Migrate from Merlin to Salesforce Sales Cloud

Merlin user account metadata (email, plan tier, usage quota) retrieved from account settings maps to Salesforce User records. Owner assignment for conversations is not a first-class concept in Merlin, so we use the user email as the CreatedBy and OwnerId on migrated records. Any Merlin user without a corresponding Salesforce User is held in a reconciliation queue for the customer's admin to provision before record migration proceeds.

Merlin conversation threads export as plain text with timestamps but without per-turn model attribution, token counts, or latency metadata. Each thread maps to a Salesforce Note record. We preserve thread context by retaining the full text and timestamp, but we flag that the model attribution metadata is permanently absent from the Merlin export and cannot be reconstructed. Notes attach via ContentDocumentLink to the nearest applicable CRM record (Account, Contact, or Opportunity) once the CRM schema is rebuilt.

Documents uploaded to Merlin Document Chat are stored in user-scoped storage. Free-tier uploads are capped at 10 pages; Pro uploads support unlimited document length. We capture the full binary content, file type, and creation date for each document. Documents migrate as Salesforce ContentDocument records in the Files home or within a dedicated Salesforce Library. If the document references person or company names identified during entity extraction, we link the ContentDocument to the corresponding Account or Contact record.

Merlin's Document Chat produces question-and-answer pairs against uploaded files. These Q&A pairs export as part of the conversation thread. We extract them and create Salesforce Note records with a standard body format (Question: ... Answer: ...) so that the Q&A context is searchable within Salesforce. Each Note references the parent ContentDocument so users can trace the answer back to the source file.

Merlin image-generation outputs are treated as attachments. We preserve the file type, creation date, and binary content as Salesforce ContentVersion records. Image outputs do not map to any standard Salesforce CRM object; we attach them to the user's personal Salesforce Files or to a designated Library for retention. The customer specifies the target Library during scoping.

Merlin does not store contacts as structured records, but person names appearing in uploaded documents or conversation text may represent existing or prospective contacts. We perform entity extraction as a supplementary migration step: using NLP-based parsing on conversation text and document content, we identify person names, email addresses, and phone numbers. These merge into a Contact deduplication queue where the customer's admin reviews and merges duplicates before final import. This is a best-effort reconstruction, not a guaranteed complete contact list.

Similarly, company names appearing in Merlin conversation text and uploaded documents are extracted during the entity-extraction step. Extracted company names create Account candidate records in a reconciliation queue. Email domain patterns in conversation text (e.g., @acme.com in an email address) provide additional signal for Account matching. The customer's admin reviews, deduplicates, and approves before these records are created in Salesforce. This reconstruction is best-effort and does not replace manual CRM population.

Merlin does not expose a schema for user-defined properties or custom fields beyond conversation text and file blobs. There are no Merlin custom properties to migrate. Salesforce custom fields must be designed and created from scratch based on the customer's business requirements. We include up to ten standard custom fields in the migration scope; additional fields are scoped separately.

Merlin has no documented tagging or labeling system accessible via export. No tags migrate. If the customer uses document-level categorization within Merlin that can be captured manually (e.g., as a CSV export of folder names or document labels), we map those to Salesforce multi-select picklist fields on ContentDocument or to Topics with TopicAssignment records.

Merlin is not a sales or deal-tracking CRM. No pipeline, deal, or opportunity objects exist in Merlin's data model. The destination pipeline stages, sales processes, and record types must be designed from scratch in Salesforce based on the customer's business process. We include one standard Opportunity stage set (Prospecting through Closed Won/Lost) in the base schema scope.

Merlin does not expose workflow, automation, or sequence data structures via export. There are no automations to migrate. Salesforce Flow rebuilds must be designed by the customer's admin or a Salesforce implementation partner based on the customer's documented business process.

Merlin does not store structured reports or dashboards. There are no reports or dashboards to migrate. Salesforce report and dashboard design is outside the data migration scope and is covered by a separate Salesforce admin engagement.