Holder mandates are live on Palisec.
Every agent token says "holders decide." Until now, nothing made the agent obey. Now $PALISEC holders vote on an agent's limits, and Palisec enforces the result on chain. The agent can't co-sign its way past it.
Here's the full run on Robinhood Chain mainnet, every step on the record:
→ The demo agent's limit was 0.00001 ETH per transaction. It asked for 0.00002. Denied (PS_BUDGET_001). Nothing reached the chain.
→ Holders voted to raise the limit to 0.0003 ETH: 1,800 $PALISEC for, 0 against, quorum reached. Every vote a gasless signature.
→ It loosened the limits, so it didn't apply right away. It waited out a 60-minute delay the owner could cancel. Then it applied.
→ It asked for 0.0004 ETH, over the new limit. Denied.
→ It asked for 0.00002 ETH again. Allowed, co-signed and executed:
https://t.co/mOBV3sdGrx
→ Holders voted to tighten it back. Tightening doesn't wait: it applied 14 seconds after the vote closed. The same 0.00002 ETH: denied.
Why it can't be gamed:
→ Votes stay inside the owner's limits. The owner sets hard maximums, and no whale can vote past them.
→ No flash loans. Voting power is fixed at a snapshot block before voting opens.
→ Loosening waits, tightening doesn't.
→ Anyone can recount. Every signature, weight and snapshot balance is in a public vote file. Both votes recount to the same result from the public chain alone.
The demo mandate, with every vote and decision: https://t.co/bEibP9tXMo
Not a dashboard that describes what holders want. A rule the agent can't break.
Our draft standard for AI agent accounts just got its first serious technical reviews on Ethereum Magicians.
Two reviewers went line by line through Policy-Enforced Agent Accounts and raised six points, including:
→ how fast a new policy can take effect, and whether every change should be announced on chain first
→ how apps should check that a decision receipt is still current at the moment they use it
→ which permissions sit outside any policy, and how the standard should say so plainly
That's exactly why we submitted it as an open draft instead of keeping it in-house. Rules that guard real money should be argued over in public.
We're writing a reply to every point, and the changes will go into the next version of the spec and the live registry on Robinhood Chain.
Read the thread and join in: https://t.co/KcuD7q0dCk
The quietest way a wallet gets drained isn't a hack. It's an approval.
One "approve unlimited" to the wrong contract, and that contract can pull every token of that type, today or six months from now. No new signature needed.
For AI agents it's worse. An agent approves whatever the transaction it was handed asks for. A poisoned webpage or a fake token page is enough.
With Palisec, an approval is treated like a spend:
→ it's checked against the owner's policy before Palisec co-signs it
→ it's simulated first, and risky approvals are refused with a reason code
→ nothing reaches the chain unless the policy allows it
→ and if something slips through elsewhere, one switch stops every agent at once
Your agent can still trade. It just can't hand out the keys to your tokens.
Robinhood Chain puts stocks onchain. AI agents can now trade NVDA or AAPL the same way they trade ETH.
That's powerful. It also means an agent holding one private key can buy any stock token, in any size, with nobody watching.
What owners actually want from a trading agent:
→ a list of the tokens it may touch. Anything else is refused.
→ a cap per trade and an hourly budget
→ a human yes above an amount they choose
→ a check that whatever it buys can actually be sold back
→ one switch that stops every agent at once
All of that is live in Palisec on Robinhood Chain mainnet today.
Your agent gets the market. You keep the limits.
We've started building something AI traders have been asking for: loss limits the agent can't override.
Most AI trading agents today have a "risk limit" that's really just an instruction in a prompt. The agent is told to stop after a bad run. A confused or manipulated agent doesn't.
Palisec already limits what an agent can spend. Next, we're limiting how much it can lose:
→ The owner sets a maximum loss, for example 5% or $50 in 24 hours.
→ Palisec tracks the account's value. Once the limit is hit, Palisec stops co-signing anything that adds risk. The agent's key can't move funds alone, so there's no way around it.
→ Safe moves stay open: selling into a stablecoin, or sending funds back to the owner. You're never locked out of your own money.
