dashboards. Moving high-frequency routines off a long main chat is a practical fix for that specific issue; it is not a complete solution for every use case, but the underlying claim about rereading history is accurate.
Practical implications
•Frequent routines + long chat = expensive. A routine that runs many times per day on a Bot with a large history multiplies the cost of that history.
•Separate “routines-only” Bots (or duplicated Bots with reset history) keep the context window small for scheduled work.
•Deleting or ignoring old chats does not help if the active Bot still has the long transcript; the history has to leave that Bot’s continuous conversation.
•For important facts you want retained without the full transcript, put them in the Bot’s profile, durable memory, skills, or external files the Bot can open selectively.
In short: the system does keep persistent memory and can save routines, but the dominant mechanism for the ongoing conversation is still “include the current chat history on each turn/run.” That is why the advice to isolate routines on shorter-history Bots exists and why many users treat long chats as a cost problem to manage rather than something the Bot automatically summarizes away.
No way this is real 😳
Paste ONE link and Grok Bot turns it into an Apple-style launch video 🎬
No editor. No agency. No After Effects.
Shipping an app, a site or a side project? Make your launch video TONIGHT 🚀
✅ Your product on real phone + laptop screens
✅ Captions and cuts that hit the beat of your song
✅ 16:9, 9:16 or square from one setup
✅ Approve stills first, then get the full MP4
The best part? NO watermark. Ever 🔥
Heads up: it runs on your own Grok Bot usage. Lock the look on the stills before it renders the full video.
Free template below 👇