Every capability

Eighty-odd things it does, and one bill

Grouped by where each one sits in the motion. The ones with a page have a page; the rest are here because a capability list that only shows what has marketing behind it is not a capability list.

Talk to sales Platform overview

ICP builder Prospect corpus LinkedIn search Post engagement mining Hiring triggers Keyword monitoring Reaction watch Intent scoring Email reveal Phone reveal Company enrichmentWebsite readingAudiences CSV import CSV exportSaved viewsContact grid Disqualify with reason

Intent signals

Keywords, reactions and replies, scored into one number the queue sorts on

Provenance per field Retention windows AI copywriting Voice matching Message templates Sequence builder Branch conditions Parallel tracts Campaigns Entry rules Campaign goals Supervised review Sending domains Sender rotation DNS verification Warmup traffic Deliverability scoring Bounce handling Complaint suppression Age ramp Daily caps LinkedIn invitations LinkedIn messages Post likes Cross-channel chains

Campaigns

Audiences × one template × senders, plus the rule that starts it

Thread folders Reply classification Auto-drafted answers Opt-out keywords Lost sweep Meeting booking Calendar sync CRM sync Slack alerts Zapier Signed webhooks Delivery retries Agent roster Editable prompts Slash commands Sub-agents Agent screen Workspace graph Run history Work items Funnel reporting Gate telemetry Credit ledger Budget alertsPrepaid top-upsClient workspacesSeat rolesAudit logMCP server Public APINotificationsPre-send gate Unsubscribe injection Postal address

Stage 1

Find

Who to talk to, and when they got warm

Lead sourcingA licensed corpus and your own LinkedIn, working two different questionsCorpus size: 700M+ contacts, licensedCorpus pull: Daily, 200 candidates per workspace Intent signalsKeywords, reactions and replies, scored into one number the queue sorts onScale: 1 to 5, on every contactQueue entry: Heat 4 and above EnrichmentEvery lead arrives with an address, or it does not arriveMetered as: enrich.reveal, 30 data creditsLatency: Same request. No pending state ICP & the fit gateMembership is binary and decided before a contact existsModel: Opus-class, not the economy defaultMetered as: ai.classify.fit, 3 action credits AudiencesLists that a campaign can point at, named for the ICP that built themBulk bound: 5,000 rows. Import, export and selectionImport batching: 250 rows per call

Stage 2

Write

What to say, in a voice that is yours

AI copywritingWritten from the evidence that made the lead warm, in your voiceReview cadence: Quality hourly, copy every 6 hoursFindings: Throttled to one a day each TemplatesThe step graph: channel, wait, branch, parallel tractObjects: template_steps, template_blocksPer campaign: Exactly one template CampaignsAudiences × one template × senders, plus the rule that starts itReview modes: Autopilot (default) or supervisedEntry: Per campaign, editable

Stage 3

Send

Out through your own accounts, at a safe pace

Email & LinkedInTwo channels, two vendors, one each, and honest about the limitsLinkedIn ramp: 5/day week one, 10 to day 30, 20 to day 60, then your own limitDomain ramp: 10 → 25 → 50 → your ceiling Sending domains & sendersA subdomain you own, verified live, with pacing per addressDomain: A subdomain, never the rootReceiving: Enabled at attach, so replies land Warmup & deliverabilityAn enforced ramp, real warmup traffic, and an honest account of its limitsLinkedIn bands: 5 / 10 / 20 / your limit at <7, <30, <60, 60+ daysDomain bands: 10 / 25 / 50 / your ceiling SequencesEmail first, then LinkedIn the same day. One entry, one chainLauncher tick: Every minuteBridge delay: Hours, same day

Stage 4

Close

Every reply, in one place, worked to a call

The unified inboxEmail and LinkedIn in one thread list, under one set of rulesPoll cadence: Every minuteEmail inbound: Webhook, on arrival Reply handlingClassified, drafted and queued. One unsent answer per threadDraft latency: On arrivalSend latency: Next eligible minute tick Booked meetingsThe calendar an agent books against, and the moment it tells everyoneBooking providers: Google Calendar, Calendly, Cal.comOn booked: CRM contact + note, alerts, funnel CRM & alertsSix CRMs, Slack, Zapier and signed webhooks. Verified before they read connectedCRMs: HubSpot, Pipedrive, Attio, Salesforce, Zoho, Go High LevelBooking: Google Calendar, Calendly, Cal.com

Stage 5

Run it

The team that does all of the above

The agent rosterFive roles, one per job. Find, qualify, write, send, checkRoster size: Five, fixedPlus: GTM CEO coordinates the five specialist roles The agent screenType a sentence; it becomes tool calls against an allowlistBelt source: MCP tools + capabilities + navigationScope check: Twice. On listing and on call The pre-send gateThe last thing before a send, and the database agreesEnforcement: Database constraint, not application logicInputs: Jurisdiction, entity type, lawful basis InsightsThe funnel, what went out, what came back, and who found themSurface: /insights, plus /settings/ops for gate telemetryRun ledger: routine_runs, one open run per job MCP & APIDrive your workspace from Claude, with your own scopesTransport: Stateless streamable HTTP, no SSEDiscovery: RFC 9728 + RFC 8414

The long tail

Everything, alphabetised, for the person doing an evaluation

If you are filling in a requirements matrix, this is the list to work from.

Age ramp Agent roster Agent screen AI copywriting AudiencesAudit logAuto-drafted answers Bounce handling Branch conditions Budget alerts Calendar sync Campaign goals CampaignsClient workspacesCompany enrichment Complaint suppression Contact grid Credit ledger CRM sync Cross-channel chains CSV export CSV import Daily caps Deliverability scoring Delivery retries Disqualify with reason DNS verification Editable prompts Email reveal Entry rules Funnel reporting Gate telemetry Hiring triggers ICP builder Intent scoring Keyword monitoring LinkedIn invitations LinkedIn messages LinkedIn search Lost sweep MCP server Meeting booking Message templatesNotificationsOpt-out keywords Parallel tracts Phone reveal Post engagement mining Post likes Postal address Pre-send gatePrepaid top-upsProspect corpus Provenance per field Public API Reaction watch Reply classification Retention purge Retention windows Run historySaved viewsSeat rolesSender rotation Sending domains Sequence builder Signed webhooks Slack alerts Slash commands Sub-agents Supervised review Suppression lists Thread folders Unified inbox Unsubscribe injection Voice matching Warmup trafficWebsite readingWhite labelWork items Workspace graph Zapier

Everything above, running on your list

Describe the company, confirm the keywords, connect a channel. The agents take it from there, and Supervised mode is available when you want to review before sending.

Talk to sales See pricing

Start with sales. $2,000 or $5,000 onboarding, then at least $250 monthly credit funding.