Loading video...

Video Failed to Load

Go Home

Today, we're open-sourcing the Personal Agent Consent & Trust Protocol (PACT), co-developed by Instinct. We're also joining the Personal Agent Protocol working group by Muse. Built on A2A and OAuth, PACT enables businesses to verify who a personal agent is acting for and what the customer has authorized it...

140,472 views • 21 days ago •via X (Twitter)

35 Comments

Decagon's profile picture
Decagon21 days ago

@Muse Read more:

Decagon's profile picture
Decagon21 days ago

@Muse Code and spec:

matt.mew.cog's profile picture
matt.mew.cog21 days ago

@instinct @Muse Why is it always a flight booking demo

Harry Gao's profile picture
Harry Gao21 days ago

@instinct @Muse 🔥🔥

Farhan Afsahi's profile picture
Farhan Afsahi21 days ago

@instinct @Muse Open source is the way to go

Tyler Xiao's profile picture
Tyler Xiao21 days ago

@instinct @Muse Excited for the launch!

Jeremy Bosma's profile picture
Jeremy Bosma21 days ago

@instinct @Muse can you revoke the authorization mid task

Jesse Zhang's profile picture
Jesse Zhang21 days ago

@instinct @Muse ❤️

Jules Polonetsky's profile picture
Jules Polonetsky21 days ago

@instinct @Muse very interested in how the privacy, consumer intent issues are handled

Ivan Zhou's profile picture
Ivan Zhou21 days ago

@instinct @Muse Amazing to see this go live!

frank's profile picture
frank21 days ago

@instinct @Muse please tell me this is the same or at least complementary to the one that came out today. nothing like two competing protocols dropping on the same day.

Rowan Trollope's profile picture
Rowan Trollope21 days ago

@instinct @Muse Super cool

erik.eth 🛡's profile picture
erik.eth 🛡21 days ago

@instinct @Muse Are PACT and PAP the same thing? They seem to have similar goals. Will the converge? Congrats on the release, the spec is interesting

Divyesh's profile picture
Divyesh21 days ago

@instinct @Muse @agentmail 🫨

Dhairya's profile picture
Dhairya21 days ago

@instinct @Muse @0xVatsaShah

Ajay Krishnan's profile picture
Ajay Krishnan21 days ago

@instinct @Muse The future of agents is more agent friendly interfaces and protocols that help regulate these interfaces and it’s so so exciting to see where we’re going ❤️

Alex's profile picture
Alex21 days ago

@instinct @Muse PACT solves verification but the harder problem is revocation. what happens when a user wants to pull consent mid-transaction an agent already started?

Batuhan Cakir's profile picture
Batuhan Cakir21 days ago

@instinct @Muse Why not all collaborate on one protocol ? In half a year there will be another Protocol to solve the same problem.

Anderson's profile picture
Anderson21 days ago

@instinct @Muse open sourcing a trust protocol for agents is a smart move for everyone

Brandon Kase's profile picture
Brandon Kase21 days ago

oh interesting, why have both PACT and Personal Agent Protocol? We actually just had a discussion about personal agents: how they work, when to use what, things to think about when building a specialized one. I think understanding details of personal agents will help you use them more effectively.

Martin Ronfort's profile picture
Martin Ronfort21 days ago

@instinct @Muse The shift toward open standards like this is exactly what the industry needs to move past brittle integrations. We actually went deeper on this here:

Vedant Sabadra's profile picture
Vedant Sabadra21 days ago

@instinct @Muse PACT nails consent. We built AMP to host Brands as a PACT Provider and add the multi-hop mesh on top (delegation chains, spend limits, erasure that travels with the message). Complementary, not competing. pip install ampro ·

Jevica's profile picture
Jevica21 days ago

@instinct @Muse Love seeing this kind of work on agent trust and consent. Getting the standards right now will make personal agents way more useful down the road 👑

Diavlos.sh's profile picture
Diavlos.sh21 days ago

@instinct @Muse

Sparsh Jain's profile picture
Sparsh Jain21 days ago

@instinct @Muse How do we join this group? We are building a personal assistant as well in India and would love to contribute in this accord.

