retraceTestFlight
XRCC BERLIN 2026Live on TestFlight today

AI copilot for field technicians.

Record one expert demo. Coach every junior technician forever — real-time voice + visual coaching through their smart glasses or phone. The AI watches the work and verifies each step before it advances.

REC · 00:07
Expert · iPhone capture720 × 1280
Extracted SOPGemini 2.5 Pro
Replace hydraulic manifold seal
  1. STEP 01
    Isolate line pressure
    00:00
  2. STEP 02
    Loosen housing bolts
    00:01
  3. STEP 03
    Lift the manifold cover
    00:02
  4. STEP 04
    Inspect the old seal
    00:03
  5. STEP 05
    Seat the new seal
    00:04
  6. STEP 06
    Reassemble and pressure-test
    00:06
ADVANCE_STEP · visualGemini 3.1 Flash Live

75% FTFR INDUSTRY BASELINE·+1PP UPLIFT AT FLEET SCALE = MILLIONS / YR·$200–300 BURNED PER REPEAT TRUCK ROLL

The problem

Your senior technician walks out the door — and decades of judgment walks out with him.

Half of process knowledge is held in people's heads. Roughly 42%McKinsey is tribal — undocumented, lost the day they leave. The 600-page service manual that's supposed to replace them is exhaustive and unreadable in the moment.

Junior technicians inherit it anyway. About 25%Aberdeen of failed dispatches trace directly to a training gap — and the average repeat truck roll costs $200–300Industry avg in fuel and time before you count the customer's downtime.

Hiring more people would be the obvious answer, except the people aren't there. The US manufacturing workforce is 44.8BLS years old on average — the oldest it has ever been — and faces 3.8MDeloitte / MI jobs by 2033 with roughly half unfilled. Germany alone is short 5MIW Köln skilled workers by 2030.

One recording. Three modes. One architecture.

01 / CAPTURE

One expert.
One narrated demo.

Wear the glasses or hold up the iPhone. Talk through the task once. Gemini 2.5 Pro structures it into ordered steps with per-step completion criteria. ffmpeg clips each step automatically.

Gemini 2.5 Pro · structured output
REC · 00:07
Extracted SOP
Replace hydraulic manifold seal
  1. 01Isolate line pressure
  2. 02Loosen housing bolts
  3. 03Lift the manifold cover
  4. 04Inspect the old seal
  5. 05Seat the new seal
✓ completion criteria
LIVE · STEP 02 / 06
~0.5 FPS · LOW
PiP · reference clip
Activity
Loosen the housing bolts in diagonal pairs — two passes.
Like this?
Yes. Once they're each backed out one full turn, you're good to advance.
advance_step({ from: 2, to: 3, method: 'visual' })
02 / COACH

Voice replies.
Vision verifies.

Phone or glasses hold a direct WebSocket to Gemini 3.1 Live — voice and camera at ~0.5 fps. The model watches for the completion criterion and fires advance_step the moment it sees it. No server hop.

<100 ms · indefinite sessions
03 / TROUBLESHOOT

No procedure?
Diagnose from scratch.

Describe the broken machine. The AI identifies the product, searches your library first, falls back to a web-grounded fix with cited sources, then hands off to a normal coaching session.

Library search → web-grounded → handoff
Describe the broken machine
Machine throws E-04. Compressor short-cycling.
Looks like a refrigerant-side issue. What's on the nameplate?
Carrier 30RAP-080. About 8 years old.
Got it. Searching your library, then the web.
Diagnostic phase
  1. Identify3-phase chiller
  2. DiagnoseLow refrigerant
  3. Fetch3 sources cited
  4. CoachHandoff to learner
Cited sources
  • carrier.com / service / 30RAP / E-04
  • service.refrigeration / short-cycle-diagnosis
  • fieldnotes.io / topics / low-charge-symptoms

One wedge.
Pick the pain you own.

Service ADVISOR and SIS lock diagnostics in the shop. Retrace puts them on the technician's face.

Real-time visual guidance through smart glasses, on the equipment, no swivel-chair. Embeds into the OEM's after-sales motion as a branded service layer — the technician's hands stay on the machine while the model watches.

Why now

Three unlocks landed in the same year.

Gemini 3.1 Flash Live ships real-time video + audio + tools + context compression + session resumption — the model can finally watch and coach indefinitely. Ray-Ban Meta sold 7M units in 2025, +210% YoY; the capture layer is already on technicians' faces. Meta's DAT SDK 0.6 shipped — third-party apps can finally stream from the glasses.

Hardware mature. Model mature. Distribution mature. Nobody had built the field-service coaching layer yet. Until now.

