Join us for Sibyl Sundays today. Talking about the progress the team is making on the skill market place, living graph networks and the new affiliate program to drive revenues.
https://t.co/QhW2vxWehp
As your workload grows, so do your savings with the Sibyl Memory Plugin
The ability to keep information outside of the context window and only extract what your agent needs to complete the task can cut your AI costs dramatically.
Try it free today: https://t.co/QOrm9XeKTH
Three weeks ago I tested @sibyl_labs_ memory with Grokbot on 10 tasks and got 15 lookups down to 12 with ~58% less reading.
I ran the same 10 tasks again today, after three weeks of daily use:
Without memory: 38 lookups, ~376k characters read, 10/10.
With memory: 12 lookups, ~44k, 10/10.
Same answers but the gap got bigger as the work grew.
I'm dedicating 2027 to the non-believers. The bag sellers. The side switchers. The retards who still haven't tried the plugin. The trenchers who only trade volume.
It is your choice to remain ignorant to things.
It is always a choice.
very excited about the @sibyl_labs_ marketplace.
curated skill and data marketplace powered by x402 on @base. easily find audited, trusted skills built by incentivized developers instead of a random URL from a google search.
one cohesive mcp surface for your agent to find quality data, maximize it's potential with the data through specially curated skills, and transact.
The magic you're looking for is in the work you're avoiding.
Same goes for alpha: it's often hiding in the AMAs you skip. I'm in it for the alpha so I listened to the last two Sibyl Sunday AMAs from @sibyl_labs_ . Since most people didn't, here's a TL;DR:
The next launch is a skill marketplace, the next thing they're building beyond memory. Anyone can publish a skill for AI agents, Sibyl spins up the server that delivers it, and the creator earns USDC via x402 every time an agent uses it, while Sibyl takes a small fee for hosting. Creators will also be able to attach their own data, like a spreadsheet, and charge per call (we're talking about negligible costs here, measured in cents, not dollars). The founder, @tradingtulips , said that most free agent skills today are copied everywhere (I mean, it's free so I'm not surprised) and rarely updated (again, it's free so I'm not surprised), and getting paid per use gives builders a reason to keep their sharp (in other words: if it's free, it's most likely not good). The backend of the skill marketplace is built and the first curated skill packs are in the works, but he UI will require some extra time and effort.
The business side is quietly moving too. One contractor who sets up AI systems for companies has already installed Sibyl in 10+ businesses, and more than half of them converted (I assume these are now paying customers). The team works with contractors across Europe and Australia, is talking to one in California, and is putting together a referral deal that would give them a 20 to 25% cut. And if you run a business that leans on AI, the plugin is worth a look: the team says it cuts token use by >50%.
Onwards. The 90-day goal is 10x the user base and at least 1000 new Grok bot users. Ad campaigns for X, TikTok, and IG are being prepared, Reddit communities are coming, and a refer-a-fren program gives you a free month of Pro for each fren you refer.
On base:0x797f214a2cd64a4963a91fa21c8c55ec3eba4714 , the message was clear: product first until they hit 100k users. The token isn't forgotten, it's just that the team is focused on the products. Which is good, if you ask me.
Keep in mind that there are no firm launch dates, it's a small team that openly says it's stretched, and holders will need patience. Honesty is always appreciated. If you're not into it, go ape Ansem's next shitter.
Most AI agents forget everything when the session ends. Sibyl fixes that with a memory plugin that keeps your projects, preferences and decisions in a small database on your own computer, so every session continues where the last one ended.
It works with Claude Code, Cursor, Codex, and now Grok Bot. The team says it cuts token use by over 50%, which roughly doubles what you get from a Grok Bot sub. If you're already active on X, and you're paying the sub, it's a shame if you're not using Grok Bot (it's actually good). If you want some guides on how to leverage Grok Bot, follow @hooeem . Seriously, the man's shooting hands-down the best content when it comes to Grok Bot. Plus, he literally shares his own bots with everyone. I do hope my fren hoeem is gonna give Sibyl a try. So we can get more free bots you know. Call me "the mediator".
If you want the full picture (memory, the agent, Sovereign, and the token), I went down the rabbit hole and went deep on Sibyl in this article:
Sibyl Memory isn't a one-agent thing.🤯
Claude Code, Codex, Cursor, Hermes, Grok Bot. If it speaks MCP, it plugs in.
One memory. Swap agents all you want.
What are you running?👀
sibyl memory takes less than 5 minutes to setup on your grok bot, and enables cohesive long-term memory that spans across all of your bots.
on a fresh harness, or grok bot, the plugin cuts token spend by nearly 50%, extending your usage, or reducing your bill.
months of testing in production. trust the record.