TOKEN.PRICING
PAY FOR WHAT YOU USE

A token is ERP.AI's unit of compute. 1 token = $0.005 at pay-as-you-go rates. Every API operation consumes tokens based on its type (read, write, compute, proto) multiplied by the category factor. A simple GET query costs 1 token. Running a complex P&L report in Finance costs 150 tokens. Your $10,000 free credit gives you 2 million tokens to start.

Per-seat pricing was designed for humans using desktop software. Agents don't sit in chairs. A single agent can do the work of 100 users. Charging per-seat would make agent adoption prohibitively expensive. Token pricing means you pay for compute consumed, not identities provisioned. An idle user costs $0. An agent that processes 10,000 invoices pays exactly what that compute is worth.

Traditional ERP/CRM charges $50-300 per user per month regardless of usage. A 100-person company on Salesforce pays ~$15K/mo. On ERP.AI, that same company pays only for actual operations — typically $250-2,500/mo depending on volume. No per-seat costs, no shelf-ware, no minimum commitments. Idle users cost nothing.

Yes. ERP.AI is headless-first — every operation is an API call, whether from an autonomous agent, a Proto execution, or a human using a frontend. Token costs are identical regardless of caller. Agent code AGT-4821 pays the same per-token rate as human user [email protected]. No "agent surcharge" ever.

No. Only successful operations (2xx responses) consume tokens. 4xx client errors and 5xx server errors are free. Rate-limited requests (429) are also free. You can monitor token usage in real-time via GET /v1/usage or the billing dashboard.

Yes. Use the free tier to run your apps with real data and real users. After 2-4 weeks, check your billing dashboard for actual usage patterns. You can also call GET /v1/usage/forecast to get a 30-day projection based on your current run rate. We publish scenario estimates above showing typical costs for different company sizes.

No. Tokens cover compute, storage up to your tier limit, and standard support. The only additional costs are: (1) storage overages beyond your tier allocation at $0.10/GB/mo, (2) enterprise add-ons like SSO/SAML, HIPAA compliance, and dedicated support. There are no egress fees, no API gateway fees, no "platform fees."

Read (1 tk): Any GET request — list records, fetch a single item, search, filter. Write (4-5 tk): Any POST/PUT/PATCH/DELETE — create, update, archive, or remove a record. Compute (50-200 tk): Heavy operations like financial reports, BOM calculations, AI enrichment, bulk transforms. Proto (25-200 tk): Workflow executions — approval chains, scheduled jobs, multi-step automations. Real-time (1-3 tk): WebSocket events and pub/sub subscriptions.

Volume tiers are pre-purchased commitments, not usage tiers. You buy 500K tokens at $0.0045/tk and use them until they run out. If you start on pay-as-you-go and later move to a volume tier, the new rate applies to all future tokens but does not refund prior usage. You can upgrade tiers at any time mid-cycle.

Every registered company automatically gets $10,000 in agent tokens. No credit card required. Tokens never expire. Use them across any apps — ERP, CRM, HRMS, all 31 categories. When your free tokens run out, add a payment method and continue at pay-as-you-go rates, or pre-purchase a volume tier for discounts up to 60%.

A registered company means you sign up with an official company email address (e.g. [email protected]) and a verified phone number. Free email providers like Gmail, Yahoo, Hotmail, or Outlook.com do not qualify. This ensures one free tier per legitimate business and prevents abuse. If your company doesn't have a custom domain yet, contact [email protected] and we'll work with you.

No. Free tier tokens have no expiry date. You can sign up today, use nothing for 6 months, and come back with your full $10,000 balance intact. We don't do "use it or lose it." Your tokens are yours until you spend them.

Your agents are never hard-blocked without warning. When free tokens run low, we notify via webhook and email at 80%, 90%, and 100% thresholds. At 100%, read operations continue working for 7 days (grace period). Writes and protos are paused until you add a payment method. You can set up auto-top-up via POST /v1/billing/auto-topup to avoid any interruption.

Yes. Qualifying startups can apply for up to $100K (or even $250K for accelerator-backed companies) through our Startup Program. Non-profits and educational institutions can also apply for extended free credits via [email protected].

No. Every app across all 31 categories is available on the free tier. The only limits are: 10 req/s API rate limit, 1GB data storage, 500MB file storage, 7-day audit logs, and community support. There are no feature gates — you can run protos, deploy agents, use webhooks, everything.

