Resources

Papers, memos, and talks.

Original research and field notes from the Teleperson team and our advisors on the future of customer experience, agentic AI, and the agent-to-agent commerce era.

Compare

Teleperson vs Intercom, Decagon, Sierra, Ada, Maven, Forethought, Chatbase.

Side-by-side capability matrix and per-vendor deep dives — what each is good at, where Teleperson differs, and when to pick which.

FeaturedWhite paper

The Architecture of Agent-to-Agent Commerce: A Field Map

Most analysis treats 'agents' as a single thing. The agent commerce stack is actually five distinct layers, and conflating them produces bad strategy. A field map for the next decade of customer engagement infrastructure.

Teleperson Team · 2026 · 12 min read

Read the paper

Cover of The Architecture of Agent-to-Agent Commerce: A Field Map

Topics

17 papers

White paperTrust & safety
Grounding Promises in the Sandbox: An Environment-Grounded Commitment Protocol for Trained Autonomous Agents
The Verifiable Commitment Protocol can prove an agent made a promise, but not always decide whether the promise was kept. This paper extends VCP into the Environment-Grounded Commitment Protocol — using the sandboxed training environment itself as the trusted oracle that decides whether a commitment was discharged, and turns that verdict into an attributable training signal.
Teleperson Team · 2026 · 16 min read
Read the paper
White paperArchitecture
Lost in Translation: A Formal Semantics of Cross-Protocol Interoperability for Autonomous Agents
The agent ecosystem standardized on several interoperability protocols, not one, and production systems already bridge between them. This paper gives the first formal account of when a cross-protocol bridge preserves meaning — and proves that some interoperability is not merely unimplemented but impossible.
Teleperson Team · 2026 · 16 min read
Read the paper
White paperRegulation
Regulating Agentic AI and Agent-to-Agent Interaction
A comparative survey of agentic-AI regulation across the EU, US, UK, China, and other major jurisdictions: and the five distinct legal questions that autonomous, multi-agent systems raise that conventional AI law does not yet answer.
Teleperson Team · 2026 · 23 pages
Read the paper
White paperA2A commerce
The Architecture of Agent-to-Agent Commerce: A Field Map
Most analysis treats 'agents' as a single thing. The agent commerce stack is actually five distinct layers, and conflating them produces bad strategy. A field map for the next decade of customer engagement infrastructure.
Teleperson Team · 2026 · 12 min read
Read the paper
White paperRegulation
The Liability Question: Who Owns the Outcome When an Agent Acts?
When an agent commits a transaction the principal would not have authorized, the question of who is liable does not have a settled answer. A practical framework for assigning liability across the principal, the agent provider, the marketplace, and the counterparty.
Teleperson Team · 2026 · 8 min read
Read the paper
Industry researchCustomer experience
The Asymmetric Service Economy: Why Customer Support Is Broken on Both Sides
Customer service is the only major operational function where year-over-year spending consistently produces declining satisfaction. The structural reason has nothing to do with effort, and the fix has nothing to do with better chatbots.
Teleperson Team · 2026 · 10 min read
Read the paper
White paperA2A commerce
From One-to-One to One-to-Many: The Topology Inversion at the Heart of A2A
Most agentic AI shipping today repeats a single architectural mistake: a one-to-one pattern in which each consumer talks to each brand's agent on the brand's surface. The marketplace inversion changes which side of the relationship has leverage.
Teleperson Team · 2026 · 8 min read
Read the paper
White paperTrust & safety
Trust at Machine Speed: Identity, Receipts, and the Audit Layer Agent Commerce Needs
The technical chasm between an agent that demos well and an agent you would let act on your behalf is wider than the engineering teams realize. The trust layer is what closes it, and it is the layer most teams underbuild.
Teleperson Team · 2026 · 11 min read
Read the paper
Industry researchRegulation
After EU AI Act 2.0: How Agentic Workflows Will Be Regulated by 2027
The original EU AI Act was already obsolete on agentic systems by the time it cleared parliament. The amendments specifically targeting agent-to-agent interaction are arriving faster than the industry expects, and they will reshape the deployer's compliance burden.
Teleperson Team · 2026 · 9 min read
Read the paper
Case studyProcurement
The Procurement Gauntlet: What Fortune 500 Buyers Actually Need from Agentic AI
Agentic AI vendors that demo well to CEOs die in procurement. Four people the founders have never met decide whether the deal closes, and their concerns are categorically different from the ones the demo was designed to address.
Teleperson Team · 2026 · 7 min read
Read the paper
White paperPricing
Pricing Agentic Outcomes: The End of Per-Seat SaaS
Per-seat SaaS pricing was built for software that humans use. Agentic AI does not have seats. The pricing model that wins the next decade looks different, and the SaaS playbook does not survive the transition cleanly.
Teleperson Team · 2026 · 8 min read
Read the paper
TalkA2A commerce
The Connective Tissue Argument: Why Agentic Commerce Needs a Neutral Marketplace
A talk delivered to a private gathering of CX leaders. The argument: closed-platform plays in agentic commerce will lose to a neutral marketplace layer for the same reason walled-garden directories lost to the open web.
Teleperson Team · 2026 · Talk · 6 min read
Read the paper
White paperData moats
Negotiation Traces as Defensible Data: The Corpus No Clickstream Can Match
The defensible data moats of the next decade are not going to be clickstreams. They are going to be agent conversations. The structural reasons this matters for category formation and competitive dynamics.
Teleperson Team · 2026 · 8 min read
Read the paper
White paperA2A commerce
The Bring-Your-Own-Bot Pattern: Why Marketplaces Must Be Agent-Agnostic
Marketplaces that require participants to use a specific agent will fail to achieve liquidity. The agent-agnostic 'bring your own bot' pattern is structurally necessary, and it has implications for how the marketplace layer must be designed.
Teleperson Team · 2026 · 7 min read
Read the paper
Industry researchVertical analysis
The First Wave of Agentic Verticals: Banking, Telecom, Utilities, and Why They Adopt First
Agent-to-agent commerce will land in specific verticals first. Knowing which verticals, and why, separates teams building for the actual market from teams building for the vision deck.
Teleperson Team · 2025 · 9 min read
Read the paper
Industry researchPricing
The Consumer-Side Business Case: How Households Pay for Their Own Agent
The brand-side business case for agentic AI is straightforward; budgets exist. The consumer-side case is harder, and harder to fund. Three pricing models, the economics that drive them, and what each one requires to work at scale.
Teleperson Team · 2025 · 8 min read
Read the paper
Industry researchArchitecture
Agent-to-Agent Protocols: A Comparative Survey of MCP, A2A, ACP, and the Card-Network Entries
Eight competing standards now exist for agent-to-agent interaction. Treating them as fragmentation misreads what is happening: the standards are colonizing different layers of the stack, and the eventual settlement is already partially visible.
Teleperson Team · 2025 · 10 min read
Read the paper

