algofin 1.0

The future of financial decisions.

A model trained for financial decisions. It reads a customer message, a news story or a report and gives a probability for each possible answer. Financial firms license it and run it on their own servers.

Join the waitlist
  • 84.8%on financial sentiment, 11.9 points ahead of Jev 1.13
  • 1 GPUis all it needs to run, inside your own network
  • 0bytes of your data sent to OpenAI, Anthropic or Google

Use cases

Anywhere your team sorts, scores or screens text. algofin 1.0 reads it once and gives a probability for each possible answer.

  • Market sentiment

    Read news, filings and research notes, and score each one for what it means for the company.

    In

    “Operating profit rose to EUR 13.1 mn from EUR 8.7 mn a year earlier.”

    Out, one of

    • Positive
    • Neutral
    • Negative
  • Customer requests

    Route each customer message to the right team, with the request it is about.

    In

    “I was charged twice for the same card payment yesterday.”

    Out, one of

    • Duplicate charge
    • Card payment fee
    • Refund not received
  • Scam screening

    Flag scam and spam messages before they reach a customer or an agent.

    In

    “Your account is locked. Verify your details at the link below.”

    Out, one of

    • Scam
    • Genuine

Against Jev

algofin 1.0 beats Jev by 11.9 points on financial sentiment and matches it on banking intent and scam detection.

Scored zero-shot on the University of Bonn’s published benchmark for Jev, with their evaluation code, unmodified.

Financial sentiment

algofin 1.0
84.8
Jev 1.13
73.0

Banking intent

algofin 1.0
79.4
Jev 1.13
79.7

Scam detection

algofin 1.0
93.6
Jev 1.13
93.8
Accuracy, or F1 for scam detection, %. Zero-shot. Jev 1.13 scores: University of Bonn, arXiv 2609.37647.

With 95% intervals

On financial sentiment, our interval sits wholly above Jev’s.

Financial sentiment

algofin 1.0 84.8, 95% interval 82.5 to 87.0
Jev 1.13 73.0, 95% interval 69.9 to 75.6

Banking intent

algofin 1.0 79.4, 95% interval 77.9 to 80.8
Jev 1.13 79.7, 95% interval 78.2 to 81.2

Scam detection

algofin 1.0 93.6, 95% interval 92.4 to 94.7
Jev 1.13 93.8, 95% interval 92.4 to 95.0
Dots are scores; lines are 95% bootstrap intervals. Accuracy, or F1 for scam detection, %. Zero-shot. Jev 1.13 scores: University of Bonn, arXiv 2609.37647.

Self-hosted

algofin 1.0 runs on your own servers. Your messages, documents and decisions never leave your network.

Your network

InMessages, news and reports

algofin 1.0

on your own hardware

OutDecisions, with a probability for each answer

Never sent to

  • OpenAI
  • Anthropic
  • Google
  • Anyone else

Licensing

Financial firms license algofin 1.0 and run it on their own infrastructure.

Hosted AI APIalgofin 1.0
Runs onThe provider's serversYour servers
Your dataSent to the providerStays on your network
Model changesOn their scheduleWhen you choose
Runs air-gappedNoYes

Run algofin 1.0 on your own servers.

Licences start at $10,000. Apply for the waitlist and we will be in touch.

As seen in