Skip to content

Guide

Set Up Your Sending Infrastructure with AI

Set up your domains and mailboxes, connect them to your outreach app, and check warmup. Your AI assistant handles supported provisioning tasks and checks progress. You approve changes and complete any manual registrar or warmup steps.

Copy this prompt into your assistant. If it is not connected yet, start with step 1.

Help me set up my sending infrastructure with builders.ac. Check what I already
have, ask what you need from me, and guide me through domain setup, mailboxes,
sequencer connection, and warmup. Verify the environment, account, and workspace.
Confirm the scope and relevant costs with me before making changes, and verify
each step. Tell me which actions I need to complete manually.
  • builders.ac account — access to your intended workspace.
  • Owned domains — access to the registrar where you registered them.
  • Mailbox capacity and sender details — paid slots, final counts, and sender names.
  • Sequencer account, if applicable — a supported account and destination workspace.

Expand any step to see what your assistant handles and what you need to do.

Connect your assistant and confirm the right account and workspace.

What happens: Your assistant connects to builders.ac and verifies your account, connection environment, and available workspaces.

Your action: Follow the MCP connection guide, sign in, and confirm the workspace. The kit currently targets UAT; its settings are not interchangeable with the MCP guide’s environment. Do not switch environments merely because another connection works.

Done when: Your assistant has successfully checked access and confirmed the intended account, environment, and workspace. If access fails, use MCP troubleshooting.

Review what you own and approve a setup plan.

What happens: Your assistant checks existing domains, mailboxes, background work, and sequencer connections, then proposes what to reuse or add.

Your action: Confirm the domains, final mailbox count per domain, sender names, and relevant costs. Buy domains separately if needed; domain purchase and renewal are not implemented here. Resolve missing access or capacity before the affected step.

Done when: You have approved a concrete plan that reuses suitable resources and clearly identifies additions. Existing mailboxes are included in the final count.

Connect owned domains and track their setup.

What happens: Your assistant connects new domains you own and tracks background setup. Existing active domains are reused; incomplete ones are inspected before repair. Connecting each new domain reserves 50 mailbox slots, but does not create mailboxes.

Your action: Confirm the selected domains and slot impact. If setup needs your attention, follow the assistant’s next action rather than repeatedly reconnecting.

Done when: The selected domains are active, or setup has reached the nameserver checkpoint in step 4. Pending or failed setup must be resolved before mailbox creation.

Update your registrar if a nameserver change is needed.

What happens: Nameservers tell the internet where your domain’s DNS records live. Your assistant shows the exact values returned for your domain and checks progress.

Your action: If required, update those nameservers at your registrar. Review the impact on existing website or email services first. Registrar changes are manual unless you have a supported, authorized integration. If still waiting, check the values and allow time for the change to spread.

Done when: Your assistant confirms the domain is active; any failed setup is explained.

Create your approved mailboxes on active domains.

What happens: On active, ready domains, your assistant creates only approved missing mailboxes. Supported capacity-occupying totals are 50, 75, or 150 per domain; existing pending and failed mailboxes can count too. Your paid capacity must cover the plan.

Your action: Confirm the final total and sender addresses. For example, adding 25 to 50 reaches 75. The assistant must ask before increasing an unsupported smaller request.

Done when: Every expected mailbox is active, with the agreed addresses and counts. Failures require review, not automatic replacement.

Choose where your mailboxes will connect.

What happens: Your assistant finds available sequencer accounts and workspaces. A sequencer manages outreach sequences; it is separate from the service hosting mailboxes. Binding a domain can happen before or after mailbox creation and can queue existing mailboxes for connection.

Your action: Confirm a valid destination. Changing a binding can move all existing mailboxes on that domain and requires your approval. Resolve missing registration or upload entitlement first.

Done when: Each domain shows the confirmed destination. Individual mailbox connections are checked next.

Check each mailbox reaches the selected sequencer.

