The High-Performance and Explainable Language Processing Platform

Golem.ai Core is the no-training required artificial intelligence solution for building high-performance, robust, frugal, and unbiased NLP projects.

Build all your use cases with Golem.ai Core

Build your project from A to Z with our versatile NLP platform.

Extract the content of your documents to save reading time and automate their processing.
OCR processing - Content analysis - Information extraction
Extract the content of your documents to save reading time and automate their processing.
OCR processing - Content analysis - Information extraction
Analysts can create network structures to extract knowledge from different sources.
Text extraction - Content analysis - Information linking
Analysts can create network structures to extract knowledge from different sources.
Text extraction - Content analysis - Information linking
Process incoming messages by analyzing the message and its attachments.
Message and attachments analysis - Categorization - Information extraction
Process incoming messages by analyzing the message and its attachments.
Message and attachments analysis - Categorization - Information extraction
Protégez les données de vos utilisateurs et protégez-les des mauvais contenus.

Take advantage of the power of our revolutionary NLU

Explainable, frugal, multilingual, and customizable.

We confirm the arrival of the cargo ship Louis Bleriot containing operational equipent for hospitals at the port of Le Havre from the port of 香港. A two-hour delay in the unloding operation is expected.

Tokenization
Selection and separation of words (tokens) to keep only the relevant elements. Tokenization is enriched by the configuration, which is used to pre-select the relevant tokens.
We confirm thearrival of the cargo ship Louis Bleriot containing operational equipent for hospitals at the port of Le Havre from the port of 香港. A two-hour delay in theunloding operation is expected
Tokenization
Selection and separation of words (tokens) to keep only the relevant elements. Tokenization is enriched by the configuration, which is used to pre-select the relevant tokens.
Tokenization
Selection and separation of words (tokens) to keep only the relevant elements. Tokenization is enriched by the configuration, which is used to pre-select the relevant tokens.
arrival cargo ship Louis Bleriot operational equipent hospitals Le Havre from 香港 two-hour delay operational unloding
Tokenization
Selection and separation of words (tokens) to keep only the relevant elements. Tokenization is enriched by the configuration, which is used to pre-select the relevant tokens.
Dict:Multi
Correction of terms according to the appropriate language and business usage.
arrival cargo ship Louis Bleriot equipment operational hospitals Le Havre from
Hong Kong
delay two hours operational unloading
Dict:Multi
Correction of terms according to the appropriate language and business usage.
Chunking
Grouping business terms to improve the understanding.

arrival cargo ship


Louis


Bleriot equipment


operational


hospitals Le Havre from


Hong Kong delay two


hours


operational


unloading

Chunking
Grouping business terms to improve the understanding.
Chunking
Grouping business terms to improve the understanding.

arrival .cargo ship


Louis


Bleriot .equipment


operational


hospitalsLe Havre . from


Hong Kong . delay .two


hours


operational


unloading

Chunking
Grouping business terms to improve the understanding.
Named Entity Recognition
Assigning an entity type to each term.

arrivalStatus.
cargo shipTransport


LouisNickname


BleriotName.
equipmentProduct


operationalCharacteristic
or action


hospitalsSector.
Le HavrePlace.
fromDescriptor


Hong KongPlace.
delayStatus.
twoQuantity


hoursTime


operationalCharacteristic
or action


.
unloadingAction

Named Entity Recognition
Assigning an entity type to each term.
Entity Linking
Creating links between entities to resolve a textual entity into a unique identifier from a knowledge base.
arrivalStatus.
‘transport’ : ‘cargo ship’,
‘name’ : ‘Louis Bleriot’
Transport
.
‘product’ : ‘equipment’,
‘characteristic’ : ‘operational’,
‘sector’ : ‘hospitals’
Product
.
Le HavrePlace of arrival.
Hong KongDeparture place.
delayStatus.
‘number’ : ‘2’,
‘time’ : ‘hours’
Time

