Greenhouse BambooHR Integration: Automate New-Hire Onboarding
Stop re-keying accepted offers. Design a controlled Greenhouse-to-BambooHR handoff that prepares employee records, catches mismatches, and gives HR a clear approval point before day one.

TL;DR: what you'll build. Greenhouse's native export to BambooHR misses fields quietly, and HR finds out on a new hire's first day. If you own HRIS or onboarding, this guide shows how to wrap a governed layer around your Greenhouse BambooHR integration on Major. The layer adds a review queue, deterministic checks, a narrow mapping agent and an HR approval gate, so every hire lands in BambooHR correctly before day one.
The short version
The fastest way to automate Greenhouse-to-BambooHR onboarding is to keep Greenhouse's native "Export to BambooHR?" option for clean hires. Then put a governed layer around it for everything else.
The native export is a one-time, one-way push. It sends the following:
- Contact details
- Start date
- Job title
- Department
- Office
- Documents
It works until a department string does not match BambooHR exactly, and then HR finds out on day one.
If you own HRIS or onboarding, build that governed layer on Major. It has four parts:
- A durable onboarding queue app
- A workflow that does the deterministic checks
- A narrow agent that only resolves ambiguous fields
- An HR approval gate before anything writes to BambooHR
The build section is below. First, it helps to see exactly where the native path fails.
Why the native Greenhouse BambooHR integration breaks
Greenhouse documents the export as manual per candidate. A recruiter marks the candidate hired, sets the export option to yes, and the payload goes once.
There is no queue to inspect and no approval step for HR. Failures are quiet, which means nobody gets an error message they can act on before the start date.
Three failure modes that show up before a Monday start
These three show up every Friday before a Monday start:
- A sourcer types a department that does not exist verbatim in BambooHR, so department or office does not sync.
- A personal email lands in the wrong field.
- A duplicate work email collides with an existing record.
Why volume makes this worse
Forty hires a month means roughly forty places for a quiet miss. Each one is small on its own, but together they add up to a steady stream of day-one cleanup for HR.
The fix is not a smarter model on every hire. It is a staging step that catches the mismatch before the write. Think of the staging step as a loading dock: every record stops there, gets checked against the manifest, and only moves into the warehouse once someone signs for it.
What the two APIs let you build against
Before you design the workflow, it helps to know what each system exposes, how it authenticates, and how it reports errors. These details decide how your checks and retries behave.
Greenhouse Harvest API v3
Custom integrations authenticate with OAuth 2.0 client credentials. You request a token from https://auth.greenhouse.io/token.
The older static-key Basic Auth is being retired, so do not start a new build on it. Anything you build on the old method will need to be redone.
The objects you need are:
- Candidates
- Applications
- Offers
- Jobs
The reference lists Mark an application as hire and List applications.
We could not confirm a documented "hired" webhook payload. A webhook is a push notification from one system to another, like a doorbell that rings when something happens. Without a confirmed doorbell, you poll instead: check recently hired applications on a schedule until you verify a webhook in your own account.
Rate limits are not stated in the docs we reviewed, so build in backoff. Backoff means waiting longer between retries when a request fails, instead of hammering the API.
BambooHR REST API
Auth is an API key tied to a user with the right permissions. The key is scoped to your subdomain, using the base URL https://api.bamboohr.com/api/gateway.php/<subdomain>/v1/.
The technical overview documents the three responses that matter here:
- 409: a duplicate. Duplicate employee emails are not allowed.
- 406: the request names a field that does not exist in your configuration.
- 429: too many requests. The response carries a
Retry-Afterheader you must honor.
Each of these maps to a specific handling rule in the edge cases section further down.
The compliance frame for new-hire data
New-hire data is employment data, not health data, so HIPAA is not the frame. The relevant guardrail is the EEOC's guidance on automated tools in employment.
Keep this workflow administrative. Mapping a field is fine. Rejecting a person is not something an agent here should do.
