One API call. One AI interview.
Already run your own ATS or pipeline? Create an interview for a candidate and get back a link, a status, and a result. Your system stays in charge; Tarkflo only runs the interview and hands back what happened.
POST /v1/interviews, from your own server, in about ten minutes.
{
"job_id": "66f1c2a9b3d4e5f6",
"email": "ananya.rao@example.com"
}{
"status": "ready",
"interview_url": "https://hire.tarkflo.com/?token=…"
}Interview as a Service, start to finish
Four steps, and the second one is the only part that happens outside your system: the candidate, on Tarkflo's own interview page.
Create an interview
One call: a job and a candidate. Tarkflo returns a link, shown once, like a key’s own secret.
The candidate interviews
They open the link in their own browser and talk to Tarkflo’s AI interviewer. Nothing on your side drives it.
Poll for status
Awaiting a slot, ready, in progress, completed. Check on whatever schedule suits you, or listen for the webhook.
Read the result
A score, a verdict, a hire-bar decision, and a scoped link to the full report. Never the rubric it was graded against.
Already run your own pipeline? Sync straight into it.
Reading a CV is a best guess, and saving a candidate is a commitment. So they are two separate calls, and nothing reaches the pipeline until you say so.
01
Extract
Send the CVs for a job. Tarkflo reads the name, email and phone from each file and sends them back, along with any existing candidate who matches. Nothing is saved, no candidate is created and no email goes out.
02
Submit
Send the details you reviewed. Each candidate enters the job's pipeline with their CV attached, and the job's workflow starts. Tag each one with your own ID so results map back to your records, and send an idempotency key so a retry never adds anyone twice.
{ "status": "ok", "name": "Ananya Rao",
"email": "ananya.rao@example.com",
"cv_id": "cv_7Qm2xN4v…" }{ "application_id": "66f2d4b8…",
"name": "Ananya Rao",
"external_id": "greenhouse-48213" }Extract reads the CVs and saves nothing. Submit saves the candidates you kept. Illustrative: the names are made up.
The API operates the pipeline. It does not configure it.
Workflows, interview templates and scoring rubrics are sets of linked settings. One API call cannot show you the checks the dashboard applies, so the API refers to them by ID and never edits them.
With the API
- Create an interview for a candidate and get back a link and, once it finishes, a score and a verdict
- Submit candidates into a job's pipeline
- Read name, email and phone out of CVs
- Read your jobs, with their descriptions and how many have applied
- Read your workflows round by round, and the interview templates they run
- Carry your own ID and metadata through to the result
In the Tarkflo dashboard
- Create and edit jobs
- Build workflows, interview templates and scoring rubrics
- Run, watch and review interviews
- Billing, members, single sign-on and API keys
Interviews run through the candidate's own link, in their own browser. The API can create that link, but it never drives the interview itself, and never sees its questions or rubric.
Access is yours to manage.
Create API keys yourself in the Tarkflo dashboard. Nothing to request, and nothing to wait for.
How access works
- Each key acts as a user with the role you choose, so it can only do what that role can
- Set it to expire in a day, a month, a year, on a date you pick, or never
- A key is shown once, when you create it, and you can revoke it at any time
- Every call lands in your audit log under the key and its own service account address
- Owners have access by default and can grant it to a role
Start with the product. Add the API when you need it.
Let candidates move when they're ready, and let Tarkflo handle the work between application and decision. Start with your next role and see how much of the process it can take off your plate.
No credit card required · Start in minutes