Text screening

Alongside prescreening calls, Bond can screen a candidate by text message. Open a patient and go to the Messages tab, then click Text Patient. An AI assistant texts the patient over SMS (via your registered number), introduces the study, asks a few eligibility questions one at a time, and records the conversation - all asynchronously, so replies can arrive minutes or hours apart. The first message always identifies your team and includes a STOP opt-out, as carrier rules require.

The whole thread is saved on the patient, so it survives a refresh and any teammate can pick it up. You can also jump in manually at any point - pause the AI, type your own replies, then hand back to the AI.

When a patient asks a question back - “how long is the study?”, “will travel be covered?” - the assistant answers from that trial's protocol, drawing on the actual document rather than guessing. If the protocol doesn't cover something, it says so and offers to have a coordinator follow up.

AI screening settings

Above the thread is an AI screening settings menu where you choose which of this patient's eligibility gaps the AI should ask about - drawn from their criteria results. You can include items that need the patient's input(no EHR data), items checked at the screening visit (consent, pregnancy test, device access), and - opt-in - items currently marked not met. Each item is a checkbox, so you can trim the list; the selection is remembered per patient and applies to the next screening you start. The AI only asks what you pick, phrased as plain self-report questions (never as a diagnosis), and never decides eligibility itself.

Patient responses

As the patient replies, their answers are captured automatically and shown in a compact Patient responses checklist on the conversation: a progress bar and an answered/total count up top, then one status icon per question - a green check foryes, a red ✕ for no, amber for unclear, grey fordeclined, and an empty circle while awaiting a reply. Hover any row to see the patient's exact words. When the captured answers line up with the patient's criteria, an Update action appears that applies them to the criteria record in one click (stamped in the audit trail). Until you use it, the answers are just recorded for your review and don't change eligibility on their own.

Escalations & attention

If the assistant hits a situation a human should handle - a patient in distress, hostility, repeated confusion, an explicit request for a person, or a very long thread - it automatically escalates: it stops auto-replying and flags the conversation. Escalated patients surface in a Needs attention list in the sidebar (with a count) and get an Escalated chip in the patient table, so you can spot them at a glance without opening each one. Click into the thread and choose Resolve & resume AIonce you've handled it to clear the flag and hand the conversation back to the AI.

Opt-out & re-subscribe

Patients can reply STOP at any time to opt out; the AI confirms and stops texting them. Opt-out is tracked per patient, so one person opting out never blocks anyone else. If a patient later wants back in, they can reply START, or a coordinator can use the Re-subscribe action on their thread. Note that once someone replies STOP, the carrier also blocks that number until they reply START - re-subscribing in the app clears Bond's record, but a genuinely opted-out number stays blocked at the carrier until the patient texts START themselves.