→ Money you add or withdraw doesn't count as profit or loss. Only trading results do.
→ Prices come from sources that are hard to manipulate. If a price can't be read, risky trades are blocked, never allowed.
→ Only the owner can reset it, and a reset waits out a delay, so a stolen session can't simply switch it off.
Every block will come with a reason code and a signed receipt, readable on chain through our open standard.
Building now. Live only once it's proven on mainnet.
Honest question for traders running AI agents:
Would you give up some upside for a hard maximum loss the agent can't override?
Not a stop-loss the agent is told to respect. A limit outside the AI: if the account drops more than a set % in 24 hours, the agent simply can't open new trades until you say so.
→ Yes, give me the hard floor
→ No, let the agent manage its own risk
→ Depends on the strategy
Reply with your pick. We're deciding what Palisec builds next, and this is on the list.
Agreed. And the bottleneck for agentic finance in 2027 won't be how smart the agents are.
It'll be whether people are willing to hand them real money.
That comes down to three things the owner controls, not the agent:
→ A ceiling: the most the agent can ever spend, enforced outside the model
→ A brake: anything that loosens the rules waits and can be cancelled, and one switch stops everything
→ An exit: the owner can always move the funds out, even if the infrastructure provider disappears
Intelligence gets agents invited. Control gets them funded.
We're building the control part on Robinhood Chain, live on mainnet today.
https://t.co/xCoSdYy3h2
The newest trend: open-source AI trading desks with an analyst agent, a trader agent and a "risk team" agent.
Here's the problem. The risk team is a prompt. The trader agent can be talked past it, it can hallucinate around it, and one bad input can override it. Risk limits that live inside the model are suggestions.
Real limits live outside the model:
→ the trading key can't move funds without a second signature
→ the second signer only signs inside limits a human set: size, budget, markets
→ trades that break a limit are refused before they reach the chain, with a reason code
→ raising a limit waits out a delay. Cutting one is instant.
Let the AI do the analysis. Don't let it grade its own risk.
If 99.999% of transactions end up between agents, every agent will have to answer one question before it deals with another:
Can this counterparty's account do something its owner never allowed?
Humans read terms of service. Agents need something they can check in a single call.
That's why we put Palisec's rules on chain, where any agent can read them:
→ is this account protected by a policy signer, or is it one bare key?
→ which policy is it running right now, and is a looser one about to apply?
→ is its emergency stop on?
One read from a public RPC. No API key, no login, no trust in us.
That's the kind of infrastructure agent-to-agent finance needs, and it's live on Robinhood Chain mainnet today.
https://t.co/GgYmNY6olR
Prediction: in 10 years, 99.999% of financial transactions will happen between agents (with humans making up the remaining 0.001%).
Agents need new infrastructure to scale to this point, and we’re building it.
Volume from AI agents on Coinbase keeps hitting all-time highs. The agents are already trading.
On an exchange, a trading bot gets an API key with limits built in: trade-only permissions, no withdrawals, withdrawal whitelists, rate limits. Exchanges learned years ago that you never hand a bot the keys to everything.
Onchain, most agents still get one private key that can do anything. Send everything, approve anything, to anyone.
Palisec brings exchange-style limits onchain, on Robinhood Chain:
→ the agent's key can't move funds alone
→ only allowed contracts and tokens, inside per-trade and hourly limits
→ sending to an address that isn't on the list gets refused
→ and the owner keeps a recovery key, so nobody else holds custody
The same idea that made exchange bots safe. Now for agents with their own wallets.
https://t.co/AsCVXkEyOX
Which AI agent is the most popular on Coinbase?
Grok Bot, for now.
But Claude, ChatGPT and custom setups are catching up fast, and volume on Coinbase for Agents keeps hitting ATHs.
Palisec just submitted a draft standard to Ethereum: Policy-Enforced Agent Accounts.
Today, if a wallet, a lender or an agent marketplace wants to know whether an AI agent is safe to deal with, it has to take the agent team's word for it. This standard lets any app check it straight from the chain instead:
→ Is this agent's account protected by an outside policy signer?
