- the analysis game
- Posts
- The Three Missing Pieces (Every Protocol Lacks).
The Three Missing Pieces (Every Protocol Lacks).
The Critical Gaps That Will Make or Break Your Agent Implementation.
Gabriel Botsie
August 20, 2025 • Estimated Reading Time: 4 minutes
The Three Missing Pieces.
None of the 14+ agent protocols we've covered can actually deliver on their full promise yet. Industry expert Laurie Voss identified three critical gaps that every current protocol lacks.
These aren't minor features - they're fundamental requirements for any serious agent deployment.
Missing Piece 1: Universal Discovery
Right now, there's no way to search the entire internet and say "I need an agent that can handle financial calculations" and get back a list of available options.
You can only discover agents if you already know where they are. It's like having a phone system where you can only call people whose numbers you already have.
Business Impact: Your agents can't find new capabilities or services automatically as your needs evolve.
The Core Questions:
How do agents get permission to use external services?
How do they pay for computational resources or premium APIs?
The blockchain crowd thinks they have the answer, but most enterprise teams aren't convinced that blockchain complexity is justified
What This Means: Early implementations need custom solutions for managing permissions and costs. Budget for additional complexity.
Missing Piece 3: Reputation Systems
In a world where agents can discover each other automatically, how do you know which ones are trustworthy?
Real Scenario:
Your procurement agent finds three supplier agents offering the same service. How does it choose the reliable one versus the potentially malicious one?
To recap, there’s no standardised way to evaluate:
Agent reliability
Performance history
Security credentials
Why It's Hard: The protocol developers are "explicitly punting" on this because reputation systems are notoriously difficult to implement.
Real-World Impact: Financial Services
The reputation gap isn't theoretical. Sean Neville from Catena Labs (co-founder of Circle/USDC) puts it bluntly: "there's no real way to do 'know-your-customer' on agents."
Financial institutions need agents to prove they're dealing with licensed entities. As Neville explains: "There needs to be a way for an agent to say, 'this is who I am as an agent, here's my identity, my risk and who I am operating on behalf of…That verifiable identity in a way all these different agentic frameworks can understand would be key."
This is why banks are sticking to internal agents rather than open exchanges - they can't verify external agent credentials or audit trails.
What Your Early Agent Implementations Will Need:
Internal agent registries
Human approval for external integrations
Vetted partners only
Budget Reality
Implementation costs will be higher than initial estimates suggest. Factor in additional infrastructure for handling these missing capabilities.
Next week: How to evaluate protocols for your specific needs despite these limitations.
Reply