Building sequences

Sequences are available to customers on the Pro plan.

You don’t assemble sequences in a node editor. Describe the cadence in chat, and Lightfield’s agent programs it for you.

Building with the agent

A typical build looks like this:

1. Describe the audience, the offer, why you’re reaching out now, and the ask.
2. Choose senders. Email goes out from sequence senders; LinkedIn from connected LinkedIn accounts. Set both up in Settings → Sequences before you activate.
3. Draft. The agent saves a sequence with a name, messaging, steps, and a send window. Nothing is live yet.
4. Test run against one contact. Each step is composed as a draft so you can read the copy. Waits are skipped and nothing is sent.
5. Publish, then Activate. The first publish leaves the sequence paused. Activate is what starts sending for enrolled contacts.
6. Enroll the contacts you want in the cadence.

Before a test run or publish, the agent may ask you to grant permissions for the tools each step uses.

The same agent can edit an existing sequence, inspect an enrollment, or debug a send that didn’t go out.

Messaging

Every sequence has four fields that frame every message:

Field	What it is
Persona	Who you’re writing to
Value proposition	The outcome or benefit
Trigger	Why this outreach is timely
Call to action	The one next step you want

The agent writes each touch from these fields plus a per-step prompt and the contact’s CRM context. There is no merge-tag template. Personalization is agent-written from the record.

Steps

A sequence is an ordered list of touches, with waits between them.

Touch	What it does
Email	Composes and sends (or parks for review) from the designated senders
LinkedIn invite	Sends a connection request
LinkedIn DM	Messages a connection

Waits are delays, not steps: Wait 3 days, or Wait for connection (up to 5 days) after an invite. A sequence may start with a wait. It must end on a touch. At most 10 email steps.

LinkedIn order

A DM cannot be delivered until the recipient is a connection, so the shape is:

Invite → wait for connection → (optional wait) → DM

If the wait-for-connection deadline passes and they still haven’t accepted, a later DM is skipped. Email steps still run. A LinkedIn-only sequence with nothing left to send is cancelled as Not accepted.

An invite to someone already connected succeeds without sending another request. Keep the invite in the sequence; don’t author a DM-only cadence.

Connection invites default to no note. Notes draw on a much smaller LinkedIn budget than plain invites.

Review manually vs send automatically

Each email, invite, and DM step has a delivery mode:

• Review manually (default): the step composes the message and parks it in For review. Approving records the decision; the send still waits for the send window and pacing.
• Send automatically: the step sends without a human in the loop.

A test run never sends and never parks, regardless of delivery mode.

Senders

Sequence sending is set up in Settings → Sequences, separately from inbox sync.

• Email. Connecting a mailbox in Settings → Mail and Calendar does not make it a sequence sender. Turn on Enable for sequences in Settings → Sequences so the account can send. A sequence only sends from the connected accounts you designate on it; there is no fallback to the rest of the workspace.
• Use connected account. When this is on, email leaves from the designated connected mailbox. When it is off, email leaves from that account’s SMTP / warmed mailboxes.
• LinkedIn. Connect LinkedIn accounts in Settings → Sequences and designate them on the sequence. Each enrollment sticks to the LinkedIn account it first sends from.

You can draft a sequence before senders exist. You cannot activate an email sequence without a designated email sender, or a LinkedIn sequence without a designated LinkedIn account.

Send window and pacing

The default send window is weekdays, 9–5, in your timezone. Sends that land outside the window wait until it opens. Overnight windows (for example 10pm–6am) are not supported.

Workspace sending limits live in Settings → Sequences → Sending limits. They apply across all sequences. A sequence-level cap is optional; only add one if you want this campaign slower than the workspace defaults. High priority only reorders the send queue. It does not raise limits or bypass the window.

Enrolling contacts

Enrollment is how a contact starts a sequence. Enroll from chat or from the contact in the CRM. The sequence must be published. You can enroll into a paused sequence; sending waits until you activate or resume.

A contact is skipped when they:

• are already active or paused in this sequence,
• have no email, if the sequence has email steps,
• have no LinkedIn URL, if the sequence has LinkedIn steps.

One contact can have only one live enrollment in a given sequence. After that enrollment finishes or is cancelled, you can enroll them again.

What stops an enrollment

Outcome	What happened
Replied	The contact replied by email or LinkedIn. This removes them from every live sequence, including out-of-office auto-replies.
Completed	They finished the last step without replying.
Undeliverable	The address bounced, failed a deliverability check, or had bounced before.
Not accepted	A LinkedIn invite was not accepted in time, and no later email step remained.
Unenrolled	Someone unenrolled them. Remaining steps are skipped and this cannot be undone.

Pause contact holds them temporarily. Unenroll is permanent. Pausing the sequence itself only holds contacts who are waiting on a step; work already in progress finishes.

Editing a live sequence

After you publish a change, enrolled contacts pick it up at their next step. You do not need to pause first.

Next steps

• Sequences: what a sequence is and how it differs from an automation.
• How For review works: the queue where Review manually steps land.