Connect AI Bullhorncrm

CData Connect AI — Bullhorn CRM Skill

⚠️ Prerequisites — load these first

• connect-ai-base skill

Before proceeding, locate and read the connect-ai-base skill. If it is not available in the current environment (not loaded, not enabled, or not installed), stop immediately. Do not proceed with the task. Tell the user that the connect-ai-base skill is required and ask them to install and enable it before continuing.

This skill provides Bullhorn CRM-specific guidance for querying Bullhorn data through CData Connect AI. It composes on top of the connect-ai-base skill, which handles the discovery workflow, error recovery, SQL dialect, and query naming convention.

Precedence

This skill replaces getInstructions for the BullhornCRM driver. Do not call getInstructions for BullhornCRM — the guidance it provides is already incorporated here. Proceed directly to schema discovery (getTables / getColumns) after identifying the Bullhorn connection via getCatalogs.

Schema

BullhornCRM is a single-schema driver. The schema name is BullhornCRM.

SELECT [ID], [FirstName], [LastName], [Status]
FROM [YourBullhornConnection].[BullhornCRM].[Candidate]
LIMIT 100

Query Process

Bullhorn is a recruitment-centric relational model built around three core entities: Candidates, JobOrders, and Placements. Most analyses start with one of these and join outward.

1. Identify the connection — if unknown, call getCatalogs and filter for DRIVER = 'BullhornCRM'.
2. Pick the core entity — Candidate for talent/pipeline questions, JobOrder for open-role/client-demand questions, Placement for hires/revenue questions.
3. Inspect columns — call getColumns on the chosen entity before writing queries. Bullhorn instances are heavily customized; do not assume the schema matches Bullhorn UI labels.
4. Sample first — run a SELECT * … LIMIT 5 to validate structure and value formats (status fields are string picklists; valid values vary by tenant).
5. Filter on date columns — DateAdded and DateLastModified exist on most major entities; always include a date predicate when result sets could be large.
6. Join outward incrementally — JobSubmission links Candidate ↔ JobOrder; Placement represents successful hires. Validate each join with a small LIMIT before adding the next.

Data Model

Key Tables

Bullhorn's data model centers on the recruiting lifecycle:

• Candidate — Primary recruiting entity (people in the talent pool).
• ClientCorporation — Companies / clients. Supports parent-child hierarchies via ParentCompanyid.
• ClientContact — People who work at client companies; linked to ClientCorporation via Companyid.
• JobOrder — Open job requisitions (may also be referred to as job listings).
• JobSubmission — Many-to-many link between Candidate and JobOrder; tracks who was submitted for which role and the submission status.
• Placement — A successful hire; one Placement represents a candidate placed into a JobOrder. Carries direct FKs to both the JobOrder (Jobid) and the client company (Companyid).

Activity Tables

• Task, Appointment, Sendout, Tearsheet — Activity tracking.
• Note — Email content is not exposed via the Bullhorn API by default. If the customer routes emails through Bullhorn Automation, those emails populate the Note table and can be queried there.

Payroll & Financial Tables

• PayCheck, EmployeePay, Deduction — Payroll records.
• PlacementCommission — Commission attributable to a Placement.
• JobOrderRateCard, PlacementRateCard — Rate cards (billing/pay rates per job or placement).

History (Edit) Tables

For auditing field-level changes. Table names vary by tenant and driver version — confirm with getTables before querying:

• CandidateHistory (may also appear as CandidateEditHistory)
• JobOrderEditHistory
• PlacementEditHistory
• ClientCorporationEditHistory

These tables can be very large. Always include filters (entity ID + date range) before querying.

Key Relationships

Important Columns

Candidate

• ID — Primary key (numeric)
• FirstName, LastName, Name — Name is a derived/display field; write FirstName / LastName separately
• Status — String picklist (e.g., Active, Inactive, Placed)
• DateAdded, DateLastModified — Timestamps for incremental queries
• Email1, Email2, Email3, MobilePhone, WorkPhone — Contact fields
• CurrentCompany — Most recent employer (text)
• YearsExperience, DesiredSalary, DesiredPayRate
• OwnerCorporationid — FK to ClientCorporation (the owning agency)
• BranchID — FK to Branch (note: uppercase ID, not the typical lowercase id)
• IsDeleted — Soft-delete flag; filter to false for active records
• IsAnonymized — GDPR/privacy flag