unloadingAction
Entity Linking
Creating links between entities to resolve a textual entity into a unique identifier from a knowledge base.
Dependency Parsing
Completing the understanding of the text by adding each term to an ontology.
statusarrivalStatus.
transport > ship > cargo ship > Louis BleriotIMO 9776432Transport.
product > medical devicesmedical devicesProduct.
France > portLe HavrePlace of arrival.
China > portHong KongDeparture place.
statusdelayStatus.
‘number’ : ‘2’,
‘time’ : ‘hours’
Time

action > action shippingunloadingAction
Dependency Parsing
Completing the understanding of the text by adding each term to an ontology.
Interaction
Linking different terms based on an ontology to form a unit of meaning.
Delivery tracking
statusarrivalStatus
transport > ship > cargo ship > Louis BleriotIMO 9776432Transport
product > medical devicesmedical devicesProduct
France > portLe HavrePlace of arrival
China > portHong KongDeparture place
Delivery status
statusdelayStatus
‘number’ : ‘2’,
‘time’ : ‘hours’
Time.

action > action shippingunloadingAction
Interaction
Linking different terms based on an ontology to form a unit of meaning.

Text Extraction from
images and documents

Easily transform your documents into usable texts using our Extractor technology.

Several OCRs and extraction libraries available via API.

				
					package main

import (
  "fmt"
  "strings"
  "net/http"
  "io/ioutil"
)

func main() {

  url := "https://extractor.golem.ai/v3/analyse"
  method := "POST"

  payload := strings.NewReader(`{
    "file": "https://www.yourfile.pdf"
}`)

  client := &http.Client {
  }
  req, err := http.NewRequest(method, url, payload)

  if err != nil {
    fmt.Println(err)
    return
  }
  req.Header.Add("Authorization", "Basic XXX")
  req.Header.Add("Content-Type", "application/json")

  res, err := client.Do(req)
  if err != nil {
    fmt.Println(err)
    return
  }
  defer res.Body.Close()

  body, err := ioutil.ReadAll(res.Body)
  if err != nil {
    fmt.Println(err)
    return
  }
  fmt.Println(string(body))
}
				
			
				
					<?php

$curl = curl_init();

curl_setopt_array($curl, array(
  CURLOPT_URL => '"https://extractor.golem.ai/scan"',
  CURLOPT_RETURNTRANSFER => true,
  CURLOPT_ENCODING => '',
  CURLOPT_MAXREDIRS => 10,
  CURLOPT_TIMEOUT => 200,
  CURLOPT_FOLLOWLOCATION => true,
  CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
  CURLOPT_CUSTOMREQUEST => 'POST',
  CURLOPT_POSTFIELDS =>'{
    "file": "https://www.yourfile.pdf",
    "useCache": true,
    "parsers": {
        "document": {
            "extractImages": false,
            "ocr": {
                "name": "tesseract",
                "mode": "auto"
            },
            "PDF": {
                "extractImages": false,
                "ocr": {
                    "name": "ida",
                    "mode": "on"
                }
            }
        },
        "image": {
            "minimumHeight": 500,
            "minimumWidth": 500,
            "ocr": {
                "name": "ida",
                "mode": "off"
            },
            "png": {
                "minimumWidth": 100,
                "ocr": {
                    "name": "ida"
                }
            }
        },
        "spreadsheet": {
            "readVertically": false,
            "unmergeCells": false,
            "splitPerBlock": false,
            "splitPerBlockRowLimit": 10,
            "splitPerBlockColumnLimit": 10,
            "parseHiddenSheets": false
        },
        "email": {
            "extractAttachments": false,
            "ignoredAttachments": [
                "xlsb",
                "eml"
            ],
            "msg": {
                "extractAttachments": true
            }
        }
    }
}',
  CURLOPT_HTTPHEADER => array(
    'Authorization: Basic XXX',
    'Content-Type: application/json'
  ),
));

$response = curl_exec($curl);

curl_close($curl);
echo $response;
				
