Caplena news
Caplena news
caplena.com

Workshop on AI-assisted text analysis (GERMAN ONLY)

 

Event

 

 

As part of a DGOF workshop series, Pascal will be holding a session on AI-assisted text analysis for market research. We'll cover different approaches for different kinds of text data and go over many practical examples.

The workshop will be in German. For more information and to register visit: https://www.dgof.de/event/dgof-workshop-ki-unterstuetzte-textanalyse-fuer-marktforscherinnen/

Chart Excel export and Performance improvements

 

Feature

 

 

Want to include our charts in PPT or Excel?

While you could always export our charts to PNG or SVG, more customization remained a constant request. This is why we've built the option to export the data underlying your charts to Excel files. Look for the table icon in the chart editor to use this new functionality.

Screenshot 2021-06-01 at 22.03.27.png

Also, we greatly improved loading speed and stability in our visualizations (charts and dsahboards) 🏎.

Revamped Upload

 

Feature

 

 

As you might have noticed, our upload got a complete overhaul πŸ’….

Screenshot 2021-05-05 at 20.13.07.png

Besides a beautiful new design πŸ‘¨β€πŸŽ¨, there are a number of new features and improvements:

  • Stability: We can now deal with much larger files, up to 200k rows at once. If you want your default limit increased, let us know;
  • Automatic column identification: We now identify the open-ended columns automatically, saving you a lot of clicks;
  • Sheet selection: Choose which sheet you want to upload, no need to create a new file anymore;
  • Drag & Drop upload: I know, we were late to the party. But hey, better late then never πŸ€·β€β™€οΈ;
  • ID column: Identify duplicate items by an ID column when appending to an existing project.
  • … many more.

Try it out and let us know what you think - and how we can get even better πŸ“ˆ.

CX training πŸ§‘β€πŸŽ“ offered by our partner CX.AI

 

Event

 

 

Doing customer experience right is hard. That's why our partner Frank Buckler from CX.AI is hosting a CX Analytics Master course. Within 10 classes, he will show you how to set up a professional feedback pipeline, from collection to predictive NPS.

We are looking forward to presenting how Caplena plays a crucial role in a modern CX workflow πŸ“š during this course.

Updates to billing logic

 

Fix

 

 

In contrast to most other SaaS services, we try to only charge credits to your account when you really use the data. We also do not charge extra for translations, which we want to keep this way. However, we of course also want to make sure that this system is not misused.

There is no change in how much our service costs, only the moment of credit-deduction is changed in some cases.

What changes

  • Projects with more than 2500 responses and activated translations will be charged immediately on upload;
  • Saving charts will deduct the credits, if the project data has never been used before;
  • Uploads through the API will be billed on upload immediately, except when uploading reviewed responses.

What stays the same

  • The pricing;
  • Every response will only be charge once - after that, you can modify, visualize and download it as often as you want;
  • You will always be informed of a credit deduction before the action through such an info dialog:

Credit deduction Dialog

New Feature: Dummy variables

By default we do not count empty responses towards your credits. But if you have additional dummy data in your files (e.g. -99 representing empty responses), you can now define up to three dummy-values on your account page which will not be counted in subsequent uploads.

Dummy variables

Why LEGO-blocks are the future of insights software

 

Event

 

 

Join our Webinar on Wednesday, 17. June 14pm CEST

Lego_Picture.png

Learn how creative professionals build modular insights pipelines for their company and innovative solutions for clients, by plugging together lean, best-in-class products. Focusing on language, text and AI, we will showcase real-world examples that exist today.

Click here to register.

We are excited to have Frank Buckler, founder & CEO of Success Drivers as our guest. Frank is a pioneer in the field of market research AI solutions, and will offer his view & expertise on this exciting topic and the future of research software.

