0
Skip to content
Intaleon
Jarlai
Team
Contact
English
Intaleon
Jarlai
Team
Contact
English
Jarlai
Team
Contact
English
Back

JARLAI

The Company Brain

From meeting minutes
to organizational coherence layer.

Jarlai captures every meeting locally on an appliance within your premises. The Coherence Engine cross-references what has been said across meetings and flags any discrepancies.

Coherence Engine in active development Jarlai Meeting AI deployed today
Discuss use cases
100% Offline Edge AIZero CloudPlug & Play

The Coherence Engine

The graph thinks along.

Every meeting feeds statements into a typed knowledge graph. It performs cross-comparisons across meetings, teams, and weeks. As soon as statements collide or chain together into a risk, a finding is generated complete with quote, source, and timestamp.

The knowledge graph at work Sketch with sample data
Every meeting feeds the graph
Live transcription Sketch with sample data
AK Anna K. · Project Management
TG Thomas G. · Product Development
What Jarlai makes of it
Summary · automated
Go-live November 30, pending external testers from October.Decision
Plan sign-off for early December · Anna K.Task
Ask Jarlai
What was decided regarding the go-live?
Source: Project Status · Aug 26 · 00:31:55

The Foundation: Jarlai Meeting AI

Every word captured. Local. Live.

The Meeting AI transcribes physical and hybrid meetings in real time, directly on the device. Decisions, tasks, and statements are extracted from the conversation, fully verifiable against the original text at any time. These exact statements feed into the knowledge graph above.

Plug & Play100% LocalAll on-device

Set up, plug in, and productive from day one. Speech recognition and the language model run entirely on the appliance, with no external interfaces in the processing pipeline.

Finding Types

What Jarlai detects.

Contradiction

Two meetings establish conflicting delivery dates for the same package.

Detected before ordering, not upon delivery.

Risk Chain

A budget freeze cuts the testing phase, putting the go-live and customer commitment at risk.

Each statement is reasonable on its own. The chain of events creates the risk.

Context

A decision in Project A quietly drives a work package in Project B.

Makes dependencies visible before they cause surprises.

Unresolved Task

Something was promised, but lacks an owner, a deadline, and tracking.

Closes the gap before it becomes a liability.

Decision never made

Everyone acts as if a decision has been made, yet it was never actually finalized in any meeting.

Prevents work based on flawed foundations.

Diverging conditions

Different prices, discounts, or payment terms are communicated to the same counterparty.

Protects margins and credibility during negotiations.

Note: The Coherence Engine is an extension of the Jarlai platform currently under development. Availability and pricing are determined as part of a pilot program with reference customers.

How it works

Five steps. A reliable memory.

01
Capture
Meetings are transcribed and summarized locally on the device. Audio and text never leave your premises.
02
Extract
Every meeting yields typed statements, classified as decided, proposed, or agreed, anchored directly to the wording.
03
Resolve
Entities are linked to your registries: project codes, equipment tags, vendors, contracts.
04
Compare
Statements are cross-referenced across meetings and, upon request, against your own documentation.
05
Report
Findings are weighted by severity and confidence: critical issues trigger immediate alerts, while all others are compiled into a unified overview.
The graph thinks

The language model perceives, the graph thinks. Comparisons are performed using deterministic graph queries rather than freely generated responses: the same query on the same data always yields the exact same result. No hallucinations.

Role-based access

What is displayed as a finding depends on a user's role and their authorized access permissions. Non-authorized statements do not generate findings for that role.

Pricing & Appliance

One device. One flat rate. No surprises.

The Jarlai Appliance

A dedicated on-premise device: speech recognition, language model, and knowledge graph run entirely on the appliance. The companion Jarlai app on your devices handles recording and display. Processing is performed exclusively local to the device.

  • Basic, Pro, and Ultra editions, depending on meeting and processing volume
  • Air-gap operation possible, including offline updates
  • Hardware remains the property of Intaleon; the system scales as needed
Hardware-as-a-Service
B2B Flat RateNo Token LimitNo User LimitUnlimited UsageHardware Included
from €599 / month Jarlai Meeting AI incl. hardware · 20% discount on 3-year contracts

You pay for capacity and features, not usage. We will determine which edition fits your volume in a brief consultation.

Request a quote

Data Sovereignty

Hardware-enforced. Not just promised.

Local Execution
The model and data reside locally on the device deployed at your premises. All processing takes place directly on-site within the enterprise.
No Cloud, No Transmission
No external interface, no vendor account, no subprocessor in the processing chain.
Unrestricted Ownership
All transcripts, insights, and summaries remain entirely in your possession, with no access by Intaleon or third parties.

Frequently Asked Questions

Before you have to ask.

Is the Coherence Engine already available?

It is currently in active development. We can already demonstrate two concrete use cases in a personal consultation. Availability and pricing will be determined as part of a pilot program with reference customers.

What happens to the hardware at the end of the contract?

Jarlai runs as a Hardware-as-a-Service solution. The appliance remains the property of Intaleon at all times and is returned at the end of the contract, much like a router from an internet service provider.

Does Jarlai also work completely offline?

Yes. Full air-gap operation is supported, including offline update packages for maintenance.

Are my data automatically backed up?

Currently, automated data backup is not included. We address this transparently in your tailored proposal so that you can factor it into your own backup strategy.

How quickly is Jarlai operational?

Plug & Play: position it, connect it to power, and be productive from day one without the need for an integration project.

Bring two meetings regarding the same project.

We will demonstrate in a conversation what a coherence layer reveals beyond that.

Schedule a consultation

100% local AI · Zero Cloud · Plug and Play

Imprint

Privacy Policy

Terms & Conditions