# Argorant — Verified B2B contact data, email verification and outbound campaigns, drivable by your own AI agent over MCP.

> Source: The Agents Index — https://theagentsindex.com/argorant (structured, researched, re-verified)
> Facts last verified: 2026-09-26

Argorant is a B2B contact database of 626.5M records that verifies work emails by live SMTP probe at reveal or export. It also sends outbound campaigns from your own mailboxes and exposes the same engine over MCP, REST and a CLI.

| Fact | Value |
| --- | --- |
| Website | https://argorant.com |
| Price | from $99/month billed monthly, $79/month billed annually |
| Free trial | Partly |

Confirming your email starts a 14-day trial with 100 credits and 300 campaign emails, no card. At day 14 campaigns pause and unused trial credits expire. Counts, searches and masked previews never consume credits; API rate is 30 requests a minute on the free start.

## What it does

**Done-for-you service** · Yes — Argorant Managed is a done-for-you engagement where their team buys domains, warms inboxes, writes and runs campaigns and works replies; it is quoted on a call rather than priced publicly.

> Done-for-you outbound, run as an ongoing engagement. Priced on your ICP and volume, in a 30-minute call.
> — https://argorant.com/pricing · Pricing page, Managed tab, 'Fully managed / Argorant Managed' card

## How much it does unattended

**Runs autonomously** · Partly — An agent can run the whole loop unattended for free actions, but every credit-spending step and every campaign launch is gated on explicit approval, and destructive tools are annotated so clients ask first.

> One credit per verified contact, only when you or your agent says go. Every spend logged.
> — https://argorant.com/ai-agents · Capabilities section, 'Reveal on approval' card

**Multi-agent** · Partly — Several agents can share one workspace's lists and credit balance with a key each, so spend is attributable per agent, but Argorant supplies no orchestration or hand-off between them.

> Attribute agent spend. Create one API key per agent or integration, and the activity trail tells you which one is expensive.
> — https://help.argorant.com/help/account-and-security/activity-tab · 'Reading it usefully' list, second bullet

**Agent permissions** · Yes — Access is scoped per capability, with free read scopes separated from credit-spending ones, tool allowlists the owner controls, and destructive tools flagged so the client prompts before sending email.

> Campaign tools act on the connected workspace only. Launching and replying send real email; those tools are annotated as destructive and open-world so clients ask for confirmation.
> — https://docs.argorant.com/docs/agents/security · 'Boundaries' bullet list, fourth bullet

## Which models it runs on

**Claude** · Yes — Claude is a first-class client with its own setup guide and a native remote MCP connector covering Desktop, Cowork and Code.

> Anything that speaks MCP: Claude (Desktop, Cowork, and Code), ChatGPT connectors, Cursor, Codex, Perplexity, Manus, OpenClaw, Hermes, and any custom runtime.
> — https://argorant.com/ai-agents · '01 Connect' panel of the tool-catalog section

**GPT** · Yes — ChatGPT connectors and OpenAI's Codex are both documented clients, with step-by-step setup against the hosted MCP endpoint.

> Use Argorant from ChatGPT: connect the MCP server, ask for any market, and get real counts, previews, and verified exports from 626.5M contacts.
> — https://argorant.com/integrations · Automation platforms grid, ChatGPT card

**Open models** · Partly — Any MCP-speaking runtime can connect, which admits an open-weight model you host yourself, but Argorant documents no open model specifically and runs no inference of its own.

> Anything that speaks MCP: Claude (Desktop, Cowork, and Code), ChatGPT connectors, Cursor, Codex, Perplexity, Manus, OpenClaw, Hermes, and any custom runtime. You decide which tools an agent may call.
> — https://argorant.com/ai-agents · '01 Connect' panel of the tool-catalog section

**Model choice** · Yes — The buyer picks the model by picking the client: Argorant is the tool layer behind one MCP endpoint and takes whatever agent you already pay for.

> Our entire search engine is exposed as a Model Context Protocol server. Connect Hermes, Claude, GPT, or any agent — let them search, filter, and push leads programmatically.
> — https://argorant.com/features · Features page, 'AI AGENTS — MCP server. Built for agents.' block

## Where you use it

**In your editor** · Partly — It reaches editor-side agents such as Cursor, Claude Code and Codex through the same remote MCP server, but there is no native IDE extension of Argorant's own.