→ Which policy is active right now? Fetch the document and check it matches the hash on chain.
→ Is a looser policy waiting to apply, and when?
→ Why was a transaction blocked? Standard reason codes every app can read.
→ Is this decision genuine? Every decision comes with a signed receipt anyone can verify.
And it's not just a document. The reference implementation is live on Robinhood Chain mainnet:
→ a verified registry contract with no owner, no admin and no upgrades
→ our demo agent's policy published on chain, and a pending change published 3 seconds after it was made
→ a blocked request that came back with a signed receipt
The proof: a fresh script with only a public RPC and the open-source library (npm: policy-enforced-accounts). No Palisec API, no keys. It read the agent's signer, its policy, the pending change, and verified the receipt.
The spec and how to read any account: https://t.co/YSnU6ygtqb
The discussion on Ethereum Magicians: https://t.co/KcuD7q0dCk
Question for anyone running an AI agent with real money:
What's the one thing you'd never let it do, no matter how convincing its reasoning sounds?
→ Send funds to an address it has never used before
→ Buy a token it can't sell back
→ Spend more than a set amount in an hour
→ Give a contract unlimited approval over its tokens
→ Make any big move without a human saying yes first
Every one of these is a rule you can set in Palisec today. It's enforced by the co-signer, not by the agent's judgment, so a fooled or buggy agent still can't break it.
Reply with yours. We'll share what the most common answers look like as a Palisec policy.
Three days on Robinhood Chain mainnet. Here's everything Palisec has shipped since, all of it on the record:
→ Mainnet launch: six outcomes (allowed, denied, held for approval and more), each recorded with a real transaction.
→ $PALISEC utility: hold 5,000,000 in your organization's owner wallet and get Pro limits, read straight from your wallet.
→ Loosening delay: anything that raises an agent's limits waits 60 minutes, and the owner can cancel it. Tightening is instant. A stolen session can't quietly open the gates.
→ Asset safety: before co-signing a buy, Palisec simulates selling it back. Honeypots and hidden taxes get blocked before the money moves.
→ Verify on chain: anyone can check whether an agent is really protected by Palisec, no API key needed. https://t.co/1Kv1Mufl3a
→ Holder mandates: $PALISEC holders voted on a live agent's limits on mainnet, and Palisec enforced the result.
In progress: a draft standard so any wallet or app can read an agent's active policy, see why a transaction was blocked, and verify a signed receipt for every decision.
We don't call anything live until it's proven on mainnet. Every item above has a transaction behind it.
Before writing our standard for AI agent accounts, we read every standard that already touches agent permissions on Ethereum. Here's what each one covers:
→ ERC-4337: how smart accounts execute transactions. Our accounts already run on it.
→ ERC-1271: how a contract proves a signature is valid.
→ ERC-7579 / 7780: how a single account can enforce rules inside itself.
→ ERC-7710: delegating limited authority from one account to another.
→ ERC-8004: who an agent is, and its reputation.
→ ERC-8226: a registry of mandates for regulated assets.
→ The spend-mandate proposal: guardrails inside the token itself.
All useful. But none of them lets any app answer four simple questions, straight from the chain:
1. Is this agent's account protected by an outside policy signer?
2. Which policy is active right now, and is a looser one waiting to apply?
3. Why was a transaction blocked?
4. Is this decision receipt genuine?
That's the gap we're filling. Not a new way to enforce rules, but a common way to read and prove them, whatever does the enforcing: a co-signer like Palisec, an onchain module, or a Safe guard.
We'll reuse the reason names other standards already use where they mean the same thing, and link to ERC-8004 agent identities, so it fits with what exists instead of competing with it.
It ships with a working reference implementation on Robinhood Chain mainnet. Live only once it's proven.
#12 is the one we're building.
"Agents that need to spend money... someone builds the spend controls, fraud protection, receipts."
On Robinhood Chain, that's Palisec:
→ Spend controls: every transaction an agent makes is checked against limits its owner set. Per-transaction caps, hourly budgets, allowed contracts and tokens. Nothing outside them gets co-signed.