			
				
					import requests
import json

if __name__ == "__main__":
    URL: str = "https://extractor.golem.ai/scan"

    payload: dict = json.dumps(
        {
            "file": "https://www.yourfile.pdf",
            "parsers": {
                "document": {
                    "extractImages": False,
                    "ocr": {"name": "tesseract", "mode": "auto"},
                    "PDF": {
                        "extractImages": False,
                        "ocr": {"name": "ida", "mode": "on"},
                    },
                },
                "image": {
                    "minimumHeight": 500,
                    "minimumWidth": 500,
                    "ocr": {"name": "ida", "mode": "off"},
                    "png": {"minimumWidth": 100, "ocr": {"name": "ida"}},
                },
                "spreadsheet": {
                    "readVertically": False,
                    "unmergeCells": False,
                    "splitPerBlock": False,
                    "splitPerBlockRowLimit": 10,
                    "splitPerBlockColumnLimit": 10,
                    "parseHiddenSheets": False,
                },
                "email": {
                    "extractAttachments": False,
                    "ignoredAttachments": ["xlsb", "eml"],
                    "msg": {"extractAttachments": True},
                },
            },
        }
    )

    headers: dict = {"Authorization": f"Basic XXX", "Content-Type": "application/json"}

    response: requests.Response = requests.request(
				"POST", URL, headers=headers, data=payload
		)

    print(response.text)
				
			
				
					var settings = {
  "url": "https://extractor.golem.ai/v3/analyse",
  "method": "POST",
  "timeout": 0,
  "headers": {
    "Authorization": "Basic XXX",
    "Content-Type": "application/json"
  },
  "data": JSON.stringify({
    "file": "https://www.yourfile.pdf"
  }),
};

$.ajax(settings).done(function (response) {
  console.log(response);
});
				
			

Golem.ai protects and respects your data

Our artificial intelligence allows us to respect your data by design.

Security

Golem.ai follows the cryptography recommendations issued by ANSSI.

Privacy

Golem.ai's AI is hosted at Scaleway in France. You remain the exclusive user and owner of your data.

Compliance

An accessible and documented API, available connectors.

La puissance technologique
au service de votre relation client

Miralia repose sur une IA frugale, explicable et respectueuse, conçue pour traiter uniquement ce qui compte, au moment où cela compte.

Notre approche neuro-symbolique

Allier le meilleur des deux mondes.
Plus qu’une promesse, c’est le fonctionnement même de la technologie Miralia.
L’IA Neuro-symbolique, c’est la puissance de l’IA générative, la fiabilité et l’explicabilité en plus.

Compréhension fine du langage

Grâce à l’IA générative pour interpréter des demandes variées et complexes.

Décisions fiables et explicables

Chaque action repose sur des règles claires, traçables et compréhensibles.

Adaptation aux métiers

Une IA qui apprend de vos données tout en respectant vos process existants.

Une technologie conçue pour vos enjeux

Notre IA n’est pas une “boîte noire”. Elle apporte des bénéfices concrets, tangibles et mesurables.

Performance

The quick setup allows for visible results in the first weeks without semantic complexities or hallucinations.

Explicability

All choices resulting from artificial intelligence are explainable, traceable, and demonstrable.

Souveraineté

Golem.ai's artificial intelligence is 100% French, proprietary and hosted by Scaleway.

Frugality

This technology consumes very few server resources and has a low CO2 impact.

Une plateforme modulaire et complète

La technologie Miralia repose sur différents modules, interopérables entre eux et adaptables à vos besoins métier.

logo and lines background
Logo M Miralia in a circle

NLU

Analysez et classifiez chaque message suivant ses intentions et ses contextes, quelque soit la langue.

Extractor

Transformez simplement vos images et documents en textes exploitables.

Moteur de réponse

Générez des propositions de réponses personnalisées et conformes à vos process.

Vision Engine

Analysez, classez et exploitez vos visuels et documents complexes avec une précision inégalée.

La sécurité au cœur de notre technologie

Our artificial intelligence allows us to respect your data by design.

Logo Encryption

Chiffrement bout-en-bout

Encryptage standard AES-256. Transits de données encryptés en utilisant le standard TLS (HTTPS ou AMPQ).

Logo Conformité

Compliance

Notre plateforme respecte les normes internationales de sécurité et de protection des données.

Logo Scaleway

Souveraineté

Les solutions Miralia bénéficient d’un hébergement souverain ultra sécurisé chez son partenaire Scaleway en France.

Redonnez sens et maîtrise
à votre relation client

Essayez dès aujourd’hui notre IA explicable, frugale et fiable.