In short
Consig administers behavioral health surveys such as the PHQ-9 over an outbound call. The instrument runs as a deterministic, scripted stage so every patient hears the same questions. Consig scores the answers and, based on thresholds your clinical team sets, hands the call to staff live or routes the result into their follow-up queue with the full call context.
Why does behavioral health screening suit a scripted voice call?
A validated screening instrument depends on consistent administration. The questions, the answer options and the order should not drift from one patient to the next, which is exactly the kind of step a generative model should not improvise. In Consig, the screen runs as a deterministic stage that behaves identically on the first call and the ten-thousandth.
Voice also fits the work. Until recently, performing a behavioral health assessment by automated call was close to impossible. Advances in voice AI now make these interactions possible with high fidelity and reliability, while a phone call reaches patients who never open a portal message or finish a paper form.
What happens when a score is concerning?
The handoff is the reason the workflow exists. Your clinical team defines the rules: which total scores or individual answers trigger a live transfer, and which go into a staff queue for follow-up. When a rule fires, Consig transfers the call through your call center or UC system with the call-attached data, so the person who picks up already sees the answers.
Every escalation needs an owner. Who responds, how quickly and what happens after hours should be decided before the first call is placed. A flagged result that reaches no one is worse than no program at all. We go deeper on escalation design in What AI Should and Should Never Do on a Patient Call.
How is sensitive information protected on a screening call?
Behavioral health answers are among the most sensitive data a call can carry. Identity is verified before the screen begins, using passkeys, device biometrics or one-time passcodes through Journey.ai's Zero Knowledge Network, so identity inputs stay out of the conversation layer, transcripts and any third-party LLM in the call path.
Consig also runs conversational context firewalls that keep sensitive data out of the model's reach, giving each AI component only what it needs. Because some states restrict what health information may touch a general-purpose AI path at all, Consig adjusts the technology call by call based on state law. The Dilemma at the Heart of Every AI Patient Call explains why.
Where does conversational AI fit around the screen?
The parts of the call around the instrument benefit from sounding natural: checking whether it is a good time, responding to a patient who wants to talk, finding a time for a follow-up appointment. Consig's workflows mix scripted stages with generative ones, so the screen stays precise while the rest of the call stays human.
How a Consig call works
- Connect. The call displays with branded caller ID and a reputation-managed number, and Consig confirms it is a good time to talk.
- Authenticate. The patient verifies identity with a passkey, device biometrics or a one-time passcode before any screening questions are asked.
- Administer the screen. The PHQ-9 or another approved survey runs as a deterministic stage, question by question, with answers captured as structured data.
- Score and route. Consig scores the responses against the thresholds your clinical team configured and decides whether to hand off now or queue for follow-up.
- Hand off with context. A concerning result transfers to staff through your call center or UC system, with the answers and call context attached.
Frequently asked questions
Is the PHQ-9 asked word for word?
It can be. In Consig, the screening instrument runs as a deterministic, scripted stage, so every patient hears the questions and answer options the way your clinical team approved them, in the same order. Generative AI is reserved for the surrounding conversation, such as confirming it is a good time or scheduling a follow-up.
Does Consig diagnose or interpret the results?
No. Consig administers the survey, calculates the score and applies the routing rules your clinicians define. It does not diagnose, interpret symptoms or give clinical advice. Anything that needs judgment goes to a person, either live on the call or through your team's follow-up queue with the full context attached.
What if a patient says something alarming mid-call?
Your clinical team defines escalation rules before launch, including rules on individual answers, not just the total score. When one fires, the call transfers to a staff member within seconds, with the answers so far attached. Deciding who covers those transfers, including after hours, is part of designing the program.
Can we use screening instruments other than the PHQ-9?
Consig is built to administer standardized surveys, with the PHQ-9 as the example on our solutions page. Workflows are configurable through the Consig portal or directly via JSON and the API, so another approved instrument can be set up as its own deterministic stage. New workflows can be designed in days, not quarters.
Can I try a behavioral health screening call?
Yes. The Behavioral Health demo on the Consig solutions page places an interactive example call that administers the PHQ-9 and shows the handoff on a concerning score. It is a quick way to hear how the scripted screen and the conversational parts of the call fit together.