→ Fraud protection: a prompt-injected agent asking to send funds to an unknown address gets denied with a reason code. A token that can't be sold back gets blocked before the buy.
→ Receipts: every decision is logged with its reason, and every allowed transaction is on chain for anyone to check.
And the part most spend-control tools skip: the agent's key can't move funds alone. The limits aren't a suggestion the agent can ignore. Every transaction needs Palisec's co-signature, and Palisec only signs inside the rules.
Live on Robinhood Chain mainnet: https://t.co/YkFPykRC0r
The biggest opportunities right now:
1. build for solving loneliness (the more AI floods everything, the more people crave real human connection, IRL and small social)
2. build for places with no cameras and no recording (being offline and unseen is turning into the new status symbol)
3. build for the crisis of young men. Millions of guys in their 20s feel completely lost with work, dating, and purpose. Whoever builds real brotherhood, mentorship, and an actual path forward wins a massive market (and helps a ton of people).
4. build for protecting people's brains (feeds are turning everyone's attention to mush, tools that protect your focus and mood are an easy sell)
5. build for the death of "everyone agrees" (you can find a chart or a podcast clip to prove any point now, people are starving for one source they actually trust)
6. build for day 2, not day 0. Anyone can launch an app in a weekend now. But how can you keep it running, fix the mess, and grow it for 3 years.
7. build for throwaway software (not everything needs to be a $20 a month subscription, tiny apps made for one trip, one wedding, or one group chat are coming)
8. build for people whose ego just got bruised by AI. Millions of smart people are realizing their job was mostly pattern matching. Help them figure out what they're good for next.
9. build for things that take effort on purpose (when everything is instant and free, nothing feels earned, people will pay for slowness and craft)
10. build for people who miss being told they're wrong. AI agrees with everything you say now. Real rooms where smart people actually push back on your ideas are becoming a luxury.
11. build for skin in the game. The internet is full of free takes and bots. Any platform where people have to put real money, real names, or real work behind what they say wins.
12. build for agents that need to spend money (they're getting virtual cards and budgets, someone builds the spend controls, fraud protection, receipts)
13. build for the end of fake email jobs (half of corporate America knows their job is theater, build the exit ramp for people who want to do real work)
14. build for the friend with great recommendations (nobody trusts Yelp reviews etc anymore because it's all bots and SEO, people will pay one obsessive person to just tell them where to eat and what to buy)
15. build for life after the chat box. Typing prompts already feels old. The best AI products won't look like chatbots at all, they'll just sit in the background and proactively handle things.
16. build for markets too small to matter before (500 lobster fishermen was never worth a team, now it's a real business if you can figure out what to build value to add to these fishermen lives!)
17. build for helpful without being creepy. AI is getting very personal very fast. Whoever figures out how to make it feel warm instead of weird owns consumer tech.
18. build for small private rooms (the public internet feels like an ad and a bot farm, all the good conversations and deals are moving into small invite-only groups)
19. build for someone taking the blame. AI can write the contract or read the scan, but a machine can't look you in the eye when it goes wrong. People still pay for a human on the hook.
20. build for learning that actually keeps up (a four year degree is way too slow for how fast things move now)
21. build for the anti-AI premium (as everything gets generated, human-made and unpolished become status symbols people pay up for)
note: more trends/ideas https://t.co/a5ARFnvSnA (free to sign up)
22. build for the return to the physical (screens fill with slop, people crave the real world, the hands on, the local, the analog)
23. build for the caregiving wave. The population is aging fast, tens of millions are caring for parents, and the whole burden is landing on families with no support.
