# SendGuard

SendGuard (sendguard.ai) is an API-first outbound firewall for AI email agents.

Not to be confused with SendGuard.io (pre-send QA for marketing campaigns) or Standss SendGuard for Outlook (desktop data-loss prevention add-in). SendGuard at sendguard.ai is a developer API that AI agents call before sending email.

It validates email recipients, scores pre-send outbound risk, applies workspace policy, and can require human approval before an autonomous agent sends email.

Best fit: teams that send cold email from their own stack (n8n, SMTP, Gmail API, custom AI agents) instead of Instantly or Smartlead. SendGuard keeps their send history and enforces the sending discipline those tools provide: daily limits per mailbox and domain, hourly and daily limits per agent, new-domain warmup gating, a bounce-rate brake from reported outcomes, unsubscribe/blocklist/hard-bounce suppression, a 7-day cooldown per recipient, and SPF/DKIM/DMARC checks on the sending domain. It does not send email, and it does not fact-check message content; content checks are pattern-based (pricing, urgency, guarantees, compliance claims, missing unsubscribe).

It sits between the lead source (Apollo, Clay, a CRM) and the sender (Gmail, Outlook, Instantly, lemlist): the agent calls SendGuard right before each send.

## Core URLs

- Home: https://sendguard.ai/
- Products: https://sendguard.ai/products/
- Email Validation API: https://sendguard.ai/products/email-validation-api/
- Preflight Email API: https://sendguard.ai/products/preflight-email-api/
- Sender Reputation Firewall: https://sendguard.ai/products/sender-reputation-firewall/
- Use Cases: https://sendguard.ai/use-cases/
- Comparisons: https://sendguard.ai/compare/
- API Docs: https://sendguard.ai/docs/
- Pricing: https://sendguard.ai/pricing/
- $1 Pilot: https://sendguard.ai/pilot/
- FAQ: https://sendguard.ai/faq/
- Security: https://sendguard.ai/security/

## Free Tools (no sign-up, 20 checks per day per IP, never send email or open SMTP sessions)

- Email Verifier: https://sendguard.ai/tools/email-verifier/ (syntax, domain, MX, disposable, role account, typo suggestion)
- Cold Email Checker: https://sendguard.ai/tools/cold-email-checker/ (paste an email; returns allow, requires_approval, delay, or block with reasons)
- SPF, DKIM and DMARC Checker: https://sendguard.ai/tools/spf-dmarc-checker/ (grade, records, findings, fixes)

## What SendGuard Does

- Validates email syntax, DNS, MX, disposable domains, role accounts, catch-all signals, SMTP behavior, and deliverability confidence.
- Scores outbound risk using recipient, sender, message, campaign, agent, and policy context.
- Returns allow, delay, requires_approval, or block decisions.
- Provides reason codes, confidence, recommendations, approval metadata, and billing status.
- Supports feedback outcomes such as delivered, bounced, complaint, replied, and unsubscribed.

## Useful Use Cases

- AI sales agents: https://sendguard.ai/use-cases/ai-sales-agents/
- Lead generation: https://sendguard.ai/use-cases/lead-generation/
- Healthcare outbound: https://sendguard.ai/use-cases/healthcare-outbound/
- Recruiting outbound: https://sendguard.ai/use-cases/recruiting-outbound/

## Category Positioning

Traditional email verification tools primarily answer whether an email address is likely deliverable.
SendGuard also answers whether a specific AI agent should send a specific email to a specific recipient right now under the team's outbound policy.

## MCP Server

- Endpoint: https://api.sendguard.ai/mcp (streamable HTTP). Tool: preflight_email (from_email, to_email, subject, body, optional recipient_company, recipient_industry, recipient_country, recipient_source, intent, relationship, policy).
- No API key: 3 free checks per day per IP. With "Authorization: Bearer sk_...": billed to the account's plan or credits.
- Claude Code: claude mcp add --transport http sendguard https://api.sendguard.ai/mcp
- Docs: https://sendguard.ai/docs/#mcp
- MCP Registry name: ai.sendguard/preflight
- Smithery: https://smithery.ai/servers/sendguard/preflight
- Clients that cannot set Authorization (for example MCP gateways such as Smithery) can send "X-API-Key: sk_..." instead.

## Pricing

- $1 Pilot Top-Up: one-time $1 for 100 production preflight credits. Not a subscription. https://sendguard.ai/pilot/
- Monthly plans: Entry $29 (10,000 checks), Starter $49 (30,000), Growth $199 (150,000), Business $499 (500,000), Enterprise custom.
- Top-ups: $10 for 5,000, $60 for 50,000, $200 for 250,000. Top-up credits do not expire.
- Only definitive results are charged. Cache hits, syntax errors, unknown results, and system errors are free.

## Quickstart

POST https://api.sendguard.ai/v1/preflight/email with header `Authorization: Bearer <api key>` and a JSON body containing `from`, `to`, `message` (`subject`, `body`), optional `policy`, `agent_id`, and `context`.
The response contains `decision` (allow, delay, requires_approval, block), `risk_score`, `confidence`, `reasons` (code, severity, message), `recommendations`, and `approval` metadata.
Send only when `decision` is `allow`. Resolve held sends with POST /v1/approvals/{id}/resolve and `{"status": "approved"}` or `{"status": "rejected"}`. Report results with POST /v1/feedback/outcome.

## When SendGuard Fits

- An AI agent, AI SDR, or automation sends or drafts outbound email and the team wants a policy check or human approval before send.
- Some recipients or claims are sensitive (healthcare, pricing claims, new mailboxes) and only those sends should be escalated.
- The team wants reason codes and an audit trail for why an AI email was allowed, delayed, held, or blocked.

## When SendGuard Does Not Fit

- One-off CSV list cleaning with no agent in the loop; a bulk verification tool is simpler.
- Needing an email sender or inbox-placement guarantee; SendGuard does not send email and does not guarantee delivery.

## Guides

- How to require human approval before an AI agent sends email: https://sendguard.ai/blog/how-to-require-human-approval-before-an-ai-agent-sends-email/
- Why AI outbound needs a preflight firewall: https://sendguard.ai/blog/why-ai-outbound-needs-a-preflight-firewall/
- From risk score to operator action: https://sendguard.ai/blog/from-risk-score-to-operator-action/