Alga Solutions
TREN
RegattaAI
Alga’s AI product family

AI modules for call and image data

RegattaAI modules are integrated into Alga products and into your organisation’s own processes. All modules are available now.

A · Conversation Intelligence

Audio → text → anonymization → analysis

These modules process call recordings end to end. Built for collections and call centre operations, they work integrated with Recall and can also be used on their own.

  1. Audio

    Call recordings from your phone system or from Recall. Mono and stereo recordings are supported.

  2. Text

    Transcription (ASR)

    Converts call recordings to text automatically, using models tuned to industry terminology.

    Technical details

    • 95%+ recognition accuracy (measured by WER)
    • Language models tuned to industry terminology
    • Mono and stereo recordings supported
    Transcript outputExample · fictional
    [00:00:12]

    Agent: Hello, good afternoon.

    [00:00:14]

    Debtor: Hello.

    [00:00:15]

    Agent: Could I speak to Jane Doe, please?

    [00:00:17]

    Debtor: Speaking.

    [00:00:18]

    Agent: Ms Doe, I’m calling from Example Asset Management. This call is being recorded.

  3. Anonymization

    Anonymization

    Masks sensitive data such as names, phone numbers and national ID numbers, so calls can be analysed with AI in line with KVKK and GDPR.

    Technical details

    • KVKK and GDPR compliant
    • Automatic masking based on named entity recognition (NER)
    • Irreversible masking
    MaskingExample · fictional
    Original
    Agent: Could I speak to Jane Doe? We have your phone number as 0555 000 00 00.
    Masked
    Agent: Could I speak to [PERSON-1]? We have your phone number as [PHONE].

  4. Analysis

    The masked text is analysed by 5 modules.

    Contact Classification

    Answers "who did the agent speak to?" (the debtor, a relative, etc.) automatically, removing manual entry errors.

    Technical details

    • Verifies who is on the line through voice signature and content analysis
    • KVKK-compliant third-party check
    • Contact quality metrics generated automatically

    Call Outcome

    Tags calls as Reached, Debt Negotiation, Payment Plan or Collection Follow-up, showing where the portfolio is heading.

    Technical details

    • Four main tags: Reached, Debt Negotiation, Payment Plan, Collection Follow-up
    • Tags tailored to each organisation
    • Input for collection forecasting models

    Non-Compliant Call Detection

    Flags insults, threats and missing legal disclosures, with an early warning before they turn into complaints.

    Technical details

    • Detects insults, threats and slang
    • Checks for legal disclosures (KVKK, etc.) that were not read in full
    • Sentiment analysis

    Callback Detection

    Picks up requests such as "call me on Friday at 2 pm" and creates the callback in the system automatically.

    Technical details

    • Captures date and time expressions with natural language processing
    • Calendar and CRM integration
    • Automatic reminders and tasks

    Next Call Guidance

    Gives agents tailored persuasion tips based on the debtor’s profile and past objections.

    Technical details

    • Debtor profile analysis
    • Counter-arguments to past objections
    • Personalised persuasion strategy
    Analysis outputExample
    Spoke with
    Debtor
    Outcome
    Payment Plan
    Non-compliant language
    None detected
    Callback
    Friday 2 pm · callback
    Next call
    Mention the instalment option

Integrated with Recall

Conversation Intelligence results appear in Recall alongside the related case file.

Explore Recall

Standalone use

The modules can also process call recordings for organisations that do not use Recall.

B · Vision Intelligence

License plate recognition

AlgaLPR

License plate recognition built for Turkish plates: an embeddable .NET library for automotive and car park use.

Technical details

  • Built for Turkish plate formats
  • Embeddable .NET library (ONNX-based)
  • Car park, service check-in and access control scenarios
AlgaLPR plate recognition example: the plate marked on the vehicle, the text read and the confidence score — illustrative product image

See the modules at work on your own recordings.

Choose RegattaAI on the request form and our team will get in touch.

Request formOur documents