What happens: Your assistant first checks work already queued by the domain binding. It uploads only when needed and verifies each expected mailbox’s connection to the selected sequencer workspace. A finished upload can still have skipped or failed mailboxes.

Your action: Review any incomplete connections and approve a repair plan if needed. Successful connections should be preserved; do not create replacement mailboxes to fix an upload problem.

Done when: Every expected mailbox is connected to the correct destination, with no unexplained missing or failed connections.

Enable and verify warmup where your sequencer supports it.

What happens: Warmup gradually builds mailbox activity before outreach. Uploading mailboxes does not prove warmup is enabled; builders.ac MCP has no supported warmup enablement tool in the reviewed contract.

Your action: Where supported, enable warmup in your sequencer’s documented interface. Check each address in its warmup dashboard and confirm the result to your assistant. Resolve provider errors there; mark unsupported warmup as incomplete.

Done when: Warmup is enabled and verified for every intended mailbox, with the verification source recorded. Enabled does not mean fully warmed up.

Review completed setup and remaining actions before sending.

What happens: Your assistant summarizes what is provisioned, connected to the sequencer, warmup-enabled, and still blocked. These are separate milestones.

Your action: Review the checklist below and any remaining actions. Before future supported campaign use, separately review email health, mailbox login and sending/receiving, warmup progress, sending limits, recipients, content, and campaign settings.

Done when: Every setup outcome has evidence or a clear next action. Provisioned infrastructure is not automatically campaign-ready. Do not launch production campaigns from this internal-testing workflow.

Use this as a reference for your own review; it does not show live infrastructure status.

  • Correct environment, account, and workspace confirmed.
  • Selected domains and expected mailboxes active.
  • Each domain uses the confirmed sequencer destination.
  • Every expected mailbox connected to that destination.
  • Warmup enabled and verified where supported; gaps recorded.
  • No unresolved failures or partial results; pending work identified.
  • Remaining checks before campaign launch clearly listed.
Technical details and references

Your assistant should discover current tool names and inputs before acting. The canonical infrastructure skill and provisioning contract describe the workflow; the kit’s authenticated end-to-end acceptance remains pending.

  • Inventory and authentication: Verify identity with me and use workspace/domain filters across every list page. OAuth collection reads support non-secret readiness checks; do not request passwords or API secrets to check progress. Secret-bearing retrieval has separate authentication requirements.
  • Background work and retries: Poll getOperation until done, then inspect state, errors, and actual resources. Finished is not necessarily successful. After an uncertain response, reconcile before retrying; reuse identical arguments and the same idempotency key for a transport retry. Replaying failed work does not restart it. Do not reconnect active domains, delete resources, or cancel slow work as a retry. See MCP retry rules.
  • Capacity and creation: Count all capacity-occupying rows, not only active mailboxes. Domain and backing tenant limits are each 150; domains may share tenant capacity. There is no public slot-balance endpoint; returned capacity refusals are authoritative. Choose explicit addresses or persona batches, not both. Failed mailboxes cannot currently be retried or removed, and their addresses remain taken. See mailbox creation and known limitations.
  • Sequencer upload: Binding returns a domain and may already queue connection work. Separate uploads cover all created, not-yet-connected mailboxes on selected domains, which must share the builders.ac workspace and sequencer destination. Preserve sending and warmup settings by omitting strategy unless a change was requested. Check each mailbox’s destination, remote ID, and state: not_uploaded, uploading, uploaded, failed, or disconnected. Preserve successful results and reconcile partial failures before an authorized retry. See sequencer setup.
  • Evidence and resuming: Empty, unmeasured, or stale health checks do not prove health; active mailboxes do not prove login or send/receive success. Label manual warmup checks as customer-confirmed unless directly verified through a documented, authorized integration. Keep a private, secret-free summary of resource and operation IDs, problem codes, and last observed states and times. Recheck environment, account, workspace, and existing work when resuming.