Branch B, Sequenced

The first 90 days

The base case is an AI study companion for enrolled students, on campus and blended. The plan below gets a working version running on a real SFU module inside four weeks, then measures it against the benchmark Oakley already trusts: the 27% study-time reduction proven at IU with Syntea. A working result on SFU's own content arrives before any incumbent platform finishes a scoping document.

Three horizons

Days 0-30Prove it

Prove it on real content

Build a working AI study-companion demo on ONE real Psychotherapy Science module, grounded in SFU's own material. In parallel, run the AI-visibility audit across the four faculty lines to baseline where SFU is and is not cited today.

Exit: a demo Kevin can use, on SFU content, not a slide.

Days 31-60Measure

Measure and extend

Stand up the outcome harness and read it against the IU 27% benchmark. Extend the companion to a second faculty. Run a faculty teaching-assistant pilot so staff see load come off, not added.

Exit: a first outcome number on real students, not a projection.

Days 61-90Roll out

Roll out and rank

Blended rollout across faculties. AI-ranking and AEO live on the query clusters that matter. If the ambition is bigger than one university, open the group-platform conversation, backed now by a real result.

Exit: proven on SFU, ready to template for the group.

12 quick wins

#	Quick win	Track	By
1	Companion live on one Psychotherapy module's real content	Companion	Day 30
2	Quiz and summary generation working inside that module	Companion	Day 30
3	AI-visibility baseline across the four faculty lines	AEO	Day 30
4	Course schema shipped on the pilot faculty's program pages	Schema	Day 30
5	Direct-answer FAQ content for the psychotherapy-degree cluster	Content	Day 45
6	Outcome harness reading study-time against the IU benchmark	Companion	Day 50
7	Faculty-expert entity pages for the pilot faculty	AEO	Day 50
8	Companion extended to a second faculty	Companion	Day 60
9	Enquiry-chat deflection live on one program page	Chatbot	Day 60
10	Course schema rolled across all four faculties	Schema	Day 75
11	AEO content pass live on the priority query clusters, DE and EN	AEO	Day 80
12	Group-reusable companion and content template documented	Group	Day 90

What we measure

Metric	Baseline	Day 30	Day 60	Day 90
Study-time on covered material	Captured at start	First read	Trending down	Toward the 27% benchmark
Companion weekly active usage	0	Pilot cohort	Two faculties	Blended cohort
AI citations for key queries	Captured at start	Gaps mapped	First lifts	Cluster cited
Enquiry-chat deflection rate	0	n/a	One page live	Measured rate

All day-30, day-60, and day-90 targets are illustrative. The real numbers come from SFU's own baseline, captured in week one, and the harness reads against the IU 27% study-time reduction as the benchmark to beat.

Honesty: what the week-4 demo actually uses

4-label calibration

The four-label calibration, applied to this plan so nothing on it is vaporware:

Landed Scaffolded Tutor chat and retrieval grounded in course material. This is what the week-4 demo is built from.

Scaffolded Quiz and summary generation. Core structure exists; activated on the pilot module.

Not Built Course generation on the fly. Not on this plan, not promised, and not needed for the companion.

Designed Multilingual parity, German and English. To be verified before any DE and EN promise; the pilot starts in one language.