> Anything that speaks MCP: Claude (Desktop, Cowork, and Code), ChatGPT connectors, Cursor, Codex, Perplexity, Manus, OpenClaw, Hermes, and any custom runtime.
> — https://argorant.com/ai-agents · '01 Connect' panel of the tool-catalog section

**On the command line** · Yes — A published npm CLI runs counts, searches, exports, campaigns and the inbox from a terminal, and needs no install to try.

> Yes — published on npm as 'argorant'. Run npx argorant count "..." with zero setup, or npm install -g argorant.
> — https://argorant.com/ai-agents · Agents FAQ, 'Is there a CLI?' answer

**In a browser** · Yes — A hosted web app at app.argorant.com holds search, lists, exports, campaigns, the unified inbox and the billing profile.

> Exports are verified at the moment of download. Credits are only spent when contacts are revealed or an export is created; every spend is logged.
> — https://argorant.com/ai-agents · '04 Export' panel showing the app.argorant.com/exports screen

## Whose machine it runs on

**Self-hosted** · No — The platform is vendor-operated on EU infrastructure and no self-hosted or bring-your-own-infrastructure option appears on any plan; only licensed bulk data can land in your own S3 or Snowflake.

> Our primary infrastructure is hosted in the European Union.
> — https://argorant.com/privacy · Section 9, International data transfers, first sentence

**Open source** · No — The terms reserve the software, models and database compilation to Argorant and grant no rights beyond using the service, so nothing is source-available.

> Argorant and its software, designs, documentation, compilation of the Database, models, workflows, and service content are owned by Argorant or its licensors and are protected by intellectual-property and database-protection laws.
> — https://argorant.com/terms · Section 10, Intellectual property, first paragraph

## What it costs to run

**How it meters** · Yes — Billing is per validated contact in credits, with a seat and campaign-email allowance on plans: counts and masked previews cost nothing and undeliverable rows are filtered before they are charged.

> 1 credit = 1 validated contact. 0.5 credit = 1 external email verification.
> — https://argorant.com/pricing · Plans tab, footnote under each plan card

**Free tier** · Partly — You can start without a card, but it is a 14-day trial rather than a standing free plan: campaigns pause and unused trial credits expire at day 14.

> When the 14 days are over, campaigns pause and unused trial credits expire; pick a plan to keep going, and everything you set up carries over.
> — https://argorant.com/pricing · Pricing FAQ, 'How does the free trial work?' answer

**API access** · Yes — A documented REST API is included on every paid plan and on pay-as-you-go, sharing one account and credit balance with the app, with rate limits rising by tier.

> Yes. The app, the REST API, the CLI and your AI agent share one account and one credit balance.
> — https://argorant.com/pricing · Pricing FAQ, 'Is API and MCP access included?' answer

**MCP server** · Yes — A hosted streamable-HTTP MCP server with OAuth 2.1 and PKCE exposes search, counts, previews, reveals and exports as tools another agent can call.

> OAuth 2.1 authorization code with PKCE. Clients that cannot be pre-registered self-register at the registration endpoint (dynamic client registration). Refresh tokens are issued.
> — https://docs.argorant.com/docs/agents/connect · Connect a client, paragraph under the Endpoints table

**Bring your own key** · Partly — You bring your own agent and therefore your own model subscription, and your own Google Workspace or Microsoft 365 mailboxes do the sending; there is no key or cloud choice for Argorant's own hosted assistant.

> Connecting your own Google Workspace or Microsoft 365 mailboxes is included on every plan, unlimited, as is warm-up.
> — https://argorant.com/pricing · Pricing FAQ, 'Are mailboxes included?' answer

## Buying it for a team

**A company can buy it** · Yes — Plans are bought self-serve by a workspace owner who invites teammates onto shared billing, and a signed DPA is offered on any paid plan for compliance sign-off.

> A signed DPA is typically required by enterprise and EU-based buyers for compliance sign-off. We're happy to provide one for any paid plan.
> — https://argorant.com/dpa · DPA page, third paragraph

**Seat model** · Yes — Each plan includes a named seat count from 1 to 20 with extra seats priced per user per month, and a seat is only consumed when an invite is accepted.

