Migrate from Merlin to Pipedrive

Merlin conversation threads (chat text with timestamps) map to Pipedrive Note records. We preserve the conversation text as Note body content and attach each Note to a Person, Organization, or Deal record resolved from entity extraction or manual assignment during scoping. Thread-level metadata (model used per turn, token counts, latency) is not available in Merlin's export and cannot migrate. If the customer requires audit trails of AI interactions, we flag this gap during scoping and advise capturing model-attribution screenshots before account closure.

Merlin documents uploaded for Document Chat migrate as Pipedrive Attachments linked to the target Person or Organization record. As a supplementary step, we perform entity extraction on document content to identify person names, company names, email addresses, and phone numbers. Extracted entities become new Pipedrive People or Organizations created during migration. Free-tier documents capped at 10 pages may have been truncated on upload; we alert on any document exceeding this limit during scoping.

Merlin image generation outputs (PNG, JPG) migrate as Pipedrive Attachments. We preserve file type, creation timestamp, and binary content. Images are attached to the relevant Person, Organization, or Deal record identified during scoping or entity extraction. No structured metadata beyond creation date is available from Merlin for these outputs.

Merlin user account metadata (email address, plan tier, usage quota) maps to Pipedrive User records. The Merlin user email becomes the Pipedrive User email for login matching. Merlin does not store role or permission data, so all migrated users default to standard Pipedrive access; the customer's admin assigns roles post-migration. Owner assignment for conversations is not a first-class concept in Merlin, so conversation Notes do not carry an Owner during migration unless manually specified during scoping.

Person names and email addresses extracted from Merlin uploaded documents become Pipedrive People records. We create Person records with name fields populated from entity extraction, email populated where found in document text, and a note flagging the source document. Any extracted phone numbers, job titles, or company affiliations map to the corresponding Person custom fields if they exist in Pipedrive or are created during migration. Entity extraction accuracy depends on document formatting; we flag low-confidence extractions for manual review.

Company names and domains extracted from Merlin uploaded documents become Pipedrive Organization records. We create Organization records with name populated from entity extraction, website populated where a domain is identified, and a note flagging the source document. Organization-Person relationships are resolved by matching extracted company affiliation data from documents against Organization names created during migration.

Merlin has no native Deal object. If extracted entity data includes deal value, deal stage, or deal owner context from document content (e.g., contract values in uploaded proposals), we create Pipedrive Deals linked to the relevant Person and Organization. The customer defines deal stage mapping during scoping because there is no source-stage to translate. Pipeline and stage structure must be configured in Pipedrive before migration begins.

Merlin does not expose a schema for user-defined properties or custom fields. No structured data beyond conversation text and file blobs is accessible for export. We cannot migrate custom properties because none exist.

Merlin does not have a documented tagging or labeling system accessible via export. Tags as structured metadata do not exist in Merlin's data model and cannot migrate.

Merlin does not support pipeline or deal stage concepts. Pipedrive pipelines and stages must be created from scratch during migration. We work with the customer during scoping to define pipeline structure, stage names, and stage probabilities appropriate for their sales process. This is a configuration task rather than a data migration task.