KOR'K Sensor

KOR'K Sensor 09 What comes next

Chapter 09

What comes next

The prototype is only one step. KOR'K is moving from Scan towards the Sensor, phase by phase, with one rule: each step must de-risk the next. The dates will follow the real milestones, not the other way round.

The roadmap

From Scan to the Sensor, in six phases

3 locks to open first ↓ we are here 0 Foundations Q2–Q3 2026 1 Traction and team late 2026 2 Sensor prototype 2027 3 Launch late 2027 or 2028 4 Scaling up 2028–2029 5 Ecosystem 2029 onwards
  1. 0

    FoundationsQ2 and Q3 2026

    GoalKOR'K Scan published, then proof that parents come back to it.

    To move onThe app online and the first retention data.

  2. 1

    Traction and teamLate 2026 → early 2027 · we are here

    GoalScan grows; a technical partner, a specialist in embedded AI, joins the project.

    To move onA CTO on board and recurring revenue.

  3. 2

    Sensor prototype2027

    GoalA sensor prototype, then a beta with families.

    To move onA working prototype and a beta under way.

  4. 3

    Sensor launchLate 2027 or 2028

    GoalCertifications, tooling, first production run, pre-orders.

    To move onA first production run delivered.

  5. 4

    Scaling up2028 and 2029

    GoalA Sensor V2, with vision and expressions, and an opening towards mutual insurers (mutuelles).

  6. 5

    Ecosystem2029 and beyond

    GoalSeveral rooms, nurseries and schools, international expansion.

Source: Notion, “Roadmap — KOR'K (Scan → Sensor)”, version of 7 July 2026. Indicative roadmap; funding sources and amounts are not reproduced here.

Audio or video

The July 2026 plan envisaged a first sensor centred on audio (language, emotions). For Nuremberg, I built a video prototype, because it reuses the pipeline that already works with KOR'K Scan: the clip goes through the iPhone, and the app analyses it like any other video. The order between audio and video has not been settled for what follows; the feedback from Nuremberg will help to decide it.

What everything depends on

The three locks

  1. 1

    The company

    Creating the SAS (the French simplified joint-stock company): without it, no grants and no schemes for innovation.

  2. 2

    Scan's retention

    The proof that parents come back: it is what opens up grants, incubators, and the Sensor.

  3. 3

    A hardware or embedded CTO

    Without one, no Sensor.

Software or hardware

Let Scan set the pace

The Sensor is expensive and risky: tooling, certification, a production run. If Scan takes off, a perfectly valid option is to double down on the software first, and to launch the Sensor only once it is funded and de-risked. I am keeping the Sensor as the goal, but it is Scan's traction that will set the pace.

What Nuremberg tests

A working mock-up

What I am taking to Nuremberg is a working mock-up. It is there to test three things: the shape of the unit, its place in a room, and the trust it inspires.

It does not prove that the unit is useful to a family. That proof will require the “bedroom” mode, which has not been coded yet, followed by trials with real families.

Sources: chapter 01, “A first contact with real users”; 06-logiciel/README.md (“bedroom” mode not coded); my reply to the first critical reading of the site (10 Nov 2026).

What research says

Videos filmed at home

A US study of 51 children, aged from 18 months to almost 7, asked parents to film four ten-minute scenes: a meal, play with others, play alone, and “what worries me”. Clinicians watched them remotely. Their conclusions matched those of an in-person consultation in 88% of cases.

The authors themselves point out the limits: a small sample and wide margins of uncertainty. And the study differs from KOR'K in two ways. Clinicians watch the videos, not an AI that writes a report. And the parent frames the child with a tripod, not a fixed camera in the corner of a room. Nothing validates these two choices yet.

I found no survey of what paediatricians would like to receive from parents.

Sources: Smith et al., Psychological Assessment, 2017, 29(3), pp. 245–252, NODA study on the remote assessment of autism (article read on 10 Nov 2026); survey of 10 Nov 2026 (recherches/2026-11-10_reperes-developpement-et-parcours-medical.md).

After Nuremberg

The next versions of the unit

  • A V2, corrected after the image tests with the real unit.
  • The “bedroom” mode: recording in 30 s slices, sorted on the unit, with framing that follows the child.
  • A custom board built around a Compute Module 5, to get back towards 10 cm along the walls.
  • An optional second unit for large rooms, if visitors see a use for it.
  • In time, analysis inside the unit, so that no video leaves the home.

Sources: design journal, steps 7, 10 and 16; decisions register.

Coming soon

What Nuremberg taught me

To be completed after 11 December, with feedback from visitors and the jury.