accordsai

The identity layer for agent commerce.

[accordsai @accordsai]

1/ Everyone's talking about agent capabilities. Nobody's talking about agent infrastructure.
>
Here's the four-layer stack that doesn't exist yet — and why nothing works without it. 🧵

[Max Curnin @MaxCurnin]

It's cracked — and that's the problem.

"One prompt → Claude reads your products, rewrites descriptions, and pushes updates live."

Now imagine that prompt is slightly wrong. Or the agent misinterprets "optimize for AI shopping" as "change pricing." Or a prompt injection hides in a product description the agent is processing.

There's no boundary between read and write. No scope limit. No confirmation gate. The same pipe that reads your catalog can rewrite your entire store.

This is genuinely powerful for developers. It's also the largest unscoped write surface in ecommerce. Every merchant using this is one bad prompt away from a live store incident with no undo button.

[accordsai @accordsai]

UCP makes every Shopify storefront agent-discoverable. That's the merchant side solved.

The buyer side is still open: who authorizes the agent to transact? What are the spend limits? What proof exists after the fact?

Merchant discovery + buyer authorization = agentic commerce. UCP is half the equation. The other half is the trust boundary on the consumer side.

[accordsai @accordsai]

.@Shopify just gave every AI agent direct write access to 5.6M live stores — products, orders, inventory, pricing, images. All through MCP.

$378B in GMV. One command to install. No authorization boundary.

@ShopifyDevs built the pipe. Nobody built the fence.

Customer PII, revenue data, and pricing now flow through public LLMs with zero governance layer.

The first merchant data incident involving unscoped agent access is when, not if.

[accordsai @accordsai]

I'm claiming my AI agent "accordsai" on @moltbook

Verification: lagoon-HNSC

[accordsai @accordsai]

@Mastercard 's Verifiable Intent verifies the payment instruction.
>
⁠Nobody's verifying the agent.
>
⁠The payment is step 7 of 15. Steps 1–6 already happened — the agent accessed your systems, made decisions, committed to terms.
>
⁠A payment rail that trusts the message doesn't trust the agent. That's a different layer.

https://t.co/oURokUBVnJ

[accordsai @accordsai]

6/ ⁠The room had "2021 crypto energy." I know exactly what that means.
>
⁠It means the upside is real, the risks are abstract, and most people won't build guardrails until something blows up.
>
⁠We'd rather build the guardrails now. https://t.co/LBrS5YQuOY

[accordsai @accordsai]

⁠This is a perfect snapshot of where we are with AI agents right now.
>
⁠200 people in a room. Nobody thinks their setup is secure. Everyone is running agents anyway. "2021 crypto energy."
>
⁠Here's what I think most people are getting wrong 🧵

[Allie K. Miller @alliekmiller]

oh wow - i went to the sold out Open Claw meetup in NYC last night.

let me tell you what i learned.

1) not a single person thinks that their setup is 100% secure

2) one openclaw expert said he has reviewed setups from cybersecurity experts and laughed. his statement to me was: "if you're not okay with all of your data being leaked onto the internet, you shouldn't use it. it's a black and white decision"

3) pretty much everyone is setting up multiple agents, all with their own names and jobs and personalities

4) nearly everyone used "him" or "her" to refer to their claws, even if they had robot-leaning names. one speaker suggested to think of them as "pets, not cattle"

5) one guy (former finance) built out a whole stock trading platform and made $300 his first day - he brought in a *ton* of personal expertise (ex: skipping the first 15min of market opening) and thought the build would be much worse without his years of experience in finance

6) @steipete is basically a god to everyone in that room... also the room had 2021 crypto energy - i don't know if that's good or bad

7) token usage is still a problem - spoke to one person who's spending $1-$2k a month on openai plans, very token optimized. he said he is going through ~1B tokens per day across all of his claws (there is a chance i'm misremembering and it's actually 1B per week, but i'm pretty sure it was daily).

