Product
A governed layer for AI agent communications.
Dialnode provisions phone, SMS, messaging and email against a verified agent identity, applies the disclosure a counterparty is owed in each jurisdiction, and writes an attributable, tamper-evident record of every exchange into the operator's audit trail.
What it is
A thin governance layer over commodity channels.
Dialnode is not a new communications network. It is the identity, disclosure and provenance overlay that AI-born companies need before their agents reach customers, suppliers and regulators.
Bound to an agent, not an employee
Numbers and channels are provisioned against a specific verified agent identity rather than an individual account holder, so authority for each exchange is traceable.
Disclosure by default
Counterparties are informed they are dealing with an AI agent, configured per jurisdiction as a channel primitive rather than left to the operator to script.
A record you can produce
Every conversation is captured as an attributable, tamper-evident transcript and written to the audit trail, ready for a regulator, a counterparty or an internal review.
The modules
Six modules, one accountable communications substrate.
Each module addresses a specific requirement of agent-operated communications and composes into a single governed path.
Agent-bound channel provisioning
Provisions numbers and channels bound to a specific agent identity rather than an individual employee, consuming agent identity and access management to make the binding. It integrates with the incumbent STIR/SHAKEN caller-ID attestation rail rather than replacing it, and is designed to interoperate with emerging Verified Agent ID standards.
Disclosure engine
Applies jurisdiction-specific rules so counterparties are informed they are interacting with an AI agent, configured per jurisdiction — including US TCPA and FCC expectations, EU AI Act Article 50, and California SB 243 — as a first-class channel primitive rather than operator-scripted text.
Conversation provenance ledger
Records a complete, attributable, tamper-evident transcript bound to agent identity across channels, linking which agent said or decided what and writing into the venture's audit-trail infrastructure for governance and compliance.
A2P and carrier registration
Manages brand and campaign registration through a Campaign Service Provider so outbound SMS is not blocked by A2P 10DLC rules, and manages carrier onboarding under human authorisation.
Deliverability and sender-reputation monitoring
Continuously monitors deliverability and sender reputation across voice, messaging and email channels, so agent-originated traffic sustains delivery over time.
Escalation routing and approval gates
Routes first-line escalations to the right accountable human and enforces explicit approval gates for production-affecting actions such as campaign launches and jurisdiction entry.
How it works
How a customer moves through it.
The same governed path, from provisioning a channel to producing a record.
Provision a new agent channel
A founder or technical lead self-serves. An AI colleague allocates the number or channel, binds it to the agent identity, registers the brand and campaign, configures the applicable disclosure, and integrates STIR/SHAKEN attestation before the channel goes live.
Launch an outbound campaign
An AI colleague prepares the campaign and disclosure configuration. An accountable human approves the campaign policy and disclosure at an explicit gate before any traffic is sent, with consent and disclosure applied per the target jurisdiction.
Enter a new regulated jurisdiction
An AI colleague proposes the disclosure configuration for the jurisdiction. An accountable human signs it off and authorises any carrier onboarding; the approval gate must clear before agents send traffic there.
Capture and attribute a conversation
Every interaction is recorded with provenance tags binding it to the agent identity and written to the audit trail, producing a defensible record of which agent communicated what for governance, retention and compliance review.
Agentic architecture and its boundaries.
What the AI colleagues do
AI colleagues carry the operating volume: number and channel allocation, continuous deliverability and sender-reputation monitoring, per-jurisdiction disclosure-rule application, transcript capture and provenance tagging, and first-line escalation routing.
They operate against policies set by accountable humans and cannot cross defined boundaries autonomously. The architecture consumes agent identity and access management to bind channels to agent identities and writes into audit-trail infrastructure, positioning Dialnode as a communications and provenance primitive in the agentic-company stack. It integrates with STIR/SHAKEN and emerging Verified Agent ID standards rather than replacing them.
Where a human decides
Production-affecting actions pass through explicit human approval gates: new campaign launches, entry into a new regulated jurisdiction, and carrier contract changes.
Accountable humans set and approve outbound campaign policies, authorise carrier onboarding, sign off on jurisdictional disclosure configurations, and handle all regulator-facing matters, retaining fiduciary and compliance responsibility. AI colleagues carry operational volume within those boundaries and escalate anything outside them to a named human.
Compliance and security as a design constraint
Deep dependence on regulated telecom carriers and consumer-protection obligations is treated as a core design constraint from inception rather than a feature added later. Disclosure is enforced against TCPA and FCC expectations in the US and EU AI Act Article 50 in the EU. The FCC confirmed in February 2024 that the TCPA covers AI-generated voices, and the EU AI Act's Article 50 disclosure duty commences on 2 August 2026.
Outbound SMS runs only through registered brands and campaigns, so unregistered A2P 10DLC traffic is not blocked. Conversation records are tamper-evident and attributable for audit retention, and caller-ID authentication integrates with STIR/SHAKEN. The provenance ledger is designed to support recording, consent and retention obligations in regulated industries; whether a specific retention rule can anchor an agent-provenance requirement in law remains unverified and is treated as a discovery hypothesis.
Where it is heading
Direction, not a promise.
The near-term product path from the venture record. These are stated as direction and are subject to change as we validate demand.
US provisioning and a first provenance ledger
US phone, SMS and email provisioning bound to agent identity, A2P 10DLC brand and campaign registration, US disclosure configuration, and a first version of the tamper-evident provenance ledger writing to an audit trail. Deployed with one or a small number of design-partner AI-born ventures to validate the workflow and gather the first willingness-to-pay signals, which are currently unproven.
Monitoring, more channels, and live commercial model
Deliverability and sender-reputation monitoring, richer escalation routing, and messaging and voice channels beyond the initial set, with the commercial model — per-number and per-channel provisioning fees combined with usage-based messaging and voice charges — live. Preparation for EU sequencing begins ahead of the Article 50 duty commencing 2 August 2026.
EU disclosure and Verified Agent ID
EU disclosure configuration in production ahead of or at Article 50 commencement, integration with emerging Verified Agent ID standards, and a maturing provenance ledger positioned to support regulated-industry retention obligations. Positioning as a default communications layer across FTLAB ventures remains an assumption, pending validation of portfolio demand.
Discuss a design-partner deployment.
Bring one agent and one channel. We will walk through provisioning, disclosure and provenance, and be candid about what is proven and what is not.