Enterprise RAG: AI on your organization's own data
Custom RAG implementation, connected to your phone system, finance, inventory, HR and office automation
RAG means the AI searches your organization's own data before answering, and shows the result together with its source. We implement it for organizations — on the organization's own server, in Persian, and connected to the very systems where your day-to-day data lives.
Six things that set our RAG apart from a generic one
Most RAGs only work on a few uploaded files. We can connect RAG to the organization's own systems, because Afrandeh is the very system that holds the organization's financial, inventory, HR and administrative data.
Phone-system connection & support-audio analysis
We connect to company phone systems (Asterisk, FreePBX and Issabel), transcribe recorded support and sales calls in Persian, and make them searchable together with the number, direction and time of each call.
Finance, admin and inventory data in RAG
We feed financial, administrative and inventory data from Afrandeh into the organization's RAG, so answers rest on your real data, not general knowledge. Every Afrandeh report becomes searchable text.
Job descriptions, personnel and org chart
We connect job descriptions, personnel information and the full org chart to that organization's own RAG, so you can ask who is responsible for what.
Personnel performance
Personnel performance records, kept in Afrandeh's Human Resources module, reach the same RAG and become queryable.
Procedures, forms and processes
We connect every procedure in Afrandeh's office automation, form builder and processes to RAG, so everyday questions like "how is this done?" are answered from there.
An after-sales support companion
RAG can be a good companion for after-sales service: it answers customer questions based on the real history of requests, tickets and support calls, and shows the source of every answer.
From organizational data to an answer with its source
Connect the sources
The phone system, Afrandeh reports and files (PDF, Excel, scans, video and audio) are connected to the organization's RAG and their text is extracted.
Four-stage search
Every question goes through semantic search, keyword search, result fusion and final ranking.
Results with a source
The most relevant passages are shown with their exact source and relevance.
This same technology is already running as a ready product in AfranDanesh: hybrid search, Persian transcription, separate companies and running on the organization's server. Learn about AfranDanesh
A RAG that is integrated with the organization itself
One platform, one database
Finance, inventory, HR, office automation and after-sales sit on one platform in Afrandeh, so RAG does not have to stitch several separate systems together to reach this data.
On the organization's own server
Every component of the RAG, from search to the speech models, runs on the organization's own server and your data stays on your infrastructure.
Company and role separation
Each company has an independent knowledge base, and each user searches only the companies they are allowed.
Built for Persian
Half-spaces, Arabic ی and ک, and three digit systems are treated as one, and homophone letters are corrected after call transcription.
What it shows today, and what comes next
Today RAG shows the most relevant passages of the organization's data together with their source and relevance. A local large language model that writes full, fluent answers is in progress and will also run on the organization's own server; tuning the model for each organization is planned.
Getting started & demo
Request a consultation & demo
Fill in the contact form or reach the Afrandeh team directly.
A short needs review
In a short session, we review your organization's data sources (phone system, Afrandeh, files).
Deployment
It is deployed on the organization's own server and connected to your data sources.
Enterprise RAG — frequently asked questions
What is RAG?
RAG means the AI searches your organization's own data before answering and shows the result together with its source — unlike general-purpose tools that answer only from their own general knowledge.
How is your RAG different from a generic RAG?
Instead of working on a few uploaded files, we can connect RAG to the organization's own systems: the phone system, and Afrandeh's financial, inventory, HR and office-automation data.
Which phone systems does it connect to?
Recorded calls from Asterisk, FreePBX and Issabel phone systems are collected through a shared folder or SFTP and transcribed in Persian, and the number, direction and time of each call are stored next to the text.
Must my organization use Afrandeh?
Not for files (PDF, Excel, scans, video and audio) and the phone system. Direct connection to financial, inventory, HR and office-automation data is for organizations that use Afrandeh.
Where is our data processed?
Every component, from the database and search to the speech models, runs on your organization's own server, and documents, transcripts and question history never leave your organization's infrastructure.
Does it write answers like ChatGPT?
Not yet. Today it shows the most relevant passages with their source and relevance. A local large language model that writes full answers is in progress, and will also run without any external service.
Does it cover several companies or organizations?
Yes. Each company has an independent knowledge base, and each user searches only the companies they are allowed.
What do I do to see a demo?
Use the "Request a consultation & demo" button on this page and the Afrandeh team will get in touch.
Design your organization's own RAG with us
Book a short session and we'll look at where your organization's data lives and how it connects to RAG.
Request a consultation & demoAfrandeh is a product of Fekrgostar Fanavar.