Skip to content
BelieverTech

Work

Four systems, described by what they do

No client logos and no invented metrics. Every capability listed here maps to code that exists and runs.

Tiered inference — one GPU, four tiers

Tier 0   every frame      YOLOWorld + ByteTrack            ~30–50 ms
Tier 1   triggered        plate OCR · re-ID · Florence-2   on detection
Tier 2   gated            live VLM narration               on scene change
Tier 3   on demand        heavy VLM deep-look · summary    operator request

Computer vision platform

2026

Real Vision

Cameras that understand what they see. Real-time detection, multi-camera identity tracking, licence-plate recognition and vehicle damage baselining — narrated in language, running entirely on the client's own hardware.

  • ~347

    detection classes

  • 4

    model tiers, one GPU

  • 7

    safety event kinds

  • 0

    footage leaves site

Per-turn pipeline — identical on every channel

audio    →  STT          browser SR, or server-side Whisper
         →  language     script + romanised marker detection
         →  context      retrieval · booking state · scoped search
         →  LLM          local model, streamed deltas
         →  TTS          sentence-assembled, barge-in cancellable
         →  caller

Voice AI platform

2026

AI Caller

A multi-tenant AI receptionist that answers the phone, the browser and web chat — with real barge-in, booking into live calendars, and the language model running on infrastructure you control.

  • 3

    channels, one pipeline

  • 4

    interface languages

  • 20 ms

    voice detection frame

  • 0

    cloud LLM APIs required

Start here

Find out what it is costing you now.

Every engagement starts with a systems audit. We map one workflow, put a number on what it currently costs, and tell you what it would take to fix — including whether it is worth fixing.

Start a projectSee the work first