ClientCorporation

• ID — Primary key
• CompanyName
• Status, Type
• DateAdded, DateLastModified
• MainPhone, BillingPhone, Fax
• ParentCompanyid — Self-referential FK for company hierarchy
• BranchID — Uppercase ID (FK casing exception, see Conventions)
• AnnualRevenue(Millions), Sector, BusinessSectors, NetTerms
• No IsDeleted column — ClientCorporation does not expose soft-delete

ClientContact

• ID
• FirstName, LastName, Name
• Companyid — FK to ClientCorporation.ID
• Email1, Email2, Email3, DirectPhone, MobilePhone
• Title, Department, Office
• Ownerid — FK to the user who owns the contact
• Reportstoid — Self-referential FK (reporting hierarchy)
• Status
• IsDeleted — Soft-delete flag
• IsDefaultContact — Whether this contact is the default for the company
• DateAdded, DateLastModified

JobOrder

• ID
• JobTitle
• Status — String picklist (e.g., Open, Closed, Accepting Candidates) — distinct from Open/Closed below
• Open/Closed — BOOLEAN column (literal name, slash included) — separate from Status. JobOrder has both, and they don't necessarily move together. Always quote with [] to escape the slash: [Open/Closed]
• ClientCompanyid — FK to ClientCorporation.ID
• Contactid — FK to ClientContact.ID (primary contact for the requisition)
• DateAdded, DateClosed, DateLastPublished, LastUpdated
• Ownerid — User who owns the requisition
• Branch, BranchID (uppercase ID)
• EmploymentType, StartDate, EstimatedEndDate, ScheduledEnd
• Locationid — FK to Location
• #ofOpenings — Literal column name (use [#ofOpenings])
• SalaryLow, SalaryHigh, PayRate, ClientBillRate, Mark-up%
• JobDescription, JobPostingURL
• IsDeleted

JobSubmission

• ID — Primary key
• Candidateid — FK to Candidate.ID
• Jobid — FK to JobOrder.ID
• Status — Submission stage (e.g., Submitted, Client Interview, Rejected)
• DateAdded, DateLastModified, DateWebResponse
• StartDate, EndDate
• BillRate, PayRate, Salary
• AddedByid — FK to user who created the submission
• Source, Comments
• IsDeleted, IsHidden
• Screening columns: ScreeningStatus, ScreeningScore, ScreeningFeedbackRating, ScreeningCompletedDate

Placement

• ID — Primary key
• Candidateid — FK to Candidate.ID
• Jobid — FK to JobOrder.ID
• Companyid — FK to ClientCorporation.ID (direct, not only via JobOrder)
• Contactid, BillingContactid, ApprovingClientContactid — Various contact references
• StartDate, ScheduledEnd, EstimatedEndDate, EmploymentStartDate, EffectiveDate, DocumentDeadline
• Status
• Salary, BillRate, PayRate, OvertimeBillRate, OvertimePayRate, DoubletimeBillRate, DoubletimePayRate
• PlacementFee(%), PlacementFee(Flat), Mark-up%, OTMark-up%
• EmploymentType, EmployeeType, TerminationReason
• DateAdded, DateLastModified

Common Query Patterns

Recently added candidates

SELECT [ID], [FirstName], [LastName], [Status], [DateAdded]
FROM [YourConnection].[BullhornCRM].[Candidate]
WHERE [IsDeleted] = false
  AND [DateAdded] >= DATEADD(day, -30, GETDATE())
ORDER BY [DateAdded] DESC
LIMIT 50

Candidates submitted to a specific job

SELECT c.[FirstName], c.[LastName], js.[Status], js.[DateAdded]
FROM [YourConnection].[BullhornCRM].[JobSubmission] js
JOIN [YourConnection].[BullhornCRM].[Candidate] c
  ON js.[Candidateid] = c.[ID]
WHERE js.[Jobid] = 12345
ORDER BY js.[DateAdded] DESC

Active placements with candidate names

SELECT p.[ID], c.[FirstName], c.[LastName], p.[StartDate], p.[Status], p.[BillRate]
FROM [YourConnection].[BullhornCRM].[Placement] p
JOIN [YourConnection].[BullhornCRM].[Candidate] c
  ON p.[Candidateid] = c.[ID]
WHERE p.[Status] = 'Active'
ORDER BY p.[StartDate] DESC

