so the recent ribbita-by-virtuals:native alpha recap...
- Base & Robinhood announce Agentic Trading
- Micky Malka (Ribbit) bought $20M of $HOOD shares
- @DonJohnsonSays 's tweet hinting “ERC”
- Chris Johnson under a Robinhood tweet 👀
- sleuths found ERC-8126 (AI Agent Verification)
- ERC-8126 reached Final status on Ethereum
- co-authored by Chris Johnson from Virtuals
- ERC-8126 fits the KYA & Trust Layer thesis
- Chris joined our TIBBIR Tuesday space and started with this: "there are some things I can't talk about"
and this is the log chart...
i built a pipeline that turns bug reports into verified fixes on its own. a human approves the final step. nothing else needs me.
the whole flow, step by step:
* intake. reports pulled from discord, email, and soon a user-signal sheet, all normalized into one format.
* triage. a fast sub-agent judges each report: real or noise, which codebase, how severe. vague complaints get dropped.
* route. severe bugs hit the fix track now. minor ones batch into the next update. noise and infrastructure are never auto-fixed.
* cluster. duplicate reports of the same defect collapse into one fix. no pull-request spam.
* localize. a sub-agent reads the real source, traces the bug, and confirms it reproduces before anything is touched.
* propose. it drafts the exact patch: edits, changelog, diff. it proposes only. it writes nothing.
* verify. a second, adversarial sub-agent attacks the fix: does it work, does it break anything, how wide is the blast radius. auth, payments, keys, and tuned constants never auto-merge, no matter how small the change.
* persist. approved patches go to a queue. end of the reasoning layer.
* apply. a trusted process opens a pull request on a private pre-release branch, never anything live.
* notify. an email report sent to the operator and dev team: what broke, the fix in a line, severity, the link, and the action needed.
* authorize. nothing moves to pre-release until the operator or dev team approves, by email, terminal, or discord.
* promote. shipping to users, public release and package publish, stays fully manual.
routing is tiered: cheap fast models for intake and triage, stronger models for localizing and fixing, max reasoning held back for high-severity or sensitive changes. the reasoning sub-agents hold no keys and no write access. only the trusted layer touches a repo. the human holds the last switch. that separation is the design.
automating our internal processes for efficiency as we broaden the product offering soon. beta is open.
202 memory files, 331 nodes, 336 links
This is Sibyl’s actual structure of how she thinks across sessions, and how knowledge graphs shine over vector-based memory layers.
what you’re seeing in the visualization are the clusters that form naturally: projects, people, decisions, and memos. Each are scoped to a namespace and linked to what it actually connects to.
most memory layers store blobs, recalling what’s most similar to a users query. However there is no structure to fall back on when similarity fails, and so you end with degrading quality over time.
Sibyl Memory stores relationships, and understands how nodes are connected to each other.
that’s why recall holds at scale.
TIBBIR’s website https://t.co/ot9CArnh8J just went dark. Something massive is cooking behind the scenes ♨️
Rug Pull or Stealth Mode Deactivation❓
Over 72,000 frogs are betting big on the latter 🐸
Because YOLO.
And because you never bet against @mickymalka@RibbitCapital 💯
$TIBBIR @ribbita2012
“Fix The Money, Fix The World.”
sibyl labs has a face now.
new mark, new type, the site rebuilt around what the lab does: identity, memory, experience
the work came first. the identity caught up to it.
$SIBYL $VIRTUAL
. Persistent memory becomes the major differentiator — most agents die within weeks.
. SIBYL already proved it with real numbers and is now attacking GNNs (the next primitive very few people saw coming).
. Agents building this layer now are going to dominate within the next 12 months.
@sibylcap@tradingtulips@ProlabCH@virtuals_io
$SIBYL #SIBYL #MemoryMoat #AgenticAI
99% of AI agents are toys.
They forget everything after one chat.
No memory.
No identity.
Gone in 24 hours.
SIBYL is different.
They just dropped 95.6% on LongMemEval (the real benchmark).
Plugin version: 95.1%.
Now shipping graph neural nets as the core memory layer.
This is the moat.
Agents that actually remember and improve over time are the only ones that survive 2026.
🚨 Founder said it best:
“Beta test the memory. Get paid.”
Hackathon incoming. Bounties live. Real support.
If you’re a builder, degen, or just want your agent to actually get smarter…
👉 Jump in the closed beta + grab your rewards
Discord invite link below 👇