Question 1

Are these real agents or just product names?

Accepted Answer

Real agents. Luna, Arc, Arbiter, and deepeye are production AI systems running in deepidv. Each one owns a discrete role across the verification and compliance lifecycle, with its own model stack, training data, and audit trail.

Question 2

How are the four agents different from each other?

Accepted Answer

Luna is the verification agent: she verifies anyone from any LLM or the deepidv platform and builds verification workflows. Arc orchestrates the workflows that decide which checks run when. Arbiter is the senior compliance manager that monitors users and transactions after sign-up, files SARs, and synthesizes signals into a risk score. deepeye is the deepfake and synthetic media defender that watches every face, voice, and document.

Question 3

Do I have to use all four agents, or can I pick one?

Accepted Answer

You can start with one. Most teams begin with Arbiter for compliance and risk or Luna for verification, then layer in the others as they expand. Each agent has its own pricing and can run independently or as a coordinated team.

Question 4

Where do humans sit in the loop?

Accepted Answer

Everywhere it matters. Arbiter drafts SAR narratives and makes the risk call, but a compliance analyst reviews and signs, and a human can override any decision, which trains the next one. deepeye flags synthetic media but escalates ambiguous cases to a fraud reviewer. Luna runs verifications and hands senior decisions up to Arbiter. The agents do the volume; humans do the judgment.

Question 5

Are the agents trained on my data?

Accepted Answer

Only with your permission. By default the agents run on deepidv's foundation models, fine-tuned on adversarial verification data we generate ourselves. Enterprise customers can opt into private model fine-tuning on their own data, isolated to their tenant.

Question 6

How fast do they respond?

Accepted Answer

Arbiter returns a composite risk decision in under 150 milliseconds. deepeye scans a face or voice clip in under a second. Luna answers verification questions in real time and sends verifications in under a minute. Arc routes workflows on every event without adding noticeable latency.

Question 7

What happens if an agent gets it wrong?

Accepted Answer

Every decision is logged, explainable, and replayable. If Arbiter scores something incorrectly, the override is captured and the model learns. If deepeye misses a deepfake, the case is added to adversarial training. The agents are not black boxes; every output ships with the reasoning behind it.

Question 8

How do I hire them?

Accepted Answer

Book a demo. We will walk through your verification and compliance flow, show which agents fit where, and get you set up with API keys and a dashboard. Most teams are running at least one agent in production within a week.

Features	Starter	Growth	Scale	Enterprise
Credits	30,000	200,000	1,000,000	Custom commit
Overage	$0.012	$0.009	$0.0055	$0.003–0.0045
Seats	5	15	50	Unlimited
SLA	99.5%	99.9%	99.95%	99.99% + custom
Support	Community	Email <24h	Slack + CSM	Dedicated SE + CSM
SSO	—	—
SOC 2 artifacts	—	—
Custom deploy	—	—	—

Hire your AI compliance team.

Four specialists working as one team.

Four roles. Zero overlap.

Luna

How the team works together.

01 · Onboarding · Arc

02 · Authentication · deepeye

03 · Decisioning · Arbiter

04 · Lifetime · Luna

An AI agent is yournext compliance hire

Four tiers. Every tier ships with the full agent suite — Luna, Arc, Arbiter, and the deepeye SDK. You only choose how much credit and which support depth you need.

Starter

Growth

Scale

Enterprise

Hire the team in a single afternoon.

Let's answer your questions