Deal Flow

Pipeline Thesis

Colony creates commercial opportunity by collapsing the entire GTM stack — prospect discovery, qualification, outbound sequencing, call intelligence, and onboarding — into a single agentic control plane. Every workflow that a ZoomProp sales team previously executed across disconnected tools (CRM, sequencer, notetaker, content calendar) now runs inside Colony as an orchestrated agent loop. The commercial proposition is time-to-revenue compression: a founder can issue one plain-English instruction at /overview and the GTM Orchestrator streams a complete pipeline action — from ICP-scored prospect list through approved outreach — in minutes rather than days. The Deployment Kit auto-generated on Closed-Won (BR-007) creates a replicable proof asset that feeds the next sales cycle. Colony therefore compounds value: each closed deal produces the case-study raw material that improves conversion on the next.

---

Target Accounts

---

Qualification Signals

• Pipedrive is already in use — the prospect's CRM is Pipedrive; Colony's bi-directional sync (docs/phase1/action_plans/06_pipedrive_integration.md, runbook docs/phase1/runbooks/pipedrive-fields.md) delivers immediate value without a migration ask. Confirmed by dry-run artifacts in runners/.artifacts/pipedrive-dry-run/.
• Founder or AE is hand-writing outreach — no sequence tool or a generic one (Apollo, Outreach) with no LLM personalisation; Colony's 5-angle message-gen agent (docs/phase1/action_plans/13_message_generator_agent.md) produces signal-, pain-, referral-, pattern-break-, and insight-led variants with per-org circuit breakers (docs/phase1/action_plans/22_circuit_breakers.md).
• Call recordings are unprocessed — meeting notes sit in Google Drive with no downstream CRM update; Colony's recording intelligence agent ingests Gemini Meet Notes and routes extracted signals to CRM fields + Knowledge Core (docs/phase1/action_plans/14_recording_intelligence_agent.md, docs/phase2/runbooks/03_recording_pipeline.md).
• Content is ungated from pipeline influence — marketing publishes across channels with no attribution loop; Colony's 6-pillar content pipeline with attribution-to-pipeline-influence tracking (docs/phase1/action_plans/16_content_agent.md) directly ties content output to deal stage movement.
• ICP scoring is manual or absent — qualification decisions made by intuition; Colony's qualification agent with ICP-score field (docs/phase1/action_plans/12_qualification_agent.md) and anti-ICP hard pre-filter (docs/phase2/action_plans/16_anti_icp_hard_prefilter.md) enforce repeatable disqualification before any sequence spend.
• Team size is small but deal complexity is high — ≤5-person GTM team managing enterprise-style onboarding; Colony's onboarding agent auto-generates the 8-asset Deployment Kit on Closed-Won (docs/phase1/action_plans/26_deployment_kit.md, docs/phase1/action_plans/17_onboarding_agent.md): pitch deck, one-pager, case-study shortlist, pricing reference, rollout calendar, stakeholder map, risk register, KPI dashboard.
• Email deliverability is degraded — high bounce or spam rates; Colony's email deliverability and warm-up module (docs/phase2/action_plans/19_email_deliverability_and_warmup.md) plus sender mailbox rotation (docs/phase2/action_plans/25_sender_mailbox_rotation.md) are concrete remediation assets.
• Multi-channel outreach is unsynchronised — LinkedIn and email sequences run independently; Colony's Unipile integration (docs/phase1/action_plans/07_unipile_integration.md, docs/phase2/action_plans/06_unipile_search_extension.md) and multi-channel sequence engine (docs/phase2/action_plans/23_multi_channel_sequence_engine.md) unify both under one approval queue.

---

Conversion Loop

1. Proof-of-value in the discovery call. Run the Pipedrive LeadBooster scraper (docs/phase2/action_plans/02_pipedrive_leadbooster_scraper.md) as a live demo: the prospect watches the discovery orchestrator (docs/phase2/action_plans/10_discovery_orchestrator.md) pull candidates from Google Places (docs/phase2/action_plans/03_google_places_integration.md), SerpAPI (docs/phase2/action_plans/04_serpapi_integration.md), and public-data adapters (docs/phase2/action_plans/05_public_data_adapters.md), score them through the matching algorithm (docs/phase2/action_plans/08_matching_algorithm.md), and surface an ICP-ranked shortlist — all inside the Colony chat at /overview. The runners/.artifacts/pipedrive-dry-run/ JSON outputs (candidates, organisations, people) serve as tangible leave-behinds.

2. Free-trial activation with immediate daily brief. On org creation (Clerk webhook → Svix sync → Cloud SQL Postgres colony-39989:…:colony), the analytics agent (docs/phase1/action_plans/18_analytics_agent.md) schedules a daily brief (pipeline today, alerts, HOT reply queue, yesterday's output) delivered via Resend (docs/phase1/action_plans/08_resend_integration.md) and visible at /overview from Day 1. The brief creates a daily pull that embeds Colony in the prospect's morning routine before any commercial conversation closes.

3. Approval queue as trust-builder. The candidate approval queue (docs/phase2/action_plans/09_candidate_approval_queue.md) and outbound approval queue (docs/phase1/action_plans/21_approval_queue.md) ensure no message or prospect exits Colony without human sign-off during trial. Per-org circuit breakers (docs/phase1/action_plans/22_circuit_breakers.md, docs/phase2/action_plans/11_circuit_breakers_phase2.md) enforce rate limits and protect sender reputation. This removes the "AI will spam everyone" objection before it surfaces.

4. First Closed-Won triggers the case-study flywheel. The onboarding agent fires on stage = Closed-Won in the 9-stage pipeline tracker, generating the 8-asset Deployment Kit stored in gs://colony-assets (CMEK + signed URLs, infrastructure/gcp/storage.tf). The case-study shortlist asset within the kit becomes the input for the content agent's next production cycle (docs/phase1/action_plans/16_content_agent.md), which publishes across the 6-pillar content system with pipeline attribution — attributing net-new pipeline influence back to that closed deal. Each won deal therefore seeds outreach credibility for the next prospect segment.

5. Expansion through multi-org and vertical depth. Accounts that close on a single-org seat are upsold through the org-switcher (docs/phase2/action_plans/27_org_switcher_hardening.md), enabling multi-brand or multi-division deployments under a single Colony instance. Healthcare-vertical accounts unlock the healthcare ICP signal-extraction module (docs/phase2/action_plans/22_healthcare_icp_signal_extraction.md) as a paid add-on tier. CSV upload (docs/phase2/action_plans/26_csv_upload_backboard_source.md) and contact-list builder (docs/phase2/action_plans/24_contact_list_builder.md) serve as self-serve expansion hooks that increase seat stickiness by embedding proprietary prospect data inside Colony's Knowledge Core (pgvector 1536-dimension retrieval, infrastructure/postgres/init/02-extensions.sql).

6. Observability as renewal evidence. Langfuse traces every LLM call; Sentry captures runtime errors (both wired via GCP Secret Manager in infrastructure/gcp/secrets.tf). At renewal, the analytics agent produces a pipeline-contribution report — sequences sent, replies received, deals influenced, content pieces published — that justifies contract expansion without requiring a manual QBR build.