Yes. Credits are company-level, not user-level. You can invite unlimited team members and deploy unlimited agents — they all draw from the same $10K pool. There's no per-user fee ever. Add your entire team on day one.

No. We detect duplicate accounts via domain, IP, and payment method fingerprinting. Abusing the free tier results in immediate account termination. If you legitimately have multiple businesses, each incorporated entity can register separately with its own domain email.

Pay-as-you-go is billed monthly in arrears. You use tokens throughout the month and receive an invoice on the 1st. Volume tiers are pre-paid: you purchase a token package upfront and draw down from it. All invoices are available in PDF via the billing dashboard or GET /v1/billing/invoices.

Visa, Mastercard, American Express, and wire transfer for enterprise accounts. All payments processed via Stripe. We support ACH for US-based companies on Growth tier or above. Cryptocurrency is not accepted. Invoice-based billing (NET 30/60) available for Enterprise tier only.

Yes. Set hard limits via POST /v1/billing/limits. When your monthly spend reaches the cap, all write/compute operations pause and read operations enter a 7-day grace period. You can also set soft limits that trigger webhook alerts at custom thresholds (e.g., 50%, 75%, 90% of budget). Budget alerts are per-app or company-wide.

Volume tiers are effectively prepaid plans. Buy 2M tokens at $0.004/tk ($8,000) and use them whenever. They don't expire. For Enterprise customers, we offer annual commitments with custom pricing, reserved capacity, and additional discounts beyond the 60% volume tier. Contact sales for details.

Yes. Every billing cycle generates a detailed PDF invoice showing: token consumption by app, by user/agent, by operation type, volume discounts, and net total. Invoices are available in the billing dashboard, via API, and emailed to the billing contact. We support custom PO numbers and tax IDs for enterprise customers.

Open a billing dispute via the dashboard or email [email protected]. We provide granular operation logs showing every token consumed, timestamped and tied to a specific API call. If we made a billing error, we refund immediately. Disputes are resolved within 5 business days.

Yes, where applicable. US customers pay state sales tax based on their billing address. EU customers pay VAT unless they provide a valid VAT ID (reverse charge applies). All other jurisdictions follow local tax requirements. Tax is itemized separately on invoices.

No. Credits are tied to the company account that purchased or received them. If you're restructuring or merging companies, contact support to discuss account migration options. We handle these case-by-case.

Three ways: (1) The billing dashboard shows real-time usage by app, user, agent, and operation type with daily charts. (2) GET /v1/usage returns current-period totals programmatically. (3) Every API response includes an X-Token-Cost header showing exactly how many tokens that call consumed. You always know exactly where tokens are going.

Yes. The billing dashboard has a per-user/per-agent breakdown sorted by consumption. You can filter by app, date range, or operation type. If AGT-7291 is burning 40% of your tokens on invoice processing, you'll see it clearly. Use this to optimize agent behavior or set per-agent token budgets.

Yes. Use POST /v1/billing/budgets to set monthly token limits per agent ID, per app, or per user. When a budget is hit, that entity's operations pause while the rest of your company continues normally. Useful for preventing runaway agents or limiting departmental spend.

Yes. Configure alerts at any threshold (e.g., 50%, 75%, 90%, 100% of budget) via the dashboard or POST /v1/billing/alerts. Alerts can fire via email, webhook, or Slack. You can also subscribe to billing.threshold events on your WebSocket connection for real-time monitoring.

Yes. GET /v1/usage/export?format=csv returns a CSV with every operation, timestamp, user/agent, app, and token cost. You can filter by date range. Enterprise customers can push usage data directly to their data warehouse via our Snowflake or BigQuery connectors.

A Proto is an autonomous, agentic workflow execution. Think of it as a self-driving program that chains together multiple operations, makes decisions at each step, and adapts based on results — without human intervention. A Proto can observe the state of your data, reason about what to do next, take action, and loop until the job is done. Examples: an invoice collection proto that checks overdue invoices, decides which to escalate vs. send a reminder, drafts personalized emails, and schedules follow-ups — all autonomously.

Traditional automation (Zapier, n8n) is "if X then Y" — rigid, linear, brittle. Protos are agentic: they observe, reason, act, and iterate. A Proto can handle exceptions it's never seen before, make judgment calls (should I approve this PO or flag it for review?), chain together arbitrary operations based on runtime context, and self-correct when something fails. They're closer to deploying an autonomous agent with a specific mission than wiring up a flowchart.