Webinar am 27. Mai mit Joyn [German]

 

Event

 

 

Jetzt registrieren zu unserem Webinar am 27. Mai mit Joyn 🎬. Thema:

Wie KI Apps dem kleinen Research-Team bei Joyn agiles Handeln auf strategischer Ebene ermΓΆglichen.

Wir freuen uns sehr Kilian Hughes, Leiter der Research & Insights-Abteilung beim deutschen Streaming-Startup Joyn als Gast mit dabei zu haben. Mit ihm werden wir (natΓΌrlich) ΓΌber Textanalyse sprechen, aber auch darΓΌber, wie Kilian's Team Automatisierungstools nutzt, um sich als Drehscheibe einer datengetriebenen Entscheidungsfindung zu positionieren.

User Management πŸ‘©πŸΎβ€βš•οΈπŸ‘¨β€πŸ’»πŸ‘©πŸ»β€πŸŽ“

 

Feature

 

 

Caplena has had multi-user capabilities for quite some time. But until now, you had to get in touch with someone from our team if you wanted to add another team member to your account. Not anymore.

User management interface

User management is now completely self service.

Specifically, you can:

  • Add new users to your organization,
  • Resend activation emails,
  • Change the access roles of your team members,
  • Remove users.

So go ahead and spread the word about Caplena in your Company! πŸ“’πŸ˜Š

Major improvement to API

 

Improvement

 

 

As more and more users make use of our API to automate data up- and download, connect Caplena to their own systems or even build entire solutions on top, we have made an effort to significantly improve the capabilities and ease-of-use of our API.

TL;DR

  • Instant automatic coding is computed by default if some conditions are met. If you do not require instant automatic coding, enable the async parameter to upload new rows, to make sure everything works as before.
  • Predicted codes will be written to the answers automatically, no need to call the codes-predicted endpoint anymore or open the frontend,
  • Switch to API key authentication if you are still using session authentication before May 15 2020.

Instant automatic coding

If you rely on the classification results to trigger some further logic, e.g. when routing support requests based on their text, or running a chatbot with them, this has just become much easier: If your plan includes API access, we will now try to do an instant classification by default. A few conditions have to be met, however:

  • You have to send a maximum of 20 responses at once and
  • You have to call the endpoint without the async option and
  • If you have reviewed answers on the project, the training 1 needs to have been triggered before at least once.

Integration of automated coding results

Until now, the results from the automatic coding had to be fetched from the codes-predicted endpoint and then written into the answer manually. Or, you had to open the frontend once in a while to get the predicted codes written.

This cumbersomeness is now obsolete: Automatically assigned codes will always be written into the answers directly after having been computed.

Authorization

Session authorization with email and password is not suited well for API use, especially from a security perspective. We have supported API keys for a while already, but until now you had to request them from our support team. Not anymore: You can now generate, disable and delete API keys directly on your Account Page.

Important: Session authentication for API usage will be deprecated on 15. May 2020. If you use session authentication currently, please switch to API keys.

Limitations and new validation

When adding new rows (POST /api/projects/{PROJECT_ID}/rows) or creating a project with rows (POST /api/projects) and calling it synchronously (without async query parameter), the following conditions are enforced newly:

  • A maximum of 20 rows can be uploaded at once
  • You cannot upload answers with reviewed: True
  • If there is no fine-tuned AI 1, or the fine-tuned AI fails to predict codes for some reason, the entire request will fail and the answers will not be saved

If you do not require instant automatic coding, we recommend enabling the async option, for which the above mentioned limitations to not apply

1 A fine-tuned AI is available if

  • there are no reviewed responses on a question and the question doesn't inherit from another one or
  • If training has been requested for your question and
    • The training has been requested less than 30 days ago or
    • Rows have been appended to the question synchronously within the last 30 days.

Two-factor authentication

 

Feature

 

 

2FA QR code

Security has always been a key priority for us, that's why we have implemented two-factor authentication. Add a second layer of security to your account, either through an App or SMS.

Go to your account page to activate it now.