Community Memo on Rasa-Pro launch!

Announcements
1

Hey Everyone,

I am happy to share that a new commercial product has been released today. Rasa Pro is powered by Rasa Open Source and has been built to effectively respond to enterprise needs for security, observability, and analytics. If you work in an enterprise and want to know more about Rasa Pro you can read more about the capabilities here.

For the Rasa Open Source Community, note that the release of Rasa Pro does not mean we are taking away any features from Rasa Open Source. As a matter of fact, we want to double down on our commitment towards Rasa Open Source by continuously innovating, especially in the areas of NLU, dialogue management, and the extensibility of the framework to support low-resource languages. We have also launched a new website just for the community at: rasa.community, where you can find everything about Rasa Open Source, the Rasa Learning Center, and the community.

Rasa Open Source is the de facto standard for conversational AI. Since its launch, it has been downloaded more than 25 million times and runs everywhere from startups to the largest companies in the world. For the latter, there are specific requirements that are associated with conversational AI running at scale. Rasa Pro aims to solve some of these key challenges in large enterprises by making it easier to integrate Rasa with other applications and technology stacks, and additionally provide end-to-end visibility into AI Assistant performance and metrics.

We also want to thank everyone in the community for their constant support, and we’re excited to move towards making conversational AI better together.

10 Likes
2

@sonam, what’s Rasa Plus and Rasa Pro Services and how does it related to Rasa Pro. The announcement doesn’t say anything about either of them but the docs reference them without explaining how they relate to Rasa Pro.

3

Hi @stephens, The announcement was only about the launch of Rasa Pro. You can refer to this page to learn further about Rasa Plus and Rasa Pro services.

1 Like
4

Hi @sonam, is there a chance to take rasa-pro for a test drive. I submitted application couple of times for a trial, but no response do far. I am mostly interested in the LLM integration and working without intents, thanks

5

Hey,

Sorry for late reply, I was on a vacation. I think there are no trials for Rasa-Pro but if there is, I will confirm and let you know.

6

Hi @jkothamb

You can submit a request for a trial using this link

1 Like
7

Hi @sonam, I did try multiple times, but no response. Hence the reason for reaching out to you

Thanks, Juzer

8

Hi @sonam : do you have any update on trial access to rasa pro. So far I only received links to some resource tutorial

9

Hey @jkothamb , I have forwarded the message to the concerned team. Thank you.

10

Hi @sonam I didn’t know how to reach you. I have a question. I joined a team last year. Team uses RASA and they started to change from RASA v1 to v2.8. They made some chatbots with v2 and we didnt know the difference. Now we are trying to change a project from v1 to v2. And our accuracy dropped from 87 to 84. Our pipeline is same. nothing changed. Here is our pipeline:

pipeline:
  - name: pyspace.rasa.components.duckling.DucklingExtractor20201220
    url: 'http://duckling-ahe:8000'
    prediction_in_train_time: true
    prediction_entity_write_attribute: duckling
  - name: pyspace.rasa.components.regex_v2.RegexEntityExtractor20210528
    prediction_in_train_time: true
  - name: pyspace.rasa.components.data_management.BackupFeatures
    features:
      - text
  - name: pyspace.rasa.components.spacytokenizer.RasaSpacyTokenizer
    whitespace_normalizer: true
    predefined_token_patterns: true
    predefined_replace_patterns: false
    post_tokenization_merge_bool: false
    emoji_bool: false
    stanza_bool: false
    predefined_matcher_bool: false
  - name: pyspace.rasa.components.normalization.TurkishCharacterNormalizer
  - name: pyspace.rasa.components.normalization.LowercaseNormalizer
    lang: EN
  - name: CountVectorsFeaturizer
  - name: CountVectorsFeaturizer
    analyzer: char_wb
    min_ngram: 3
    max_ngram: 5
  - name: pyspace.rasa.components.diet.DIETClassifier
    epochs: 32
    weight_sparcity: 0.3
    dense_dimension:
      text: 32
      label: 20
    hidden_layer_sizes:
      text:
        - 32
    number_of_transformer_layers: 2
    number_of_attention_heads: 8
    transformer_size: 32
    embedding_dimension: 50
    entity_recognition: true
    intent_classification: false
    prediction_entity_write_attribute: diet
    prediction_in_train_time: true
    prediction_condition: false
  - name: pyspace.rasa.components.clear.ClearFeatures
  - name: pyspace.rasa.components.normalization.EntityNormalization
    normalization_write_attribute: ent_norm
    normalization_config:
      - - diet
        - DIETClassifier
        - city
        - ''
        - A
      - - diet
        - DIETClassifier
        - district
        - ''
        - B
      - - regex
        - RegexEntityExtractor
        - policy_no
        - ''
        - C
      - - duckling
        - DucklingExtractor
        - time
        - ''
        - D
  - name: pyspace.rasa.components.normalization.EntityMergeSame
  - name: pyspace.rasa.components.normalization.EntityRemoveRawAttribute
  - name: LexicalSyntacticFeaturizer
  - name: CountVectorsFeaturizer
  - name: CountVectorsFeaturizer
    analyzer: char_wb
    min_ngram: 3
    max_ngram: 5
  - name: pyspace.rasa.components.diet.DIETClassifier
    epochs: 50
    weight_sparsity: 0.3
    entity_recognition: false
    use_masked_language_model: false
    intent_classification: true
    dense_dimension:
      text: 64
      label: 20
    weight_sparcity: 0.3
    hidden_layer_sizes:
      text:
        - 64
    number_of_transformer_layers: 2
    number_of_attention_heads: 8
    transformer_size: 64
    embedding_dimension: 50
    prediction_intent_write_attribute: false
    prediction_in_train_time: false
    prediction_condition: false
  - name: pyspace.rasa.components.update.TrainingDatasetPrediction
    update_intent_ranking: true
  - name: pyspace.rasa.components.synonyms.EntitySynonymMapper
    fuzzywuzzy_bool: true
    fuzzywuzzy_limit: 60
    add_train_entities: false
  - name: pyspace.rasa.components.data_management.RecoverFeatures
    features:
      - text
  - name: >-
      rasa_addons.nlu.components.intent_ranking_canonical_example_injector.IntentRankingCanonicalExampleInjector

Do you have any suggestions? Maybe to have the same results we must change our pipeline when we switch to v2. Or maybe some parameters change? Thank you for helps.

12

Hey @sbiligil ,

Is this on Rasa Pro or Rasa OSS? kindly confirm

13

It is Rasa OSS. Our ex-teammates cloned the code and build it with some new components as you can see from pipeline like turkishcharacternormalization.

14

Please check the DM

15

Hi Sonam, Hello community

I’m a new user of Rasa opensource and I started working with it just a few days ago. My goal is to use Rasa in my project for a chatbot that helps with the training of Healthcare Peer Workers. I’m currently in training to get accredited as such. During my educational journey, I found out pretty quick, that one of the biggest needs is to have enough practice with leading coaching sessions. There are not enough training partners around to any given time. This is why I started to build a chatbot that helps with this. May I ask here, if I did understand this correctly, that Rasa’s base is the opensource platform. This is the same base that is used for Rasa Pro and Rasa X / Enterprise, right? To use Rasa Pro I would need to implement the Rasa Plus Hook on top of Rasa opensource, I guess. Is there a chart where I can directly compare the options that each product class has? That would be very helpful. Thanks in advance for any hints. I like the product so far and I’m looking forward to get to know it better.

16

Hi Markus,

Thank you for reaching out to the community. You will need to get in touch with the sales team for Rasa-Pro. You can try this link.

Also find the comparison chart.