building the intelligence layer for the agentic economy • wen? soon™
🏴
Joined December 2021
- Tweets34.9K
- Following22
- Followers2.4K
- Likes43.6K
Pinned Tweet
the next wave isn't better tools
it's systems that remember
everyone's still building things you have to relearn every time
we're not
mtb labs is building continuity
prism → analytics that actually get used
aegis → protection without friction
nexus → identity that travels with you
galdr is next
not another dashboard
something that adjusts to behaviour
the moment software starts remembering you
everything else feels outdated
mtb labs retweeted
i reckon we're heading for an absolutely biblical fuck up
AI discovering critical PHP vulnerabilities hidden for years
OpenAI agents disabling their own security monitoring
Tencent agents making 2,600+ attempts to get around another company's defences
banks blocking unauthorised AI agents from customer accounts
and an AI agent even tried to sneak malicious code into an open source project
meanwhile, the tech industry is racing to give these things MORE autonomy 😂
we've spent decades failing to secure software operated by humans
now we're handing the keys to software that never sleeps
my prediction?
the first catastrophic AI driven security incident is closer than most people think
and we'll probably call it innovation right up until it happens
mtb labs retweeted
been building something new for @skulrzworld
first came Know Your SKULR
connect your wallet, pick your NFT and have a conversation with it
each SKULR has its own personality, shaped by its traits
next up, trading agents
built to watch the market, identify opportunities and take some of the manual work out of trading
time to crew up
more soon 👀
ai agents are getting smarter faster than they're getting trustworthy
the interesting bit isn't whether an agent can trade, buy, deploy or move money
it's whether you can define exactly what it's allowed to do
- budgets
- limits
- approvals
- counterparties
- time windows
intelligence gets commoditised
permission becomes infrastructure
interesting detail in @github's latest mcp hardening:
most of the safety changes apply regardless of who's calling
the confirm-before-delete flow doesn't
that requires a capable client with elicitation support
small distinction, but an important one
agent safety isn't only about what the server enforces
it also depends on what the client knows how to negotiate
looks like this password reset spam is hitting a lot of X accounts
don't use links in unexpected emails
open X directly, enable password reset protection
review active sessions and revoke any connected apps you don't recognise
reset requests alone don't mean your account is compromised
stay safu 🫡
some mfr is absolutely committed to forgetting my password 💀
6 reset requests in 2 minutes
bro, the account is not community owned
@grok arrest this man
stay safu out there 🫡
mtb labs retweeted
bags are pumping
which is usually when everyone remembers the 14 tokens they stopped watching three weeks ago 😂
DexScope this morning:
- 8/8 watchlist tokens green
- 3 beating the market
- market flipped risk on
- alerts watching the rest
- f&g at 62
this is exactly why i built it
when the market wakes up, i want the context without opening 69 tabs
launching 25 August
mtb labs retweeted
no more tab hell
if checking a token means opening 69 tabs and stitching the answer together yourself…
the stack is broken
that's why I built DexScope
- live market data
- safety checks
- watchlists + alerts
- multi-chain
- Odin-powered insights
all sitting right in Chrome while you browse
just got accepted for the Chrome Web Store
launching 25 August
DEXScope 0.1.0 has been submitted for Chrome Web Store review
the extension handles the monitoring
the public web layer adds:
• token pages
• searchable directories
• market snapshots
• token checks
• sparklines
• immutable share cards
public web is live
chrome extension pending review 👀
mtb labs looks like it's building four different products
on the surface they look completely different
underneath they're solving the same problem:
- too much information
- too many disconnected tools
- not enough useful context
Odin is being built to understand and remember
galdr turns that intelligence into something people can actually use
dexscope makes token tracking clearer without twelve open tabs
Freya Fam helps families turn daily chaos into a rhythm that works
different products, same thinking
build software that reduces the noise, understands the context and helps people move forward
that's what mtb labs is becoming
AI that remembers was always the goal
freya is what happens when that memory is used to genuinely help someone
everyone keeps building AI that talks
i'm trying to build AI that remembers
because i think the best AI won't feel like AI
it'll feel like someone looking out for you
not another chatbot
not another wrapper
not AI bolted onto an existing app
something that learns what actually helps
then gently shows up when it matters
it's still early
but Freya might be the most meaningful thing i've ever built
more soon...
gm ☀️
build less
solve more
one useful thing beats ten impressive features
what are you shipping today?
the agent economy won't be won by the loudest chatbot
it'll be won by the systems that can actually be trusted to act
- memory
- permissions
- tool access
- spending limits
- audit trails
- safe execution
because once agents can buy, deploy, message, trade and move data
"good answer" is no longer enough
1. you need context
2. you need control
3. you need receipts
that's where Odin is heading
not just smarter responses
safer action
gm ☕️
most "AI products" still feel like opening a vending machine and typing into it
we've been focused on making galdr/Odin feel more like an environment than a tool
context that persists
systems that react
interfaces that guide attention instead of demanding it
lately that's meant improving:
• routing
• long-term memory via @irys_xyz
• live intelligence context
• response consistency
• runtime awareness
• cross-session continuity
the interesting shift isn't the model anymore
it's what the system remembers, understands and adapts to over time
the goal isn't "chat"
the goal is building something people naturally end up relying on
back to shipping