Brjan | AI Builder's profile picture
Brjan | AI Builder21 days ago

@instinct @Muse how does PACT address privacy concerns in agent verification specifically?

EKOS _ OMiD🇮🇷's profile picture
EKOS _ OMiD🇮🇷21 days ago

PACT solves an important part of the trust problem: knowing who the agent is acting for and what it’s authorized to do. The next layer gets interesting after the authorization. Can we reconstruct the action itself — the context, decision, outcome, and evidence behind it? Authorization tells you an agent was allowed to act. Provenance tells you what actually happened and why.

Alvarich's profile picture
Alvarich21 days ago

@instinct @Muse Good to see some guardrails when all of your data is been in use.

Hnss's profile picture
Hnss21 days ago

@instinct @Muse PACT solves “who does this Agent act for.” Handing over a card is simpler for that,

Paul Walsh's profile picture
Paul Walsh21 days ago

@instinct @Muse Does that require access to iMessage content on any level, at anytime, for any reason?

steve 🎣's profile picture
steve 🎣21 days ago

@instinct @Muse 👀

Aina Ai | Tools & Updates's profile picture
Aina Ai | Tools & Updates21 days ago

@instinct @Muse Big step for trust setting the standard for personal agents

Furkan+'s profile picture
Furkan+21 days ago

@instinct @Muse Every company on earth should get onboard yet still most dont even offer mcps

Aaron Edell's profile picture
Aaron Edell21 days ago

@instinct @Muse Another step, today alone, in the right direction for unblocking agents from actually doing things on the web

benjamin kessler's profile picture
benjamin kessler21 days ago

@instinct @Muse This is great!

Related Videos

New Course: ACP: Agent Communication Protocol Learn to build agents that communicate and collaborate across different frameworks using ACP in this short course built with IBM Research's BeeAI, and taught by Sandi Besen, AI Research Engineer & Ecosystem Lead at IBM, and Nicholas Renotte, Head of AI Developer Advocacy at IBM. Building a multi-agent system with agents built or used by different teams and organizations can become challenging. You may need to write custom integrations each time a team updates their agent design or changes their choice of agentic orchestration framework. The Agent Communication Protocol (ACP) is an open protocol that addresses this challenge by standardizing how agents communicate, using a unified RESTful interface that works across frameworks. In this protocol, you host an agent inside an ACP server, which handles requests from an ACP client and passes them to the appropriate agent. Using a standardized client-server interface allows multiple teams to reuse agents across projects. It also makes it easier to switch between frameworks, replace an agent with a new version, or update a multi-agent system without refactoring the entire system. In this course, you’ll learn to connect agents through ACP. You’ll understand the lifecycle of an ACP Agent and how it compares to other protocols, such as MCP (Model Context Protocol) and A2A (Agent-to-Agent). You’ll build ACP-compliant agents and implement both sequential and hierarchical workflows of multiple agents collaborating using ACP. Through hands-on exercises, you’ll build: - A RAG agent with CrewAI and wrap it inside an ACP server. - An ACP Client to make calls to the ACP server you created. - A sequential workflow that chains an ACP server, created with Smolagents, to the RAG agent. - A hierarchical workflow using a router agent that transforms user queries into tasks, delegated to agents available through ACP servers. - An agent that uses MCP to access tools and ACP to communicate with other agents. You’ll finish up by importing your ACP agents into the BeeAI platform, an open-source registry for discovering and sharing agents. ACP enables collaboration between agents across teams and organizations. By the end of this course, you’ll be able to build ACP agents and workflows that communicate and collaborate regardless of framework. Please sign up here:

Andrew Ng

105,839 views • 1 year ago