- Collect configurable participant responses
- Draft pairs and triplets
- Simulate fictional participants
- Prepare editable messages and reminders
- Keep session history and evaluation records
Facilitator’s guide
Practise once. Then run it for real.
A complete first-time walkthrough for rehearsing the Impact Collective loop safely, preparing a live session, forming useful pairs or triplets, and following commitments through to evidence.
Orientation
You facilitate; the app prepares the evidence.
The app helps people articulate what matters, drafts possible groups, prepares communications, captures personal commitments, and records follow-up. It does not replace facilitator judgment.
- Judge whether a conversation is genuinely useful
- Approve or remove a suggested group
- Send invitations and create WhatsApp groups
- Respond to consent, context, and safeguarding concerns
- Check evidence before verifying completion
The operating ruleAI may simulate and draft. A facilitator approves, communicates, adapts, and verifies. Nothing is automatically published or sent.
Collective visibility, clear ownership
Every active facilitator can review every live and archived session. When you open another facilitator’s session, the desk switches to read-only observer mode. Private test rooms remain personal sandboxes.
A · Safe rehearsal
Run the simulated tutorial session
The tutorial has its own private test room. It deliberately includes weak matches so you can practise noticing problems and making better choices. Complete all eight steps in order.
- 01
Rehearse a 250-person On Purpose event
In the Facilitator desk, choose Rehearse On Purpose, then Start rehearsal with 250 fictional participants. Use the normal event-wide matching, card review and reveal steps.
What to notice: The separate fictional session leaves your real event intact. Loading data makes no AI request; later AI drafting uses the configured service and may incur charges. Public intake is closed. Open fictional participant links to practice the participant experience. - 02
Sign in to your facilitator workspace
Open the Facilitator desk and enter the personal sign-in code supplied by Nadia or another administrator.
What to notice: The same code opens your named workspace on another trusted device. You can read collective session history, while only the named lead facilitator can change a session. - 03
Start the guided simulation
Open the Guided simulation and choose Trigger Articulate · collect 8 responses.
What to notice: Eight fictional people appear in a private rehearsal. Their phone numbers are intentionally invalid and no real message can be sent. - 04
Inspect the participant responses
Select several people and compare what has their attention, what help they need, what they offer, and their realistic capacity.
What to notice: You are looking for a practical reason to talk—not merely a shared topic. - 05
Generate the deliberately weak first pass
Choose Trigger Choose · generate first pass. Review why the suggested pairs are weak, then tick the people who need another look.
What to notice: The tutorial makes poor fits visible so you can practise applying judgment rather than accepting a score. - 06
Run the additive second pass
Choose Run additive second pass. Compare each new suggestion with the original and approve only the useful round-two drafts.
What to notice: Existing suggestions remain visible. The app adds alternatives without silently rewriting history. - 07
Rehearse consent and the handoff
Simulate the Yes responses, then create the simulated WhatsApp groups and review the welcome messages.
What to notice: You see both the facilitator and participant views. In a live session, these messages are still reviewed and sent manually. - 08
Capture commitments
Choose Simulate commitment forms and inspect each person’s action, date, and expected evidence.
What to notice: Each participant owns one personal commitment. A group promise is not substituted for individual responsibility. - 09
Verify evidence
Simulate the day-7 pulses, distinguish Done from Moving, Stuck, or Paused, then choose Verify completed evidence.
What to notice: Only a Done commitment with matching evidence and facilitator verification contributes to the north-star measure.
Safe tutorial boundary
The tutorial uses fictional people and invalid +351 000… numbers. Actions such as “send” or “create group” advance the rehearsal and create an internal message record; they do not contact WhatsApp.
B · Live delivery
Run a real session
Use these six phases as your runbook. The first two happen before group formation; the final phase may happen days later.
Before inviting participants
Prepare the session
- Open the Facilitator desk and confirm that your name and organization are correct.
- In Pilot settings, give the session a clear name and describe what the gathering is trying to make possible.
- Choose the Project template or Inquiry / double-diamond template, then edit, add, remove, or reorder questions.
- Mark which questions are required, used for matching, and suitable to share with a matched group.
- Choose an AI preset if you plan to test AI-assisted matching. Terra · low is the recommended baseline.
- Save the session configuration and copy the participant form link.
Before or during the gathering
Collect articulations
- Share the session-specific form link—not the generic home-page link.
- Tell participants that a rough answer is welcome and that receiving a match is not an obligation.
- Use the People tab to confirm responses are arriving and to notice missing or unclear information.
- Keep contact details at the facilitator desk. They are not shared merely because the app suggests a group.
- Export the CSV if you need a separate working copy or preparation for another community platform.
When enough responses have arrived
Draft the room
- Use fast topic-first matching for an immediate, legible baseline; use AI-assisted matching for a context-aware comparison.
- When you already know who should meet, select those people in Second pass and use Create exactly this group. This facilitator-created draft uses no AI and leaves every existing group unchanged.
- Review the Room board as a first draft. Look for a useful shared question plus complementary needs, offers, or perspectives.
- Open Groups & WhatsApp to edit the reason, opening question, and useful outcome.
- Remove a poor draft. Never keep a weak group simply to ensure everybody is matched.
- Use Second pass for anyone unmatched, mismatched, or worth considering for an additional conversation.
- If testing AI, rate each untouched suggestion Useful, Mixed, or Poor before substantially editing it.
In the room or after facilitator review
Form the conversations
- Approve only groups you are willing to stand behind as facilitator.