Open job orders (using either Status or Open/Closed)

-- By Status picklist
SELECT [ID], [JobTitle], [Status], [DateAdded]
FROM [YourConnection].[BullhornCRM].[JobOrder]
WHERE [Status] = 'Open'
  AND [IsDeleted] = false
ORDER BY [DateAdded] DESC

-- Or by the boolean Open/Closed column (slash in the name requires brackets)
SELECT [ID], [JobTitle], [Status], [Open/Closed]
FROM [YourConnection].[BullhornCRM].[JobOrder]
WHERE [Open/Closed] = true
  AND [IsDeleted] = false

Recruiting funnel: JobOrder → JobSubmission → Candidate → Placement

SELECT jo.[JobTitle], jo.[Status] AS JobStatus,
       js.[Status] AS SubmissionStatus,
       c.[FirstName], c.[LastName],
       p.[ID] AS PlacementID, p.[StartDate]
FROM [YourConnection].[BullhornCRM].[JobOrder] jo
LEFT JOIN [YourConnection].[BullhornCRM].[JobSubmission] js ON js.[Jobid] = jo.[ID]
LEFT JOIN [YourConnection].[BullhornCRM].[Candidate] c ON js.[Candidateid] = c.[ID]
LEFT JOIN [YourConnection].[BullhornCRM].[Placement] p
       ON p.[Candidateid] = c.[ID] AND p.[Jobid] = jo.[ID]
WHERE jo.[DateAdded] >= '2025-01-01'
ORDER BY jo.[DateAdded] DESC
LIMIT 100

Client revenue rollup (Placement × ClientCorporation)

SELECT cc.[CompanyName],
       COUNT(p.[ID]) AS Placements,
       SUM(p.[BillRate]) AS TotalBillRate
FROM [YourConnection].[BullhornCRM].[Placement] p
JOIN [YourConnection].[BullhornCRM].[ClientCorporation] cc ON p.[Companyid] = cc.[ID]
WHERE p.[StartDate] >= '2025-01-01'
GROUP BY cc.[CompanyName]
-- ORDER BY on aggregate aliases (TotalBillRate) is not supported by the driver; sort client-side

Candidate pipeline by submission status

SELECT js.[Status], COUNT(*) AS CandidateCount
FROM [YourConnection].[BullhornCRM].[JobSubmission] js
WHERE js.[DateAdded] >= '2025-01-01'
GROUP BY js.[Status]
-- ORDER BY on COUNT(*) / aggregate aliases is not supported by the driver; sort client-side

Stored Procedures

The BullhornCRM driver exposes file-attachment procedures: DownloadFile and UploadFile. Both involve disk-based or stream-based parameters and have specific cloud-compatibility caveats.

DownloadFile

Downloads a file attached to a specified Bullhorn entity.

Cloud-compatible call shape — omit both OutputFolder and FileStream; the response contains file content as base64. This matches the established CData download pattern across drivers.

{
  "catalogName": "YourBullhornConnection",
  "schemaName": "BullhornCRM",
  "procedureName": "DownloadFile",
  "parameters": {
    "EntityType": "Candidate",
    "EntityId": "12345",
    "FileId": "67890"
  }
}

UploadFile — Not currently supported in cloud

UploadFile accepts either FileLocation (server-side disk path, unavailable in cloud) or Content (Java InputStream, unable to pass through the MCP interface). There is no base64 string alternative for the upload path on the BullhornCRM driver. Treat file uploads as a current limitation — do not invent parameter names the driver does not accept. Support for file uploads via stored procedures is planned — check for updates if this capability is needed.

Write Operations

The BullhornCRM driver supports a limited set of writes — the most common are INSERT INTO Candidate and UPDATE JobOrder. Other tables may or may not be writable; check the column-level Readonly flag via getColumns (Readonly=false means writable) and test in a sandbox before relying on it.

INSERT example

INSERT INTO [YourConnection].[BullhornCRM].[Candidate]
  ([FirstName], [LastName], [Email1], [Status])
VALUES
  ('Jane', 'Smith', 'jane@example.com', 'Active')

UPDATE example

UPDATE [YourConnection].[BullhornCRM].[JobOrder]
SET [Status] = 'Closed', [Open/Closed] = false
WHERE [ID] = 12345

Write access is controlled by two layers: the Connect AI connection's readonly setting (CData side) and the authenticated Bullhorn API user's permissions (Bullhorn side). If a write fails, check both.

