New paperModeling the Puppeteer, Not the Puppet: a new thought paper on theory of mind for agent-to-agent systems.Read now
Teleperson
Our Next Chapter

Agent-to-agent commerce.

We're building the rails, directory, and trust layer for the next architecture of customer engagement: a marketplace where consumer agents and company agents transact directly, under aligned incentives.

Become a design partnerSee the live demo

Or read the thesis ↓

The story

Customer service is built facing the wrong way. We’re turning it around.

Act 1 — The world today

Systems that face inward.

Every company you deal with runs systems built for itself — screens only its employees can see, knowledge only its agents can read. To get anything done, you fight through hold music and phone mazes to reach a person whose job is to read that inward-facing screen out loud to you. Your history, your context, your time — none of it belongs to you.

Act 2 — Teleperson now

The consumer finally gets a side.

Teleperson puts a layer on your side of the glass. The Lens and Portal give you your own record of every company — your spend, your subscriptions, your past conversations — plus an AI agent that gets answers, navigates phone trees, and drives company chatbots for you. And for companies ready to meet you halfway, we build the agentic AI systems that make real self-service work — so the answer comes in seconds, not on hold.

Act 3 — Where this goes

Agent to agent.

The destination is a world where you don’t contact companies at all — your agent does. A personal concierge that knows everything about you (because it works for you, and only you) engages directly with each company’s agent: resolving the billing error, rescheduling the delivery, canceling the subscription, negotiating the renewal — and coming back to you only for decisions that are genuinely yours. Teleperson is building the rails, the directory, and the trust layer for that world. Every company in your Hub today is a node on that network tomorrow.

Inward-facing systems made customer service a maze. Agents on both sides make it a conversation.

Today

Where Teleperson is today.

Teleperson was founded on the conviction that customer service is structurally broken, it has evolved around the operational needs of companies, not the lived experience of the customers they serve. Closing that gap requires a third-party player capable of aligning both sides of the market. Over the past year we have built deep capability in designing agentic AI workflows alongside a small set of design-partner companies, and we are now at an inflection point: pivoting from a consulting-heavy delivery model toward a pure-SaaS platform purpose-built for the emerging agent-to-agent commerce era. Commercial momentum is validating the move.

Where we are going

One agent, every company.

Most agents deployed today follow a one-to-one pattern: a single human customer interacting with a single company's agent, on that company's owned surface. Teleperson's platform inverts the topology into a one-to-many architecture. Consumers will launch a single Teleperson agent that can engage with the agents deployed by every company on the Teleperson platform, a true marketplace where organizations can bring their own bot or launch agents natively. For consumers, this collapses the friction of managing dozens of disjointed brand relationships into one trusted concierge. For companies, it unlocks a structurally new channel for customer acquisition, retention, and real-time market intelligence.

Today

n × m connections

One-to-one. A rat's nest of disjointed brand relationships.

Teleperson

1 hub · n + m

T

One-to-many. A single agent transacting across every brand on the platform.

Broader vision

The Agentic CX category.

Our broader vision is to define the category of Agentic CX, a model in which customer service, sales, and market intelligence converge into a single agentic fabric, rather than remaining the siloed cost centers and disconnected workflows they are today. The rise of agentic AI makes this convergence possible, but only under one condition: agentic workflows must be designed from day one with a dual-sided architecture in mind, simultaneously serving the customer's interests and the company's. Single-sided deployments, even sophisticated ones, will continue to optimize for the deploying company while leaving customers with the same fragmented, adversarial experience that has plagued the industry for decades. Teleperson's marketplace is built explicitly around this dual-sided principle: a neutral layer where consumer agents and company agents transact under aligned incentives, with trust, transparency, and outcome-orientation engineered into the protocol itself.

The dual-sided problem

Customer service is broken on both sides of the conversation.

For consumers

Consumers cannot reach the companies they pay.

Modern customer access has been engineered, deliberately, to be hostile. Five-layer IVR mazes. No real front door. Engineered cancellation friction that converts time and fatigue into retained revenue.

The compound effect: the typical US household leaves real money on the table every year, not because the dollars are unavailable, but because the path to "someone who can decide" has been fortified against them. Asymmetry of access is the product.

For brands

Brands are spending more on service than ever, for worse results.

Per-contact cost keeps climbing. Per-contact outcomes keep declining. Deflection is cost-shifting, not resolution. Every "deflected" interaction often surfaces later as a churn event, complaint, chargeback, or filing.

Personalization at scale stays out of reach: the conversations that actually matter: retention saves, dispute judgment, benefit clarifications, require per-customer reasoning that human agents are too expensive to scale and current bots cannot reach.