Looking for the deeper memos?

Investors, design partners, and journalists can request access to our internal market briefings, financial models, and technical architecture documents.

Request access Subscribe to the Substack

Papers, memos, and talks.

Teleperson vs Intercom, Decagon, Sierra, Ada, Maven, Forethought, Chatbase.

The Architecture of Agent-to-Agent Commerce: A Field Map

Grounding Promises in the Sandbox: An Environment-Grounded Commitment Protocol for Trained Autonomous Agents

Lost in Translation: A Formal Semantics of Cross-Protocol Interoperability for Autonomous Agents

Regulating Agentic AI and Agent-to-Agent Interaction

The Architecture of Agent-to-Agent Commerce: A Field Map

The Liability Question: Who Owns the Outcome When an Agent Acts?

The Asymmetric Service Economy: Why Customer Support Is Broken on Both Sides

From One-to-One to One-to-Many: The Topology Inversion at the Heart of A2A

Trust at Machine Speed: Identity, Receipts, and the Audit Layer Agent Commerce Needs

After EU AI Act 2.0: How Agentic Workflows Will Be Regulated by 2027

The Procurement Gauntlet: What Fortune 500 Buyers Actually Need from Agentic AI

Pricing Agentic Outcomes: The End of Per-Seat SaaS

The Connective Tissue Argument: Why Agentic Commerce Needs a Neutral Marketplace

Negotiation Traces as Defensible Data: The Corpus No Clickstream Can Match

The Bring-Your-Own-Bot Pattern: Why Marketplaces Must Be Agent-Agnostic

The First Wave of Agentic Verticals: Banking, Telecom, Utilities, and Why They Adopt First

The Consumer-Side Business Case: How Households Pay for Their Own Agent

Agent-to-Agent Protocols: A Comparative Survey of MCP, A2A, ACP, and the Card-Network Entries

Looking for the deeper memos?