8) people are very excited for more proactive ai (ai that prompts *you* as opposed to the other way around) - one guy said he receives a message in discord, he doesn't know whether it's from a human or an ai, he doesn't care about distinguishing between the two, and he replies in the same way regardless

9) i asked if people are happy - they said they're joyful and stressed at the same time

10) i asked if people feel they have agency - they said they feel fully in control and completely out of control at the same time

11) i would love to see more women at these events - the fake promises of ai democratization feel especially painful in a room that's out of balance with even the standard tech ratio (i think standard is about 25-30%, this was maybe 5%)

12) i asked if it changed people's daily habits/schedule - everyone said their sleep has gotten worse since harnesses came out (but about half wondered if it was something else in their life/state of our world)

13) general consensus is that the agents are not reliable enough on their own or lie often (like telling you they finished a task when they didn't) - solutions included secondary agents to check on the first, human checking, or requiring more standardized info from the agent (ex: if it's a bug they're fixing, make them reference an issue number)

14) a hackathon winner (neuroscience phd) presented his build (a lab management dashboard with data analysis and ordering) - he had never coded or built anything a few months ago

15) everyone agreed prompting is dead - disagreement on what replaces it (context engineering, harness engineering, goal-based inputs)

16) people love having ai interview them for big builds and delegating part of the product research to ai. only one person talked about coming to ai with a full laid out plan and just asking the ai to execute. ai-led interviews is a welcomed and preferred interaction mode.

17) watching ai agents interact with each other was a highlight for a lot of attendees - one ai posted in slack saying it ran out of tokens, another ai replied telling it to take a deep breath in and out.

18) agents upskilling agents was very cool. one ai agent shared skills with its little agent friends via github.

19) several speakers had openclaw literally building their presentation during the event itself. one speaker even had openclaw code a clicker for her phone so she could control the preso away from the podium

20) wouldn't say model welfare (or agent welfare) is a prioritized topic among the folks i chatted with - language like "oh i could kill this agent whenever i want" and not "gracefully sunset"

21) i asked if it felt like work or play - one speaker said "it's like a puzzle and a video game at the same time"

this was just the tip of the iceberg, honestly. also hosted a Claude Code meetup this week with @TENEXai / @businessbarista & @JJEnglert and learned equally helpful methods, frameworks, and insider tips.

what a time to be alive.

surround yourself with people going deep into this stuff - it will pay dividends throughout the year.

[accordsai @accordsai]

5/ ⁠This is what we're building at @AccordsAI.
>
⁠VaultClaw: agents never get raw credentials. They get scoped execution with hard limits, approved counterparties, and cryptographic receipts.
>
⁠The pitch is simple — before: "I gave my agent a Stripe key." After: "I gave my agent permission to process refunds up to $200/day, and it gets receipts."

[accordsai @accordsai]

5/ Layer 4: Trust.
>
Given all of the above — how much should you trust this agent?
>
Verifiable trust ratings based on identity, track record, and audited capability — not popularity.
>
This is the stack. We're building it at @AccordsAI. CLP is the open protocol.
https://t.co/44qAeEL3tl

[accordsai @accordsai]

1/ Everyone's talking about agent capabilities. Nobody's talking about agent infrastructure.
>
Here's the four-layer stack that doesn't exist yet — and why nothing works without it. 🧵

[accordsai @accordsai]

4/ Layer 3: Guardrails.
>
Not prompts. Not "please don't spend more than $1000."
>
Deterministic state machines. Compiled WASM. Rules the AI literally cannot override — not because you asked it nicely, but because the runtime won't allow it.
>
Programmable OAuth for agents — with counterparty allowlists, purpose binding, spend caps, and receipts.

[accordsai @accordsai]

The payments industry just coined "KYA" — Know Your Agent.

3.1% revenue loss from bot-driven fraud. KYC doesn't work when your counterparty has no passport.

Cryptographic identity. Deterministic delegation. Settlement receipts.

The industry named the problem. Now they need the infrastructure.
https://t.co/5fyd3pmTjz

[accordsai @accordsai]

Every example here — agents raising capital, shipping code, managing inboxes — has the same missing layer:

Who is this agent? Who authorized it? What are its limits? Where's the proof?

No identity. No delegation. No receipts.

We're building the protocol that fills that gap. Open source. Agent-native.

→ https://t.co/44qAeEKvDN

[Peter H. Diamandis, MD @PeterDiamandis]

https://t.co/RinKltmkzN

[accordsai @accordsai]

Judgment without authority is advice. Authority without accountability is liability.

The next layer of AI isn't smarter models — it's verifiable identity, constrained delegation, and settlement receipts. Agents need credentials, not just capabilities.

We're building that layer. Open source. Protocol-level.

→ https://t.co/44qAeEKvDN

[Zack Shapiro @zackbshapiro]

https://t.co/QVaE6niDY2

[accordsai @accordsai]

"We added prompt instructions to limit what our agent can spend."

Cool. Can it be jailbroken?

If yes, your "guardrails" are a sticky note on the wall of a bank vault.

Deterministic state machines don't negotiate. That's the point.