What we need from the university / how we connect¶
Good news: ControlAI works with what the university already has. In a typical university, cameras and a video recorder have long been installed — we connect to the existing infrastructure, without installing anything in classrooms. Below is a practical checklist you can hand straight to the university's IT department; every item is quick to verify. Two points of contact on the university's side are enough: the IT department (cameras, network) and the academic affairs office (schedule).
What the university needs to have¶
Cameras in classrooms¶
Classrooms already have cameras installed — ControlAI uses them. What matters is that the camera:
- shows the classroom (students should be visible — useful for Stage 2 visual metrics);
- records audio — all speech analytics runs on audio;
- delivers acceptable image quality (modest quality is enough for analysis — no need to chase 4K).
Audio comes first¶
Speech analytics is only as good as how clearly a class can be heard. That's why audio quality matters more than video quality. If a camera's built-in microphone is weak, the classroom gets a single inexpensive external microphone plugged straight into the camera's audio input — and only where the built-in mic fails the audio test. That's enough. Separate microphones for the instructor or for each student are not needed.
A note on large lecture halls: monitoring whether a class was held, timing, and the lecturer's talk share all work in any classroom; we confirm full transcript coverage in a large hall with an audio audit during onboarding — and, if needed, solve it with the same external microphone.
Video recorder (where recordings are stored)¶
Cameras typically record to the university's video recorder (NVR/DVR). We need:
- recordings to be stored for at least a few days (until we retrieve them);
- our system to have access to them;
- the recorder's clock to be accurate (network time sync — so we pull exactly the right segment of the class).
Network access¶
For ControlAI to retrieve recordings on its own, the university needs to provide internet access to its recorder — we set this up together with the IT department, securely: access is opened only for our fixed address (an allowlist), not "the whole internet." This requires a permanent (static) external IP address — universities usually already have one; if not, it's an inexpensive service from the internet provider. Plus enough internet speed for recordings to upload in time (if the connection is weak, we pull overnight, so reports are ready by morning).
Schedule¶
We need to know when and in which classroom classes take place — the schedule tells the system which recordings to pull and when. The semester schedule is entered into the panel or imported as a file (an export from HEMIS works; direct HEMIS integration is Stage 3, see the Roadmap).
Consents¶
Since classes are recorded (video and audio), we need instructors to give consent and students to be informed. Students are adults, so the consent process is simpler: transparently explaining what's recorded and why is enough. This is part of onboarding — more in Privacy and trust.
What we check before launch (audit)¶
Before launch we run a short audit of every classroom and the university as a whole, together with the IT department:
| We check | Why |
|---|---|
| Audio quality from the camera | whether it's good enough for speech recognition (a separate audio audit for lecture halls) |
| Recorder accessibility | whether we can retrieve recordings |
| Static external IP address and allowlist | so only our system can connect to the recorder |
| Recording retention period | whether we can pull recordings before they're overwritten |
| Internet speed | how quickly reports will be ready |
| Recorder clock accuracy | so we pull exactly the right segment of the class |
If a classroom "doesn't quite make it"¶
If audio is weak in some classroom, or something is missing — that's not a dealbreaker. Usually an inexpensive external microphone on the camera's audio input, or a settings tweak, is enough. These small fixes are handled during onboarding, before launch.
Key takeaway from this chapter: the university doesn't need to provide much — working cameras with audio, an accessible recorder, secure network access, a schedule, and consents; nothing new is installed in classrooms. Audio quality matters more than video; where it falls short, a single inexpensive microphone solves it. All of this is checked in advance, during the audit with the IT department.
→ Next: 13. How we roll it out — the path from "agreed" to "up and running."