> A seat is only consumed when the invite is accepted. Pending invites are free, which means you can invite the whole team and let them join at their own pace.
> — https://help.argorant.com/help/account-and-security/invite-teammates · 'Inviting somebody' section, closing paragraph

**Pooled budget** · Yes — One workspace balance is shared: teammates draw on the same plan credits, lists and exports, and a contact revealed once is not charged again for anyone else.

> A workspace is one shared surface: teammates share the plan, the credits, the lists and the exports. That is the point. A contact revealed by one person is revealed for everyone, so nobody pays twice for the same lead.
> — https://help.argorant.com/help/account-and-security/invite-teammates · Opening paragraph

**Admin controls** · Yes — An owner constrains the agent directly: workspace-scoped keys, per-tool allowlists, masked previews by default and Member versus Manager roles, with per-user feature grants for the assistant and verifier.

> Keys and OAuth grants are workspace-scoped. You control which tools an agent may call, plan limits cap throughput, previews are masked by default, and every action lands in an audit log your admin can review.
> — https://argorant.com/ai-agents · Agents FAQ, 'How is access controlled?' answer

**Audit log** · Yes — A Profile Activity tab records reveals, exports, verification runs, sign-ins and key creation with cost and actor, and the vendor states no spending path is unlogged across app, API, CLI or MCP.

> Every credit movement is logged. There is no unlogged spending path, in the app, the API, the CLI or MCP.
> — https://help.argorant.com/help/account-and-security/activity-tab · Closing line of 'Cost control that is better than watching'

**Single sign-on** · No — Sign-in is email and password or Google only; no SAML, OIDC enterprise connection or SCIM provisioning appears on the pricing page, in the docs or in the help centre's account and security section.

> Go to app.argorant.com/signup and register with email + password, or continue with Google.
> — https://help.argorant.com/help/getting-started/create-your-account · Numbered signup steps, step 1

## What happens to your code

**Opt out of training** · Partly — Uploaded lists and CRM records are contractually kept out of the sellable database unless you opt into the Chrome-extension contribution programme, but no clause anywhere states whether customer data trains Argorant's own models.

> Apart from this opt-in program, we do not incorporate your Customer Data, CRM records, or uploaded lists into the Database.
> — https://argorant.com/terms · Section 9, Your data and contributions, third paragraph

**Data residency** · Partly — Primary infrastructure sits in the European Union, but transfers to non-adequate countries and a US corporate entity are covered by SCCs rather than avoided, and no region is selectable by the buyer.

> Our primary infrastructure is hosted in the European Union. When personal data is transferred to countries that have not received an adequacy decision — for example to service providers or in connection with our US-based corporate entity — we rely on appropriate safeguards
> — https://argorant.com/privacy · Section 9, International data transfers

**Getting out** · Yes — Exports and saved lists stay downloadable as CSV from the app or CLI, account data is supplied on request in a portable format, and prepaid credits survive cancellation for 365 days.

> Your exports. Every verified CSV you have created stays on Exports and can be re-downloaded at any time, including from the CLI with argorant export download.
> — https://help.argorant.com/help/account-and-security/export-your-data · Bulleted list of what you can download yourself, first bullet

**Certifications** · Partly — GDPR, UK GDPR and CCPA are addressed with a standard DPA, SCCs and documented safeguards, but no SOC 2 or ISO 27001 certification is claimed anywhere on the site, docs or help centre.

> Our DPA governs how Argorant processes personal data on your behalf under GDPR, UK GDPR, and the CCPA.
> — https://argorant.com/dpa · Legal / Data Processing Agreement page, standfirst

## Provenance

Every fact above comes from the vendor. Nothing here is independently corroborated yet.

## Coverage elsewhere

None recorded. Every fact above comes from the vendor and is not independently corroborated yet.

## What changed

- 2026-09-26 — Capabilities · Commercial terms · Free tier · Price

## Correcting this record

Anything marked "Not established" is a question we have not answered — it is not a No. If you represent Argorant, or anything here is wrong or out of date, you can supply it with a source: the operations this site accepts from an agent are listed at https://theagentsindex.com/.well-known/agents.json — that catalogue needs no key to read, and each entry states its own auth.

Supplying evidence is free. Nothing that can be bought here changes what a field says, whether a gap is closed, or how this record is scored.