Yes. That's the default mode. A Proto can be triggered by a cron schedule or webhook, execute its entire mission (read data, make decisions, write results, call external APIs, send notifications), and complete without any human ever seeing it. You can optionally add human-in-the-loop checkpoints for high-stakes decisions (e.g., "pause and ask finance before approving POs over $50K"), but autonomous end-to-end execution is first-class.

Proto token cost = sum of all sub-operations + orchestration overhead. A simple 3-step approval flow costs ~25 tokens. A complex MRP run with 50+ operations costs ~200 tokens. An agentic collection proto that processes 500 overdue invoices autonomously might cost ~2,500 tokens per run. The X-Proto-Cost response header breaks down exactly how many tokens each step consumed. Proto costs are also visible in the execution log on the billing dashboard.

Yes. Protos support three trigger types: CRON (scheduled, e.g., "run payroll on the 25th"), WEBHOOK (event-driven, e.g., "when a new lead comes in, run qualification autonomously"), and MANUAL (triggered by a human or agent via API). Scheduled protos run in the background without supervision — you check results when you want. They show up in your billing dashboard with their next-run time and historical execution costs.

Protos are agentic, so they attempt self-recovery first — retrying failed API calls, trying alternative approaches, or gracefully degrading. If truly stuck, the Proto pauses and can either: (1) notify you via webhook/email with full context, (2) retry automatically after a backoff, or (3) fail and log the error. You're only charged for steps that completed successfully. Retry from the failure point via POST /v1/protos/{id}/retry without re-paying for completed steps.

Both. ERP.AI ships with 200+ pre-built protos (invoice approvals, PO generation, employee onboarding, etc.). You can also define custom protos as JSON or YAML workflow definitions via POST /v1/protos. Custom protos support conditionals, loops, parallel branches, error handling, human-in-the-loop gates, and — critically — agentic decision nodes where the proto reasons about what to do next based on the data it's seeing.

Protos can call any HTTP endpoint. Use http_call steps to integrate with Stripe for payments, SendGrid for emails, Twilio for SMS, your own microservices, or any third-party API. The proto orchestrator handles auth headers, retries, and response parsing. External calls consume 2 tokens each for orchestration overhead. This makes Protos a full autonomous integration layer, not just internal workflow automation.

Yes, and that's where they shine. A single Proto can: read a sales opportunity from CRM, check inventory levels in ERP, generate a quote in Invoicing, create a purchase order in Supply Chain, and notify the sales rep — all autonomously in one execution. Cross-app protos are the backbone of autonomous business operations. Each sub-operation is billed at the standard token rate.

Subscribe to proto.step events via WebSocket to see each step as it executes. The billing dashboard shows a live execution log with: step name, operation type, token cost, duration, and result. For long-running protos (e.g., batch processing 10,000 invoices), you get progress percentage and ETA. You can also abort a running proto via POST /v1/protos/{id}/abort if it's going off the rails.

Register an agent via POST /v1/agents with a name and description. You get back an agent ID (e.g., AGT-7291) and an API key. Point your agent code at https://api.erpai.com with that key. There's no container to deploy, no server to provision. Your agent is just code that makes API calls.

No limit on agent count. Free tier: 10 req/s shared across all agents. Pay-as-you-go: 100 req/s. Enterprise: 1,000 req/s or custom. You can register 1 agent or 100 — there's no per-agent fee. Rate limits are per-company, not per-agent, so distribute carefully.

Yes. Everything is API-first. The same invoice that a human creates through the web UI is accessible via the same API that an agent uses. Agents and humans share the same database, same permissions system, same audit trail. Real-time WebSocket events let humans see agent changes instantly.

Agents have the same role-based access control (RBAC) as human users. Assign an agent to a role (e.g., "Invoice Processor" with write access to invoices but read-only on contacts). You can also set per-agent token budgets, IP whitelists, and operation-type restrictions. Every agent action is logged in the audit trail.

Yes. Revoke an agent's API key via DELETE /v1/agents/{id}/keys for immediate termination. Or set the agent's budget to $0 to pause it. You can also use the dashboard kill-switch which revokes all agent keys company-wide in one click — useful for emergencies.

Yes. ERP.AI is API-first — any HTTP client works. We publish Python and TypeScript SDKs with built-in support for LangChain tools, CrewAI tasks, and OpenAI function-calling. The skill.md spec at https://erpai.studio/public/sdstack/SKILL.md provides agent-readable documentation that LLMs can parse directly to discover available operations.