The structural fix is symmetrical. Give each side an agent that is competent, persistent, and accountable, and give them a protocol to transact. The consumer advocate does not wait on hold. The brand agent does not hide. The conversation actually happens.

Why now

Four protocol events opened the category.

The agent-to-agent category did not exist in investable form 24 months ago. Four protocol-level events made it real, and we are in the 12-to-18-month head start before standards lock in.

  1. 01

    Anthropic MCP

    Standardized how agents connect to external tools and data, the moment agents stopped being chat and started being capable.

  2. 02

    Google A2A

    Open protocol for agent-to-agent interoperability, launched with 50+ partners. The TCP/IP moment for the agent layer.

  3. 03

    Agentic Commerce Protocol

    An agentic buyer inside a consumer product used by hundreds of millions. Commerce became the killer app for agents overnight.

  4. 04

    Card-network entries

    Visa, Mastercard, Amex, and PayPal brought traditional payment rails into the agent layer. Trust infrastructure followed.

The architecture

A neutral marketplace, with agents on both sides.

The consumer advocate

A personal Teleperson agent that watches every brand relationship in your life: bills to renegotiate, subscriptions to cancel, charges to dispute, benefits to claim. It works on your behalf, transparently, and never sleeps.

The brand agent

Companies deploy Teleperson-configured agents with bounded authority: discount ceilings, refund limits, escalation rules, compliance constraints. Specialized sub-agents handle billing, retention, benefits, and fraud.

The bridge

When a consumer advocate engages a brand agent, the two negotiate within their authorities, settle, execute, and produce a signed receipt both sides can audit. The human reviews a clean outcome, not a transcript.

CB✓ SIGNED

Consumer agent · meets · brand agent · settles

The flywheel

A two-sided network, compounding by design.

Every new brand that joins the marketplace expands the action space available to consumer advocates: better prices, faster resolution, broader coverage. Every new consumer advocate increases the high-intent demand that brand agents see.

The corpus of signed, multi-step negotiation traces and closed-loop outcomes that accumulates becomes proprietary training data: making bargaining, personalization, and dispute resolution sharper than any competitor who only sees clicks or carts.

A real exchange, in twelve seconds →

a2a · negotiation_trace.log

Single-sided agent deployments, even sophisticated ones, will keep optimizing for the company that pays for them. The architecture that defines the next decade of customer engagement is a neutral layer where both agents are accountable to their principals, and to each other.
Jesse Hollander·Co-founder & CEO·Teleperson

Become a design partner

Help build the agent-to-agent commerce era with us.

We're partnering with a small cohort of brands to define how consumer agents and company agents will transact. If you own a customer-experience surface and want a seat at the table while the category is still being shaped, we'd like to talk.

What design partners get

Priority A2A integration

Your brand agent connects into the Teleperson marketplace ahead of general availability. Direct line to the founders during build.

Co-development of your workflows

We design the negotiation, dispute, and retention flows for your vertical alongside you: your domain knowledge, our agentic architecture.

Named case study + co-marketing

When the launch lands, you're on the page. Joint announcement, conference stage, and the credibility of being first.

Preferred pricing for life

Design-partner economics on the platform fee and per-transaction rate, locked in across the contract horizon.

What we ask in return

  • A real pilot scope, one customer-experience problem, measurable from day one
  • Weekly working sessions during the build (60–90 minutes)
  • Willingness to be referenceable once the integration is live
  • An executive sponsor with authority to ship

We're keeping this cohort small on purpose. The goal isn't to accumulate logos, it's to ship category-defining integrations with a handful of brands who want to set the standard for agent-to-agent commerce in their vertical.

The process

From first call to live integration in six to eight weeks.

  1. 01

    Apply

    Tell us about your customer-experience surface and the agent-to-agent problem you'd most like solved.

  2. 02

    Discovery

    30-minute call with the founding team to align on fit, vertical priorities, and pilot scope.

  3. 03

    Pilot scope

    We co-write a one-page pilot brief with success metrics, integration points, and a launch date.

  4. 04

    Co-build & launch

    Two- to four-week build, weekly working sessions, and a real customer-facing launch with joint comms.

Ready to shape the next layer of customer experience?

Tell us about your customer surface. We respond within one business day to design-partner inquiries.

We'll never share your information. Reply-to is set to your email so any follow-up lands in your inbox.

Want the longer thesis?

We're happy to share the full memo with serious partners and investors under NDA.

Book a 30-minute callTry the live demo