- If people will meet later, copy each invitation and send it manually. Participants can choose Yes, Not now, or Pass.
- If people have already met in the room, choose Group met · open commitments.
- Create any WhatsApp group yourself, paste the reviewed welcome, and mark the group created only after doing so.
- Give groups about 25 minutes and treat the suggested opening question as a starting point, not a script.
At the end of each conversation
Move from talk to action
- Ask each person to record one action they personally own, a realistic date, and observable evidence.
- Use Reminders for invitations or commitment requests that need manual follow-up.
- Do not turn a tentative idea into an assigned task. A valid outcome can be a small test, a decision, or an honest choice not to proceed.
- Use Follow-on support when a group wants a deeper session or a different mix of skills.
- For projects outside the first event, copy the generic Challenge Lab link from Follow-on support. Review every open-link application before adding it to or setting it aside from the Challenge Lab list.
On the agreed follow-up date
Follow up and verify
- Send or copy the participant’s pulse link and ask whether the commitment is Done, Moving, Stuck, or Paused.
- Offer help when someone is Stuck; do not treat it as failure or silently mark it complete.
- For Done commitments, check that the submitted evidence matches the action and date.
- Tick verified only after that check. The dashboard’s north-star count updates from verified completed commitments.
- When moving to a genuinely new event, archive the current session and start fresh. The old history remains available and read-only.
On Purpose
Match across the whole event
New On Purpose events match up to 250 confirmed participants in one cohort. Choose Use all eligible participants, prepare pairs and triads, then draft wording in batches. Those batches preserve the chosen groups and never limit who can meet.
Review every card before approving and revealing the event. Progress survives a refresh, and each model request is deliberate. Before reveal, Refresh cohort includes late arrivals and invalidates private drafts so matching can consider everyone again. After reveal, conversation membership is locked.
Event reviewers receive access to the full cohort only after the event lead assigns them. Existing events created with the earlier template keep their saved room-based workflow; create a new event to use event-wide matching.
Verification boundary
The local 250-person rehearsal uses fictional participants and simulated AI. Live matching quality, cost, and latency still require an authorized pilot; deployment is a separate step.
Optional comparison
Test the models without risking a live room
Model tests belong in a new private test room. Use the same questions and participant room when you want a fair comparison, and record your judgment before polishing the wording.
Lower-cost, deeper reasoning
Useful for larger synthetic rooms and a comparison against the stronger Terra model.
Recommended baseline
A balance of interpretation and speed for assisted matching tests.
Fast Terra baseline
Use to compare reasoning against the original pilot matcher behaviour.
- In Pilot settings, select the preset and save the session configuration.
- Open Test room, choose New private test room, and generate 6–12 synthetic people—or load the fixed sample.
- Run AI matching. Before editing, rate each suggestion Useful, Mixed, or Poor and note why.
- Review the model and reasoning label on each match and the AI run history in Pilot settings.
- For another model, start a new private room and repeat. Never use Reset on a room whose evaluation evidence you want to keep.
Keep beside you
Live-session checklists
Before
During formation
After
When something is unclear
Troubleshooting
I cannot open the tutorial.
Sign in with your personal facilitator code. If you do not have one, ask Nadia or another administrator to create or reset it.
No participants appear in a test room.
Choose Load fixed sample for repeatable data, or Generate with AI for fictional answers to your current questions. AI generation requires an empty test room.
AI matching is unavailable or times out.
Use fast topic-first matching. It remains available without AI and never deletes approved groups. A failed AI request also leaves the existing draft intact.
A suggestion is plausible but not useful.
Edit or remove it. Use Second pass to consider alternatives. Shared interest alone is not enough; look for a useful question and a credible contribution each person can make.
A participant does not want the introduction.
Respect Pass or Not now without requiring an explanation. Do not create the joint introduction or share contact details unless everyone agrees.
I changed computer or want to use my phone.
Open the Facilitator desk on the trusted device and use the same personal code. Each device receives its own session; signing out one device does not sign out the others.
I am an administrator.
Administrator is an additional privilege on your facilitator account. Open Settings to invite people, reset personal codes, rotate the shared administrator key, correct a session’s lead facilitator, manage protected exports, disable access, or sign someone out on every device.
Can I open another facilitator’s session?
Yes. Choose it from Session history. You can review participants, matches, commitments, settings, and evidence, but only the lead facilitator can change that session.
A session is linked to the wrong facilitator.
An administrator should open Settings, find Session ownership, select the correct lead, and confirm the change. The change is recorded, restores lead-only writing, and updates the facilitator name used in future participant messages.
How do I invite a project that missed the first event?
Open Follow-on support in the Facilitator desk and copy the generic Challenge Lab link. It requires no account. New applications appear in the same tab, where the session lead can add them to the Challenge Lab list, set them aside, or return them to the candidate pool without deleting them.
How do I preserve or reuse a session?
Open Settings in the Facilitator desk. Download a versioned session package for recovery evidence, a CSV for analysis, or choose Duplicate as private test. A private copy keeps room-plan, approval, reveal and question history, but receives new capability links and resets contact choices because the original consent does not authorize disclosure in a different room. It never overwrites live data, sends messages, or calls AI.
How does an administrator back up the whole system?
Open Settings and use Data exports, backup, and safe restore. The versioned JSON backup includes room-plan and consent evidence plus deletion safeguards, while excluding sign-in codes, administrator keys, device sessions, capability links, secrets and external credentials. Store downloads securely because participant contact information may be included.