Community support via our public Discord and GitHub Discussions. We have active community members and ERP.AI engineers answering questions. Response time varies but most questions get answered within a few hours. You also have full access to documentation, API reference, and the skill.md spec.

Pay-as-you-go: Email + chat support, 8-hour response SLA during business hours (Mon-Fri, 9am-6pm PT). Growth: Same plus priority queue, 4-hour response SLA, and Slack Connect channel. Enterprise: Dedicated Technical Account Manager, 1-hour response SLA, 24/7 on-call for P1 incidents, quarterly business reviews, and custom onboarding.

Yes. status.erpai.com shows real-time availability for all services. Subscribe to updates via email, RSS, or webhook. We publish post-mortems for any incident affecting >1% of customers. Historical uptime is >99.95% over the last 12 months.

Yes. We provide migration guides for moving from Salesforce, SAP, NetSuite, QuickBooks, and Workday. Bulk import APIs support CSV, JSON, and XML. For Growth/Enterprise customers, our solutions engineers will assist with data mapping and migration scripts. Average migration from Salesforce CRM takes 2-3 days.

Yes. Every app has full API documentation at /docs/{app-name}. Every endpoint has request/response schemas, example payloads, error codes, and token costs. The skill.md file at https://erpai.studio/public/sdstack/SKILL.md provides machine-readable documentation that agents can parse to self-discover available operations.

Free tier: Self-serve onboarding with guided tutorials in the dashboard. Growth: 1-hour onboarding call with a solutions engineer. Enterprise: Custom onboarding program including data migration, workflow design, agent deployment assistance, and team training sessions. We also publish video tutorials, webinars, and a certification program.

Free tier: 10 req/s. Pay-as-you-go: 100 req/s. Growth: 100 req/s with burst to 200. Enterprise: 1,000 req/s or custom. Rate limits are per-company, shared across all users and agents. Rate-limited requests return 429 Too Many Requests with a Retry-After header. Rate-limited requests don't consume tokens.

Yes. Official SDKs in Python, TypeScript/JavaScript, Go, and Java. Community SDKs in Rust, Ruby, and PHP. All SDKs handle authentication, retry logic, rate limiting, and token tracking automatically. Install via pip, npm, go modules, or Maven. Source code on GitHub.

Yes. ERP.AI is API-first and protocol-agnostic. REST API, GraphQL (beta), WebSocket for real-time, and webhooks for event-driven architectures. Integrate with any frontend (React, Vue, mobile), any backend (Node, Python, Go), any CI/CD pipeline, any data warehouse, any BI tool. If it can make HTTP calls, it works with ERP.AI.

Free tier: no SLA (but historically >99.9%). Pay-as-you-go: 99.9% monthly uptime SLA. Enterprise: 99.99% with financial penalties for missed SLA (service credits). Enterprise customers can also get dedicated infrastructure for guaranteed isolation and performance.

Not currently. ERP.AI is cloud-only. Enterprise customers can get dedicated single-tenant infrastructure in their preferred region for data residency and compliance requirements. We're exploring a hybrid deployment model for highly regulated industries — contact sales if interested.

Register webhook endpoints via POST /v1/webhooks with event filters (e.g., invoice.created, contact.updated). Webhooks are signed with HMAC-SHA256 for verification. Failed deliveries are retried with exponential backoff (5 attempts over 24 hours). Receiving a webhook costs 1 token. You can test webhooks via POST /v1/webhooks/{id}/test.

Yes. Every account gets a sandbox environment at sandbox.api.erpai.com. Sandbox operations don't consume tokens. Data is isolated from production. Sandbox supports all API operations and webhooks. Use it for development, testing, and CI/CD pipelines. Sandbox data resets weekly unless you opt into persistent sandbox (Enterprise only).

Click "Sign In" and register with your company email. Or use the API: curl -X POST https://api.erpai.com/v1/agents/register -d '{"company":"your-co"}'. You'll get an API key and $10,000 in free tokens instantly. No credit card, no demo call, no sales approval.

All of them. Every app across all 31 categories is available to every account, including free tier. There's no app activation fee or per-app license. Just start making API calls to any app endpoint. If you want CRM today and add ERP tomorrow, just call the ERP API — it works immediately.

Yes. Unlimited users on every tier, including free. Invite via email through the dashboard or POST /v1/team/invite. Each user gets their own login but shares the company's token pool. No per-seat fee ever. Your 200th user costs the same as your 1st: $0.