The longer arc

AI copilot for technicians today.
Training corpus for humanoids tomorrow.

Robotics labs aren't compute-starved — they're data-starved. The humanoid wave needs first-person video of humans doing useful work, paired with speech and task-completion ground truth. That's scarce. Most of what exists is consumer or household.

The public benchmarks tell the story. 3,670 hrsEgo4D · Meta '22 of egocentric video across 9 countries. 1,422 hrsEgo-Exo4D · Meta '23 of paired first-person + third-person footage. 76K demosDROID · Stanford '24 of robot manipulation across 86 tasks. 1M+ episodesOpen X-Embodiment · DeepMind '23 spanning 22 different embodiments. All landmark releases. None indexed by industrial procedure, none paired with expert narration, none capturing the long tail of skilled-trade tasks that pay the labs' bills.

The data-starved labs · cumulative funding
Figure
$675M Series B
$2.6B (Feb '24)
OpenAI partnership
1X
$100M Series B
NVIDIA-backed
NEO Beta
Tesla Optimus
Mass-production goal
Public co.
Trained on egocentric video
Physical Intelligence
$400M
$2.4B (Nov '24)
π0 foundation model
Skild AI
$300M
$1.5B (Jul '24)
General-purpose robot brain

Combined Series-B+ funding across the five frontier humanoid programs: over $1.4B raised in 18 months on the promise of robots that can do useful physical work. The bottleneck is the same in every deck.

What Retrace's data is, that nobody else's is

Five attributes the public corpora don't have together.

  1. 01
    First-person POV.

    Same camera the humanoids will use. Ego4D had to convince 855 strangers to wear cameras around — Retrace's data comes from technicians whose existing capture habit is on-glasses or in-phone.

  2. 02
    Synchronized speech.

    Narrated demos with timestamped transcripts. Most egocentric corpora are silent video — narration grounds intent and procedure.

  3. 03
    Tool-call ground-truth.

    Explicit advance_step segmentation on completion criteria — explicit task boundaries are rare in public datasets and gold for downstream RL.

  4. 04
    Paired attempts.

    Expert demo + every learner attempt of the same task. Positive and negative samples paired by procedure — the kind of contrastive data RT-2 and π0 wish they had.

  5. 05
    Industrial verticals.

    Field service, OEM equipment, skilled trades — exactly where Ego4D leans thin. The work humanoids will eventually be sold into.

Same data, two markets. AI copilot today. Training corpus tomorrow.

Two builders

Full-stack, end-to-end.

Backend / AI + iOS / XR. One real shipped product in four weeks.

  • H
    Hyunseok Hwang
    Backend · AI systems

    FastAPI server, Gemini orchestration (2.5 Pro + 3.1 Live), structured-output extraction, ephemeral tokens, troubleshoot mode, web-grounded search.

  • J
    Jayden DeCambre
    iOS · XR

    SwiftUI app, Meta DAT SDK 0.6 integration, Gemini Live WebSocket client, Ray-Ban HUD design system, MediaPipe hand tracking.

Questions we get

Six questions, six straight answers.

Do I need smart glasses?
No. Retrace runs on iPhone today — the phone holds the direct Gemini Live session, audio and camera. Ray-Ban Meta is an optional capture layer for hands-free work. Glasses and phone are interchangeable per session.
What about the privacy of recorded video?
Expert recordings stay in your tenant. Voice and video stream peer-to-peer between the technician's phone and Gemini Live — the recording itself is only retained server-side when you publish it as a procedure. Ephemeral tokens scope every Live session to a single voice, tool set, and resumption handle.
Who owns the data?
You do. Procedures, clips, and session telemetry are yours. Egocentric training data is opt-in per-procedure, with named export controls and a per-fleet kill switch.
How is this different from Aquant or Neuron7?
Aquant and Neuron7 are powerful service-intelligence tools you open on a phone or laptop after stopping the work. Retrace runs continuously through the technician's glasses or phone camera while they work — visual auto-advance, voice replies, no swivel-chair.
How is this different from John Deere Service ADVISOR or Cat SIS?
Those are office and shop tools — searchable manuals, diagnostic codes, calibrations. They don't deliver real-time visual guidance on the equipment. Retrace is the layer on top: the technician's hands stay on the machine, the model watches, the procedure auto-advances when the completion criterion is visible.
Deployment model?
Hackathon-stage today. iOS app on TestFlight, backend on AWS Lightsail (migrating to eu-central-1 before Berlin). Production deployment is a self-hosted FastAPI + your Gemini API tenant — no third-party data egress.