How we roll it out (onboarding)¶
Onboarding a university follows a clear eight-step path — from "agreed" to "up and running." Nothing about the academic process needs to change: ControlAI fits into what's already there — cameras are already mounted, the video recorder is already installed, the class schedule is already set.
Step by step¶
1. Audit¶
We check the university's classrooms and infrastructure: audio quality from the cameras (with a separate check for large lecture halls), video recorder accessibility, recording retention period, internet speed, and recorder clock accuracy (details in What we need from the university). The output is a list of what needs fixing.
2. Additional setup where needed¶
Where audio is weak — mainly in large lecture halls — we add an inexpensive external microphone to the camera's audio input; where needed, we adjust camera and access settings. These are small, targeted fixes, not a refit: we install nothing new in the classrooms themselves.
3. Consents and disclosure¶
We collect instructors' consent and inform students that classes are recorded and analyzed; since students are adults, the consent process is simpler, but transparency is mandatory (details in Privacy and trust). For instructors, ControlAI is a personal assistant: personal statistics, an automatic summary of every class, and fair evaluation instead of subjective spot checks — that's how it should be introduced to the team.
4. Setting up the university in the system¶
We enter the university's structure into the panel: faculties, classrooms and cameras, degree programs and years of study, groups (with their language of instruction — Uzbek, Russian, or mixed; reports arrive in the group's language), instructors, and, most importantly, the semester class schedule — imported from a HEMIS export or entered manually in the panel. It's the schedule that switches on automatic recording retrieval. At this stage, panel invitations go out — to deans, the quality-control office, heads of department, and instructors.
5. Configuring the methodology¶
Together with the quality-control office (ichki ta'lim sifati nazorati bo'limi), we upload the university's own lesson observation form, its curricula and syllabi, and set the review priorities. From this point, the "AI-nazoratchi" evaluates every class against this specific university's own standards — like an open (demonstration) lesson, except not once or twice a year, but every day.
6. Trial run: one classroom¶
We launch one classroom in full — the first scheduled classes, from recording to a finished report — and review the live results together with the rectorate and the quality office. This resolves questions and shows that everything works as it should.
7. Go-live¶
We connect the remaining classrooms of the chosen faculty. From this point on, ControlAI analyzes every scheduled class and shows the results in the panel.
8. Ongoing support¶
After that, the university runs on its own, and we stay close by: we show the team — from the rectorate down to heads of department — how to use the panel, monitor (through our internal service role) that processing is running smoothly, and help with tasks from the Operations section (camera offline, microphone needed, and so on).
What determines the timeline¶
Most of the time goes into the audit and targeted equipment fixes; entering the university into the system and configuring the methodology run in parallel — so we give an exact timeline only after the audit, since it depends on the scope of the fixes. The trial run removes risk before the full launch, which is why go-live goes smoothly.
Where to start: the pilot¶
We suggest starting small — a pilot on 5–10 classrooms in one faculty, essentially the same steps at a smaller scale:
- Infrastructure audit — cameras, video recorder, audio in the selected classrooms;
- Connection and schedule upload — the faculty is entered into the system, recording switches on by schedule;
- First reports — the rectorate and quality office see, on real classes, how it works: whether each class was held, timing, speech metrics, a completed observation form.
The pilot requires no restructuring of the academic process and changes nothing for students; based on its results, the university decides whether to scale to the remaining faculties — relying on its own data rather than a presentation.
Key takeaway from this chapter: onboarding is audit → small equipment fixes → instructor consent and student disclosure → entering the university and its schedule → configuring the methodology → a one-classroom trial run → faculty launch → ongoing support. The academic process doesn't change; ControlAI fits into what's already working. The easiest place to start is a pilot on 5–10 classrooms in one faculty.
→ Next: 14. Privacy and trust — how data and consents are protected.