Wavelink Operations Dashboard

1. Poll Tickets

Queries HubSpot for unprocessed tickets in the Wavelink Sales pipeline created today.

Step Functions Triggered every 5 min by EventBridge → Step Functions

Logic Flow

1. Validate config (API key, pipelines)
2. Search tickets: pipeline = Wavelink Sales, created today, ai_status not set
3. Return list of ticket IDs (max 50 per batch)

Data Transformations

• Pipeline config string → list of IDs
• Ticket objects → array of ID strings

APIs

POST /crm/v3/objects/tickets/search

Metrics

None

2. Enrich Ticket

Enriches ticket with contact and company data. Resolves "end user" companies to their reseller.

Step Functions Runs in parallel for each ticket

Logic Flow

1. Fetch ticket details, contacts, and companies
2. Check if primary company type = "end user" or "end customer"
3. If yes: look up contact's other associated companies (v4 API)
4. Find first non-end-user company (reseller)
5. Update ticket association to reseller if found
6. Return enriched data bundle

Data Transformations

• Company type: case-insensitive match ("End Customer", "end user", etc.)
• Contact → all companies (v4 API, not just primary)
• End user company → resolved reseller company

APIs

GET /crm/v3/objects/tickets/{id}
GET /crm/v4/.../associations/contacts|companies
PUT /crm/v4/.../associations/... (if resolved)

Metrics

None

3. Extract End Customer

Uses AWS Bedrock (Claude 3.5 Sonnet) to identify end customer mentioned in ticket subject/content, then searches HubSpot for that company.

Step Functions After enrich

Logic Flow

1. Skip if: no company, company is end customer, or no ticket content
2. Send subject + content to Bedrock AI with extraction prompt
3. AI returns company name or "NONE"
4. Search HubSpot companies by name (CONTAINS_TOKEN)
5. If found: associate company to ticket, add conversation note
6. If not found: add "could not be identified" note

Data Transformations

• Ticket subject + content → Bedrock prompt
• Bedrock response → extracted company name
• "NONE" or empty → null

APIs

bedrock:InvokeModel (Claude 3.5 Sonnet, temp=0)
POST /crm/v3/objects/companies/search
PUT /crm/v4/.../associations/...

Metrics

EndCustomerExtracted, EndCustomerFound, EndCustomerNotFound, EndCustomerSkipped

4. Call Webhook

Sends sanitized ticket data to Subrosa AI webhook for processing. Handles skip conditions (no company, Fortinet).

Step Functions After extract end customer

Logic Flow

1. Extract email, company_id from enriched data
2. Sanitize text: unescape HTML, remove control chars, normalize whitespace
3. Skip if no company_id or company_id = "131" (Fortinet)
4. Look up end customer company for ticket
5. POST payload to Subrosa AI webhook (30s timeout)

Webhook Payload

Metrics

WebhookSuccess, WebhookSkipped, WebhookFailed

5. Update Ticket

Updates ticket with processing status and adds a conversation note documenting the action taken.

Step Functions Final step in pipeline

Logic Flow

1. If webhook skipped (no company): set ai_status = skip message
2. If webhook skipped (Fortinet): set ai_status = skip message
3. If webhook called: don't set ai_status (Subrosa AI sets it)
4. Add single-line conversation note (e.g., "Sent to AI — email, Company 123")

Data Transformations

• Skip reason + email + company → one-line message
• Falls back to Engagements API if ticket has no conversation thread

APIs

PATCH /crm/v3/objects/tickets/{id}
POST /conversations/v3/.../messages (or Engagements v1 fallback)

Metrics

None

6. Apply AI Replies

Scheduled Lambda that creates HubSpot quotes/deals from AI responses and posts formatted content to ticket conversations.

EventBridge Every 1 minute

Logic Flow

1. Search for tickets with processable ai_status (FULL, PARTIAL, NONE, etc.)
2. If ai_items_json present: create Deal (AI Generated pipeline) + Quote + Line Items
3. Associate deal to ticket (type 27), quote to deal (type 63)
4. Format ai_reply_draft as HTML paragraphs
5. Format ai_items_json as styled HTML table with pricing
6. Post combined content to conversation thread
7. Update ai_status: e.g., "FULL - Applied to conversation + quote"

Line Items Table Columns

Quote Creation

Pipeline: AI Generated (1276261847), Stage: Quote Created, Expiry: +30 days, Sender: Sales @ Wavelink, Template: Basic (443041226216)

Metrics

AIRepliesApplied, QuotesCreated, QuoteValue

7. Sync Presales Activity

Syncs ticket category from presales pipeline tickets to associated deals' presales_activity field.

EventBridge Every 5 minutes

Logic Flow

1. Search for tickets in presales pipeline with category set
2. For each ticket: get associated deals
3. Compare deal's current presales_activity with ticket category
4. If different: update deal property
5. If same: skip (idempotent)

Data Transformations

• Ticket hs_ticket_category → Deal presales_activity (direct copy)

APIs

POST /crm/v3/objects/tickets/search
GET /crm/v3/objects/tickets/{id}/associations/deals
PATCH /crm/v3/objects/deals/{id}

Metrics

None

8. Sync Deal Closedate / Team

Bidirectional sync between HubSpot deals and NetSuite opportunities. Handles closedate, stage, and team members.

Webhook HubSpot deal change   EventBridge Every 5 min (team poll)

Sync Directions

Team Sync Logic (HS → NS)

1. Get deal owner + collaborators from HubSpot
2. Resolve each to NS employee: explicit map → exact name → LIKE → split name
3. Preserve existing salesRoles from NS team
4. Idempotency: skip if all HS members already on NS team (subset check)
5. Debounce: skip if make__last_sync_datetime < 30s ago

Team Sync Logic (NS → HS)

1. Get NS salesTeam + BDM + CM fields
2. Map each to HubSpot owner by name
3. Primary → hubspot_owner_id, others → collaborators

Metrics

DealsSyncedToNS, DealsSyncedFromNS, DealSyncSkipped, DealSyncFailed, DealCloseDateSynced, DealStageSynced