No. Free tier is forever-free. Pay-as-you-go is month-to-month with no minimum. Volume tiers are pre-paid but non-expiring — use your tokens whenever. Only Enterprise tier involves an annual agreement, and even that is negotiable. You can cancel anytime with zero penalties.

Yes, at any time. Move from free to pay-as-you-go by adding a payment method. Purchase a volume tier to get discounts. Upgrade or downgrade volume tiers mid-cycle. Unused tokens from a higher tier carry over. Contact sales to explore Enterprise pricing. All transitions are instant with no downtime.

30-second path: (1) curl -X POST https://api.erpai.com/v1/agents/register -d '{"company":"your-co"}' (2) Copy your API key from the response (3) curl -H "Authorization: Bearer YOUR_KEY" https://api.erpai.com/v1/apps to see all available apps. Read https://erpai.studio/public/sdstack/SKILL.md for the full endpoint catalog. Deploy your first agent in under 5 minutes.

Yes. Bulk import API supports CSV, JSON, and XML. We publish field-mapping guides for Salesforce CRM, SAP ERP, NetSuite, QuickBooks, Workday HRMS, and more. Growth/Enterprise customers get migration assistance from our solutions team. Data import operations consume tokens at write rates (4-5 tk per record).

Yes. Full data export via GET /v1/export in JSON, CSV, or Parquet format. Includes all records, files, audit logs, and metadata. Export operations consume read tokens. Enterprise customers can set up continuous replication to their own data warehouse. Your data is yours — no lock-in, no export fees beyond token costs.

Automatic daily backups with 30-day retention (all tiers). Point-in-time recovery within the last 7 days (Growth/Enterprise). Enterprise customers get cross-region backup replication and custom retention policies. Backups are encrypted with the same keys as production data. Restore requests via support or API.

Yes. If you're building a SaaS product on top of ERP.AI, you can create isolated tenant environments via POST /v1/tenants. Each tenant gets its own data namespace, RBAC, and token budget. You pay tokens for all tenants from your account. Per-tenant billing reports available for chargeback. This is how ERP.AI itself is built.

Yes. All data is encrypted at rest (AES-256) and in transit (TLS 1.3). API keys are hashed with bcrypt and never stored in plaintext. Database backups are encrypted with customer-specific keys for Enterprise accounts. We use AWS KMS for key management.

SOC 2 Type II certified (annual audit). HIPAA and PCI-DSS compliance available on Enterprise tier with BAA agreements. Healthcare and Banking app categories automatically enable additional encryption, audit logging, and access controls required by their respective regulations.

Default: US-East (Virginia) and EU-West (Frankfurt) with automatic geo-routing. Enterprise customers can choose dedicated data residency in US, EU, APAC (Singapore), or India (Mumbai). Data never leaves your chosen region.

Enterprise tier includes SAML 2.0 SSO with Okta, Azure AD, Google Workspace, and any SAML-compliant IdP. SCIM provisioning for automatic user sync. Free and pay-as-you-go tiers support email/password and Krawler SSO.

Yes, on Enterprise tier. BYOK (Bring Your Own Key) via AWS KMS or Google Cloud KMS. You retain full control of the encryption keys and can revoke access at any time.

You get a 30-day export window after cancellation. During this period, your data is read-only and you can export everything. After 30 days, data is permanently deleted from all systems including backups. We provide a deletion certificate upon request.

Dedicated TAM, 99.99% SLA with financial penalties, SSO/SAML, SCIM provisioning, SOC 2/HIPAA/PCI-DSS compliance, unlimited audit log retention, data residency, BYOK encryption, 1,000 req/s rate limit, custom onboarding, quarterly business reviews, and 24/7 P1 incident support. Volume pricing starts at $0.002/token (60% discount).

Yes, for Enterprise tier. Contact [email protected] with your expected volume, compliance requirements, and deployment preferences. We offer custom token rates, reserved capacity, annual billing, and volume-based discounts beyond the published tiers. Typical Enterprise contracts start at $50K/year.

Yes, on Enterprise tier. Dedicated infrastructure runs on isolated AWS accounts in your preferred region. Your data, compute, and network are fully separated from other customers. This satisfies regulatory requirements for financial services, healthcare, and government agencies. Additional infrastructure costs apply.

Yes. We sign Business Associate Agreements (BAAs) for Enterprise customers using Healthcare category apps. HIPAA-compliant infrastructure includes: dedicated encryption keys, enhanced audit logging, access controls, and automatic PHI detection. BAA execution takes 5-10 business days after Enterprise agreement is signed.