Bullhorn-Specific Conventions

• Prefer literal true/false on boolean columns. Some boolean columns reject numeric 1/0 (a few such as IsDeleted do accept them, but not all) — e.g. WHERE [Open/Closed] = 1 fails with "Cannot decipher where clause as Boolean." Use WHERE [IsDeleted] = false, WHERE [Open/Closed] = true. This applies to both SELECT filters and UPDATE SET clauses.
• ORDER BY cannot reference aggregate functions or aggregate aliases. The CData BullhornCRM driver rejects ORDER BY COUNT(*) DESC, ORDER BY <aggregate_alias> DESC (e.g., ORDER BY TotalBillRate DESC where TotalBillRate is a SUM), and ordinal references like ORDER BY 2 DESC — all surface as "Invalid column name 'COUNT'/'SUM'/etc. for table '<TableName>'". GROUP BY with aggregate expressions in the SELECT clause works correctly; only the ORDER BY clause is affected. Workaround: drop the ORDER BY on aggregates and sort client-side, or ORDER BY a non-aggregate column already in the result set.
• FK column casing is mixed, not uniform. Most FK columns use a lowercase id suffix (Candidateid, Jobid, Companyid, ClientCompanyid, Ownerid, Contactid, Reportstoid, AddedByid, Categoryid, ParentCompanyid), but several use uppercase ID: BranchID, MasterUserID, ClientContactID. ExternalID is an external/legacy reference, not a true FK. Always confirm casing with getColumns before joining — Bullhorn does not have a single consistent rule.
• Primary keys are always uppercase ID. Only FK columns vary; the table's own primary key is consistently ID.
• Column names commonly contain special characters. Slashes (Open/Closed, Culture/Perks), hashes (#ofOpenings), percent signs (Mark-up%, Tax%, PlacementFee(%)), parentheses (Reportingto(contact)id, WillRelocate?), question marks. [] quoting is required, not stylistic. Plain identifiers will fail to parse on many columns.
• JobOrder.Status and JobOrder.[Open/Closed] are distinct fields. Status is a VARCHAR picklist (Open / Closed / Accepting Candidates / etc.); [Open/Closed] is a separate BOOLEAN. They don't necessarily move together in legacy data — check both for "is this job open?" questions, and confirm which one the user means.
• Status fields are string picklists, not numeric codes. Filter with literal strings ('Active', 'Open', 'Submitted'). Picklist values differ by Bullhorn configuration; confirm with SELECT DISTINCT [Status] FROM <table>.
• IsDeleted is present on most transactional tables but NOT on ClientCorporation. Candidate, ClientContact, JobOrder, JobSubmission, and Placement all expose IsDeleted and should be filtered to false for active queries. ClientCorporation does not have this column.
• History tables are append-only and very large. CandidateHistory (or CandidateEditHistory — name varies by tenant; confirm via getTables), JobOrderEditHistory, PlacementEditHistory, ClientCorporationEditHistory track every field change. Always filter on entity ID and a date range.
• Custom fields are pervasive and use mixed casing. Bullhorn tenants are heavily customized — every entity has many customText, customInt, customFloat, customDate, customTextBlock slots. Casing of the first letter varies within the same table (e.g., customDate1 vs CustomDate10); always confirm exact names via getColumns. Encrypted custom fields (CustomEncryptedText1 through CustomEncryptedText10) exist on Candidate and Placement for PII data.
• Note table is sparse without Bullhorn Automation. Email content is not in Note by default — only present when the tenant uses Bullhorn Automation to ingest emails.
• Permission errors return HTTP 403 with "No read rights". If a query against an entire entity returns 403 / "No read rights", the API user lacks read scope for that object; route the user to their Bullhorn administrator. If only specific fields return errors (e.g., "No read rights. Shift" on JobOrder), drop those fields from the SELECT and re-run.
• Use DateAdded for incremental loads. It's monotonically increasing and exists on every transactional entity. DateLastModified is also available where mutations are common.
• Join on numeric ID / id columns, not names. CompanyName, JobTitle, candidate names are not guaranteed unique.
• DownloadFile works in cloud; UploadFile does not. Omit both OutputFolder and FileStream on DownloadFile for a base64 response. UploadFile has no base64 alternative for the upload path — support is planned but not currently available.