Build the Greenhouse to BambooHR onboarding workflow on Major
Major is the enterprise platform where agents build the software they run on. For this handoff, the repeatable steps become a deterministic app and workflow, and the model is reserved for the one judgment call.
Reason once about how "Eng - Platform" maps to "Engineering," store the rule, and reuse it on every later hire. That keeps model usage limited to the cases that actually need judgment.
- Connect Greenhouse and BambooHR from Major's connector catalog, confirm which operations your connector version exposes, and if one is missing add a custom API connector against the Harvest v3 base URL or the BambooHR gateway URL, with credentials held in Major's credential proxy and never in workflow code. You should see both connectors listed as connected, with no API keys or tokens visible anywhere in your workflow definitions.
- Build the app that holds state: an onboardingqueue table keyed on `greenhouseapplicationid` as the idempotency key, with fields for work email, start date, job title, raw department, raw office, mapped department, mapped office, mapping confidence, approval status, BambooHR employee ID and last error, plus a second onboardingtasks table for the pre-start checklist (equipment, IT accounts, manager intro), and use this app as the HR review screen. You should see an empty queue and an empty task list in the app, ready for the first hired application.
- Add the agent for exceptions only, so that when a department or office value does not match, it proposes the closest valid BambooHR value with a confidence score or says it cannot, and it never writes to BambooHR, never invents a value outside the provided list, and never approves a record. You should see the agent's proposal and confidence score appear on the queue row, with the record still waiting for HR.
- Set the workflow trigger to poll Greenhouse for newly hired applications every 30 minutes, insert a row for each one not already queued, pull the candidate and offer fields, then run the deterministic checks with no model: look up the work email in BambooHR to catch the 409 case, and compare the raw department and office against BambooHR's configured values, which require an exact match. You should see new hires appear in the queue within 30 minutes of being marked hired, with clean matches filled in and mismatches flagged for the agent.
- Wire in the HR approval gate so every record sits in
pendinguntil an HR user approves, edits or rejects it, and on approval the workflow calls the BambooHR Employees endpoint, stores the returned ID and creates the checklist tasks. You should see the approval status change, a BambooHR employee ID stored on the row, and the onboarding tasks created for that hire. - Test the failure paths before real hires depend on them: force a 429 and confirm the retry waits for the delay, send a duplicate work email and confirm it routes to HR, send an unknown field and confirm the write is held, and run the poll twice to confirm no second row appears. You should see each test case land in the expected state, with the reason recorded in the last error field instead of a quiet drop.
- Roll out with permissions by using SSO and role-based access to limit candidate data to HR and HRIS admins, scoping credentials to reading applications, candidates and offers in Greenhouse and creating or updating employees in BambooHR, and logging every insert, agent proposal, approval and write with identity and timestamp. You should see only HR and HRIS admins able to open the queue, and an audit entry that answers "who approved this record" for every hire.
Build an onboarding queue app that stages every newly hired Greenhouse application, checks the work email, department and office against BambooHR, and holds each record for HR approval before creating the employee.
Start building this on Major and describe the onboarding queue app in a sentence.
Why the idempotency key matters
An idempotency key is a unique label that guarantees the same action only happens once, no matter how many times you try. It works like a coat-check ticket: hand in the same ticket twice and you still get one coat back, not two.
Here, the Greenhouse application ID is that ticket. Because the workflow polls every 30 minutes, it will see the same hire more than once, and the key stops that from becoming two staged rows.
Why the agent stays narrow
The agent only touches the one step that needs judgment: matching an unfamiliar department or office string to a valid BambooHR value. Everything else is a rule a workflow can apply the same way every time.
This also keeps the design inside the EEOC guardrail. The agent maps fields. A person approves the record.
Edge cases to handle before go-live
Each of these comes from a real response code or a timing gap in the handoff. Handle them in the workflow, not by hand on the day of the start.
Rate limits and duplicates
- 429 from BambooHR. Retry with the
Retry-Afterdelay. No tight loops. - 409 duplicate. Never overwrite. Route to HR with both records visible so a person decides whether it is a re-run or a real duplicate.
