Implementing an AI copilot with patient-facing agents: our talk at the AI Fertility World Conference in Cascais
We presented Session 10 at the 3rd AI Fertility World Conference in Cascais: Implementing an AI Copilot with Patient-Facing Agents. Most of the conference was about the lab (embryo selection, time-lapse imaging, models that grade and rank). Our session was about what happens before any of that, at the clinic's front door. The full talk is below.
The recording is narrated with an AI-generated voice reading the talk's script. The demo inside it uses a synthetic patient, and no real patient data appears.
The argument
An embryo model can only rank the embryos of patients who made it through the front office. Between a first inquiry and the lab there are five steps: the inquiry is answered, a consult is booked, the workup is done, a cycle starts, and only then does the lab see the patient. Each step loses people, and most of the losses happen at the phone, not in the lab.
The patients who leave are not only the ones nobody called back. A systematic review of 22 studies covering 21,453 patients found that psychological burden accounted for 14% of the reasons patients gave for stopping treatment, and that it was among the reasons that recurred at every stage of treatment.[1] A front office that answers slowly, or answers without care, adds to that burden.
So the most expensive AI problem in a fertility clinic is not embryo ranking. It is the inquiry nobody answered and the cycle that never started.
What is in the talk
- The funnel before the lab (0:34). Where patients are lost between a first inquiry and a started cycle.
- Fertiligent in 60 seconds (1:27). One AI agent across phone, fax, text messages, the web and email, with the clinic deciding what it says and does.
- An embryology report, explained (3:10). A recorded demo with a synthetic patient. The Day 5 report arrives from the lab system. A clinician reviews the explanation drafted for the patient in English, beside its Portuguese rendering, and approves it. The patient receives a text that her results are ready, reads them in her portal in Portuguese, and asks Eva, our patient companion, about them. Eva answers with the wording the clinician approved. When the patient asks about her chances, Eva does not guess: it tells her the care team will talk with her and puts the question in the coordinator's queue.
- How it fits with your EMR (6:08). The EMR stays the source of truth. Staff work in the Copilot, patients use the assistant, and staff decide what the assistant says and does.
- Why now (7:39). Agents changed how software is built: engineers decide what gets built, and agents handle how. We think clinic operations are next, with staff setting policy and approving, and the agent doing the routine work.
Read more
Each part of the talk has a longer write-up on this site:
- After-hours phone answering for fertility clinics
- Referral intake automation for fertility clinics
- Automating patient follow-up in a fertility clinic
- Emotional support in fertility care
- Eva, the patient companion
- Fertility EMR integration
To see it on your own clinic's questions, book a demo or try Eva.
References
Sergei Gorlovetsky, CEO, Fertiligent
