Skip to content

Category: Ember AI · Foundation Models

One model family, owned end to end — powering every Coreidea platform.

Ember is Coreidea’s in-house AI model family: Reason for deep analysis, Core for everyday work, Vision for multimodal, Lingua for translation, Guard for sensitivity and redaction, Edge for on-device inference, and Embed for retrieval — all served behind one private gateway with evaluation gating and residency-aware routing.

Built for

  • Coreidea platforms
  • Operating teams
  • Invited partners
Status
Live
Ledger step
Ledger step 4 of 4
Family
Ember family
Access
Public product site
Listed
4 / 4
Specialized models
7
Private gateway
1
Calls audited
100%
Third-party SDKs in prod
0

What this page can show today

  • The register marks this track Live.
  • A public product URL is published on this page.
  • Capabilities below are the published feature list for this track.

Limits of this page

  • This page does not publish financials, headcount, or client names.

Key features

  1. 01

    A full model family

    Ember Reason for deep reasoning, Core for everyday work, Vision for multimodal, Lingua for translation, Guard for sensitivity and redaction, Edge for on-device, Embed for retrieval.

  2. 02

    One private gateway

    Every model call across the group is brokered by a single gateway. Keys, rate limits, fallbacks and audit live in one place.

  3. 03

    Evaluation-gated releases

    Model and prompt changes ship only after passing task-level evaluation suites. Quality is measured, not assumed.

  4. 04

    Residency-aware routing

    Chinese personal and business data is served by in-jurisdiction Ember deployments; export workloads route through compliant channels.

Published links

Technology stack

Python · FastAPI · Apple Silicon · PostgreSQL

Related platforms

Family on the register: Ember family Same family first, then tracks at the same lifecycle stage.

  • Same family · Live

    Ember Legal

    Legal Technology

    Invite only

  • Same family · Beta

    Ember Live

    Real-Time Model Surface

    Invite only

  • Same family · Beta

    Ember Desktop

    AI Work Operating System

    This site only

  • Same family · Beta

    Ember Agent OS

    Multi-Agent Operations

    This site only

This cell on the investor register

Status and family write into the /investors URL. The holdings table lists every public track in that cell.

Open Live Ember family holdings
4 / 4

JSON fields are copied from this row in src/data/platforms.ts. No financials.

How this track works

Audience on the left, the published Ember stack in the middle, capabilities on the right. Every node is copied from this track in the public register.

How this track moves work from its published audience through Coreidea and Ember into listed capabilities.

Inputs

Coreidea and Ember

This track does not name an Ember model on its published stack.

Outputs

Select a node to inspect the published role and capability.

Renting intelligence means renting your roadmap.

Companies that bolt third-party AI onto sensitive workflows inherit foreign terms of service, unpredictable pricing, opaque model changes and data-residency exposure.

  • Model churn breaks products

    Vendor model updates silently change behavior under your features. Quality regressions arrive unannounced.

  • Residency is non-negotiable

    Chinese personal and business data must stay in-jurisdiction. Generic cloud AI offers no credible answer.

  • No audit, no defense

    When a regulator or client asks “what did the AI do and why”, most stacks have nothing to show.

Owned models, measured quality, one accountable gateway.

Ember exists so Coreidea platforms never depend on a third party for their core intelligence — and so every AI action is attributable, evaluated and residency-correct.

  • A family, not a monolith

    Seven specialized models do what one generalist cannot: the right capability, latency and cost profile per task.

  • Evaluation-gated releases

    No model or prompt change ships without passing task-level eval suites. Quality is a number, not a feeling.

  • Residency-aware by design

    The gateway routes by data sensitivity and jurisdiction: in-China data to in-China deployments, on-device when it never needs to leave.

  • Audited to the call

    Every invocation — model, version, policy, redaction state — lands in an audit trail. AI actions are attributable by default.

How a request moves through Ember.

  1. Classify

    Ember Guard scores sensitivity and strips identifiers where policy requires.

  2. Route

    The gateway selects the model and venue: Edge on-device, or Reason / Core behind the private boundary.

  3. Execute

    Structured outputs with schema validation; tool calls are explicit and auditable.

  4. Verify

    Task-level checks gate the result before it reaches product surfaces.

  5. Record

    The full decision context is appended to the audit trail for replay.

The Ember model family.

  • Ember Reason & Core

    Deep analysis and everyday generation — the cloud pair behind the private gateway.

  • Ember Vision & Lingua

    Multimodal understanding and regulated-grade translation across the group’s platforms.

  • Ember Guard & Embed

    Sensitivity classification, deterministic redaction and retrieval embeddings as shared services.

  • Ember Edge

    The on-device runtime for Apple Silicon — offline-first inference where data must never leave.

Where Ember is already at work.

  • SkillSwipe

    Challenge generation and assessment assistance with audit-grade attribution.

  • CarbonBridge

    Regulatory NLP and supplier-data extraction with in-jurisdiction processing.

  • LabForge

    Lingua drafts under locked terminology; Guard enforces passthrough lists.

  • CoreAI & Apertura

    Sensitivity-aware routing in CoreAI; on-device intelligence in Apertura via Edge.

Questions, answered.

Is Ember available to external customers?

Ember primarily powers Coreidea platforms. Invited partners can access it through CoreAI; broader availability follows the invite program.

Why build models in-house?

Control over quality, cost, residency and roadmap. Our platforms make compliance promises that rented intelligence cannot keep.

How is quality maintained?

Every change passes task-level evaluation suites before release, and production behavior is continuously sampled against the same benchmarks.

What about data residency?

Routing policy is jurisdiction-aware: in-China data is served by in-China deployments, and high-sensitivity content runs on-device via Ember Edge.

Intelligence you own is intelligence you can promise.

Talk to us about the Ember family, the gateway architecture, or partner access through CoreAI.

Patent filings bound to this track

Only applications whose public id is this track id or that id plus a suffix. Empty means this page does not invent a hop.

All platforms