Bad fields and partial writes
- 406 unknown field. Hold the write and flag the record. The native export's weakness is dropping the field quietly.
- Partial write. If the employee is created but checklist tasks fail, retry the tasks using the stored
bamboohr_employee_idinstead of recreating the employee.
Repeats and changes
- Repeat polls. The Greenhouse application ID key stops a second staged row for the same hire.
- Edited offer. Re-pull offer fields at approval time rather than trusting the staged copy. Offers can change between the hire and the approval, and the start date or title HR approves should be the current one.
Metrics that show it works
Track these numbers from the first week:
- Time from hired to BambooHR record created
- The share of hires that needed agent-assisted mapping
- How many records sit in the review queue, and for how long
- Mismatches caught before a write
- Manual re-keys per hire
The agent-assisted share should fall over time as saved mappings cover more department and office strings. That is the cost curve this design is built for: front-loaded reasoning, then flat.
In plain terms, the first few weeks teach the system your organization's naming habits. After that, most hires pass the deterministic checks and go straight to HR approval without any model call.
For the surrounding workflows, see AI agents for HR and AI agents for recruiting.
Before you go live
- [ ] Greenhouse is connected with OAuth 2.0 client credentials, not the retiring Basic Auth key.
- [ ] The onboarding queue is keyed on the Greenhouse application ID, and a repeat poll creates no second row.
- [ ] Raw department and office values are checked for an exact match against BambooHR's configured values before any write.
- [ ] The agent can only propose values from the provided BambooHR list and cannot write, approve or invent values.
- [ ] Every record stays pending until an HR user approves, edits or rejects it.
- [ ] 429, 409 and 406 responses each have a tested handling path: honor the retry delay, route duplicates to HR, hold unknown fields.
- [ ] Offer fields are re-pulled at approval time, and failed checklist tasks retry against the stored BambooHR employee ID.
- [ ] Access is limited to HR and HRIS admins through SSO and roles, and every insert, proposal, approval and write is logged with identity and timestamp.
Start building on Major
Build the onboarding queue, the field-mapping agent and the HR approval gate for your Greenhouse-to-BambooHR handoff on Major, with scoped credentials and an audit trail from the first hire.
Related articles
- Automate Claim Document Signing in Guidewire ClaimCenter with DocuSign
- Automate Marketo Lead Audiences in Google Ads with AI
- Automate ZoomInfo-to-Outreach Lead Handoffs on Major
About the author
Ritvij Gautam is Head of Growth at Major.
Last reviewed November 6, 2026 by Ritvij Gautam.
Frequently asked questions
- What information can Greenhouse send to BambooHR when a candidate is hired?
- Greenhouse's native export sends first and last name, personal and work email addresses, home/work/mobile phone numbers, start date, job title, department, office (mapped to BambooHR's Location field), and attached documents. Optional custom fields can add starting salary, date of birth, hourly rate, or contract value. Department and office only sync if the values match BambooHR's configured names exactly.
- Can Greenhouse and BambooHR sync employee data both ways?
- No. Greenhouse documents this as a one-way export: data flows from a hired candidate's Greenhouse Recruiting profile into BambooHR, triggered manually per candidate when a recruiter marks them hired and enables the export option. BambooHR does not send employee updates back into Greenhouse through this integration.
- How should HR handle missing or mismatched employee fields?
- Never let automation guess silently. Stage the record in a review queue, flag the specific field that failed to match (commonly department or office, since BambooHR requires an exact string match), and require an authorized HR user to correct or approve it before any employee record is created, preserving the original source values and an audit trail.
- How can you automate a Greenhouse-to-BambooHR onboarding handoff with Major?
- Major holds staged hire records in an app with full state and audit history, runs a workflow that polls Greenhouse, checks for duplicates and field mismatches deterministically, and only calls an agent to propose a mapping for ambiguous fields. An HR approval gate sits before every write to BambooHR's Employees API, with credentials scoped through Major's credential proxy.