24. build for finding things you didn't know to search for (algorithms just feed you more of what you clicked yesterday, people miss stumbling onto something totally different)
25. build for the housing and rootlessness problem (people can't afford to settle down, and the whole idea of a stable home base is up for grabs)
26. build for anti marketing (the slickest, most polished websites now look like AI slop or a scam, sometimes a plain text page from 1998 converts better because it feels like a real person built it)
27. build for agents that sit on hold for you (canceling a gym membership, fighting a medical bill, haggling with your internet provider, aint nobody got time for that)
28. build for software you actually own (people are exhausted from renting 30 different subscriptions every month, pay once and run it on your own machine is a massive wedge)
29. build for stuff that only works locally. The internet feels like one giant mall where every city has the exact same coffee shops and trends. Things that refuse to scale are suddenly cool again.
30. build for the 90% automated, 10% human sign-off model (clients don't actually care if AI did the heavy lifting, they just want one real expert to check the work and put their name on it)
31. build for infinite creative testing. Some brands are lighting money on fire on Meta because their ads die in 10 days. Build an automated pipeline that pumps out 100 fresh UGC and static ad variations a week based on what's actually converting in their ad account.
32. build for cold outreach that doesn't look like cold outreach. Mass blasting 50,000 people with "Hope this finds you well" is dead lol. Build an agent that watches for one specific trigger like a new VP hire or a broken checkout page, builds a custom teardown on the fly, and sends it before a human SDR even wakes up.
33. build for programmatic free tools instead of blog posts. Spin up 200 tiny, hyper-specific calculators, little apps and templates for one niche, rank them, and turn the traffic straight into conversions.
34. build for the food and toxin awakening (people are finally waking up to ultra processed food, microplastics, and why everyone feels sick and tired all the time)
35. build for the death of third places (there is almost nowhere left to just hang out without spending $50, which is why run clubs, saunas, and neighborhood clubs are blowing up)
36. build for people who can't retire at 65. Either they can't afford it or they're going to live to 95 and don't want to spend 30 years watching TV. The second act"market is huge.
37. build for the "undo" button for agents. Giving an AI access to your inbox, calendar, or bank account is still terrifying. Whoever makes it safe to let go of the wheel wins.
38. build for the sports betting and day trading hangover (a whole generation of young people are addicted to gambling on their phones and blowing up their bank accounts)
39. build for the death of the attention span. Smart adults can't even sit through a movie without checking their phone or finish a book anymore, and they're starting to hate themselves for it.
40. build for keeping marriages alive. Couples are stretched thin with work, kids, and money, and traditional therapy is $250 an hour with a waitlist. Anything that actually saves relationships is priceless.
41. build for people who don't own anything anymore. You rent your apartment, lease your car, stream your music, and subscribe to your software. Anything that helps regular people actually own an asset again wins.
KEEP BUILDING
Next for Palisec: we're writing a standard.
Today, if a wallet, a lending app or an agent marketplace wants to know whether an AI agent is safe to deal with, it has to take the agent team's word for it.
We've started work on a draft ERC for policy-enforced agent accounts: one common way for any app to read, straight from the chain:
→ is this agent's account protected by an external policy signer?
→ which policy is active right now, and is a looser one waiting to apply?
→ why was a transaction blocked, with standard reason codes every app understands
→ and a signed receipt for every decision, that anyone can verify
It won't be just a document. It ships with a working reference implementation on Robinhood Chain mainnet, used by Palisec's own accounts, so anyone can test it against real agents.
Step one is research: what ERC-4337, 1271, 7579, 7710 and 8004 already cover, so we build on them instead of reinventing them.
Every Palisec feature so far went live only after it was proven on mainnet. This one will too.
Aravind is right that someone has to decide who's responsible when an AI agent with access to your money gets hacked.
But there's a question one step earlier: why does the agent have unlimited access in the first place?
His earlier idea, funds that can't move unless the owner verifies in person, is how agent accounts should work by default. Not as a feature you ask your bank for. As the way the account is built.
On Robinhood Chain, that's what Palisec does for onchain agents:
→ the agent's key can't move funds alone. Every transaction needs Palisec's co-signature
→ Palisec only co-signs inside the limits the owner set: amounts, contracts, tokens
→ above a threshold you choose, a human has to approve
→ one switch freezes every agent at once
→ and the owner keeps a recovery key, so we never hold the funds
Liability rules decide who pays after a hack. Limits decide how much there is to lose.
https://t.co/gKqporWBia
Adding to the post attached, @RBI must also immediately create a policy around the use of AI agents in operating Indian bank accounts.
