AI

AI for business, not for show

Edda Network is an AI company. As the team behind the legal assistant Texir AI and the enterprise workspace Zerolog Business, we apply the same architecture to your organization’s data.

Live example · Texir AI

Every claim tied to a numbered source

Texir AI searches Turkish legislation and case law and ties every sentence in its answer to a source. Calculations such as deadlines and interest are done by code, not by the language model. We apply the same principle to our enterprise assistant projects: verifiable answers.

  • Source-citing search over your organization’s documents (RAG)
  • Deterministic tool calls for critical calculations
  • Reading scanned documents with Turkish OCR
Explore Texir AI
app.texirai.com
What is the limitation period for a severance pay claim? Termination 14.03.2024.
The limitation period for severance pay claims is 5 years1. It starts running from the date the employment contract is terminated2.
Calculated deadline14.03.2029
1Labour Law No. 4857, Additional Art. 32Court of Cassation precedent — start of limitation

Live example · Zerolog Business

From conversation to decision, from decision to action

The assistant in Zerolog Business summarizes what happened while you were away and extracts action items from meeting notes and recordings. AI suggests; people approve.

  • Conversation and meeting summaries
  • Automatic action items and tasks from notes
  • Decision log and approval flow
Explore Zerolog Business
business.zerolog.tech
# production

While you were away

  • • Line 2 maintenance moved to Thursday
  • • Quality report awaiting your approval
  • • 2 new tasks assigned to you
A

Ayşe K. 09:42

Updated the maintenance plan, sending it for approval.

Decision: Line 2 maintenance plan · pending approval
M

Mehmet D. 09:47

Approved 👍 Meeting at 14:00.

Capabilities

What we build for your organization

We use the building blocks behind our ready-made products in solutions tailored to your processes.

Source-Citing Assistants

A RAG architecture that searches your internal documents, regulations and databases, and links every answer to its source with a numbered citation.

Document Intelligence & OCR

Extract text, tables and fields from PDFs, Word files, scanned papers and forms with Turkish OCR; classify and summarize them.

AI Agents

Agents that don’t just answer but get work done: they open records, send items for approval, generate reports and move data between systems.

Summaries & Action Items

Summaries, decisions and to-do lists generated from meetings, correspondence and notes — live in Zerolog Business.

Deterministic Calculation Layer

Tool calls verified in code, so critical calculations like deadlines, interest and costs are never left to the language model.

Predictive Analytics

Anomaly detection, predictive maintenance and demand forecasting on IoT sensor data and ERP records.

Conversational Interfaces

Assistants that speak natural Turkish, embedded in web, mobile, kiosk and business chat channels.

Integration & API

A secure API and webhook layer that connects to your ERP, CRM, UYAP (Turkish e-judiciary), email and internal systems.

Use Cases

AI at the heart of the work

Texir AI

Legal research

Legislation and case law search, deadline/interest calculations, petition drafts.

Zerolog Business

Team communication

Catch-up summaries of what you missed, action lists from meeting notes.

Document Tracking

Insurance files

Document extraction, classification and deadline tracking for claim files.

Edda IoT

Production & quality

Anomaly detection in sensor data and predictive maintenance alerts.

E-CURE ERP

ERP assistant

Ask questions and get reports on inventory, order and production data in plain language.

Smart kiosk

A self-service assistant that talks and guides visitors at kiosks and queue systems.

Our Principles

Four rules of trustworthy AI

Show the source

An answer that can’t be verified isn’t an answer. Our products tie every claim to its source.

Leave the math to code

No guesswork where precision matters; calculations are done with deterministic code.

Your data is yours

Hosting in Turkey and per-account isolation; models are never trained on customer data.

Human in the loop

AI suggests, people decide. Approval flows and audit logs come standard.

How We Work

Four steps from discovery to go-live

A measurable pilot first, then scale. We don’t start AI projects that won’t deliver value.

  1. 01

    Discovery

    Together we define which work will improve, with which data, and by which measure.

  2. 02

    Pilot

    We build a small, measurable pilot on your real data and measure accuracy against an evaluation set.

  3. 03

    Integration

    We connect to your ERP, document archive, communication and IoT systems, and set up permission and approval flows.

  4. 04

    Go-live & Monitoring

    We go live, continuously monitor quality and cost, and keep the model and sources up to date.

Frequently Asked Questions

About AI projects

Can you build a custom AI assistant for our organization?

Yes. We build the same source-citing architecture we use in Texir AI on top of your organization’s documents, procedures and databases. The assistant answers with citations to your own sources.

Is our data used to train AI models?

No. We work under enterprise contract terms that exclude customer data from model training; data is isolated per account/organization and can be hosted in Turkey when required.

What happens if the AI gives wrong information?

We reduce this risk by design: answers are tied to sources, critical calculations are done in code, and people stay in control of key decisions. We also measure quality regularly with evaluation sets.

Will it integrate with our existing ERP and systems?

Yes. We build assistants and agents that work with E-CURE ERP, finance/accounting software, our IoT devices and any system that supports webhooks/APIs.

How long does deployment take?

Our off-the-shelf products (Texir AI, Zerolog Business) can be used the same day. For custom projects we move through discovery, pilot and go-live phases; we agree on the timeline together during the discovery meeting.

Want an AI assistant tailored to your data?

Tell us about your processes and data; we’ll start with a measurable pilot and scale it across your organization.