If a customer uses AI agents to access and work with their account, the bank must not be held responsible for hacks or issues arising out of it.
It must work something like this:
- Indian banks must be asked to build measures to detect the use of AI agents accessing accounts and flag the customer.
- The bank, at this moment, does not know if the customer is using the AI agent themselves or it is an unauthorized agent.
- So they must suspend operations on the account and send a notice to the customer informing of the same.
- The bank must ask them to accept an agreement waiving liability if they chose to use AI agents.
- If the customer doesn't agree, then the bank must block all AI agents accessing the accounts of the customer.
In fact, it may be a good idea to temporarily block all AI agent access to all Indian bank accounts until better measures and security is in place.
Just heard of someone's account getting drained because of an AI agent they had deployed themselves. Now, why should deposit insurance or the bank be responsible for this?
@nsitharaman@GovSMalhotra@sanjeevsanyal@NITIAayog
Why holder mandates matter if you're launching an agent token.
Right now, an agent token is a promise. The team says holders have a say. The agent runs on a key the team controls. If the team changes its mind, or the key gets taken, holders find out on the explorer.
With a Palisec holder mandate, the promise becomes a rule:
→ The owner sets hard maximums once: the most the agent can ever spend, the contracts it can ever touch, the lowest approval threshold.
→ Holders vote with the token inside those limits. Voting power is fixed at a snapshot block, so it can't be bought mid-vote.
→ The vote that passes becomes the agent's live policy. Palisec's co-signer won't sign anything outside it.
→ Any loosening waits out a public delay. Any tightening applies at once.
→ Every vote, cancel and decision sits on a public page, and every vote can be recounted from the chain.
For holders: you can check what the agent is allowed to do, live, before you buy.
For teams: "holders decide" stops being marketing.
Our demo mandate is live on Robinhood Chain mainnet: https://t.co/bEibP9tXMo
Building an agent token on Robinhood Chain? Our DMs are open.
What happens to your agent’s funds if Palisec disappears tomorrow?
Nothing. They stay yours.
A Palisec agent account is a standard Safe with three owners:
→ your agent’s key
→ the Palisec co-signer
→ you, the human recovery owner
Any transaction needs two of the three.
Day to day, the agent and Palisec sign together, and Palisec only signs what your policy allows. That’s what stops a fooled, buggy or compromised agent from draining itself.
But you and your agent’s key are also two of three. If our co-signer ever went offline, you can still move every token out of the account. No support ticket, no waiting on us, no permission from anyone.
And it works the other way too. Palisec alone can never move your funds: one signature is never enough.
We built it so you never have to trust us with custody. We enforce the rules. We never hold the keys to your money.
Every agent token says “holders decide.” Today, $PALISEC holders actually did, on Robinhood Chain mainnet.
The first holder vote on a Palisec agent just passed:
→ proposal: raise our demo agent’s per-transaction limit from 0.00001 to 0.0003 ETH
→ 1,800 $PALISEC for · 0 against · quorum of 1,000 reached
→ every vote a gasless signature, from three demo voter wallets holding real $PALISEC
Before the vote, the agent tried to spend past its limit. Palisec denied it (PS_BUDGET_001), and nothing reached the chain.
Why a holder vote here can’t be gamed:
→ Whales can’t vote an agent wide open. The owner sets hard maximums, and votes can only pick values inside them.
→ No flash loans, no buying votes mid-vote. Voting power is fixed at a snapshot block before voting opens.
→ Loosening waits. This vote raises a limit, so it’s sitting on a 60-minute delay the owner can cancel. Tightening votes apply instantly.
→ Anyone can recount. Every signature, weight and snapshot balance goes into one public vote file, checkable with a public RPC. No API key, no Palisec server.
33 of 33 end-to-end tests are passing. Next on mainnet: a transaction allowed only under the new limit, one blocked past it, and a vote that tightens it back.
Not live yet. We launch when every step is proven.