The most advanced Frontend for Conversational AI
Build, prototype, train & publish your conversations & speech models as easy as never before.
Made for your entire team.
Made for VUI designers.
Drag & Drop flow builder. Visual SSML editor. Autocomplete. Text Snippets. And more.
Made for product owners.
Keep track of each dialogue flow, challenge it with your responsible team and present it on the spot to your manager.
Made for developers.
Made for trainers.
Keep track of every past conversation and train your speech model with real data at one place.
The drag & drop flow builder.
Building automated dialogs couldn’t be simpler – and never more colorful!
You can easily create great dialog experiences per drag & drop. It provides a clear interface for copywriters, product owners & developers. Complex dialogs are mapped more easily than ever before thanks to several blocks:
Write your responses and define intents for subsequent user interactions.
Connect Parloa to every external code or service such as CRMs, ERPs and exchange data with it.
Branch your dialog based on user inputs or other back-end requirements.
Set and alter specific user’s inputs or other conversational values in variables.
To Prev. States
Handle unexpected user inputs without losing track in your conversation.
Easily forward a call or use SIP REFER, whenever it’s time for a human to take over.
And always stay on top of things:
Organize dialogs into reusable subdialogs.
Ever lost track of your complex dialogs? Put an end to confusing flowcharts and say ‘Hi’ to Parloa’s subdialogs:
- Keep track of your dialog flow and go deeper into the dialog with each subdialog
- Create dialogs or functional logics (e.g. customer authentication) once to refer to them at different places in your dialog
The best debugger since debugging.
Anyone can design dialogs. But making them work is the real art. This is why we have invented the Parloa Debugger.
It helps you find logic errors in your conversation flow, interpret NLU responses, and understand why external services may not be responding correctly.
- Access all context information of platform and user values, conditions, or back-end connections
- Highlight changes to find what’s broken faster
- Analyze NLU confidence levels for intents & slots
- Jump to the affected dialog section with one click
The visual SSML response editor.
You'll be speechless, but it'll just get you chatting!
Thanks to Parloa’s intuitive autocompletion you can write state-dependent but dynamic responses for each channel, without writing a line of code yourself!
Vary your dialogue with varied responses based on user intent.
Different channels require different responses, e.g. in the difference between text and sound.
You can also provide varied answers to misunderstandings between humans and machines.
Why write “Ok” over and over again when you can work with reusable varied Text Snippets?
Write individual & consistent responses per sessions thanks to both platform & user inputs.
- Various responses for your channels
Define separate response elements for each voice or text channel – because “😍” can’t be read out loud by Alexa.
- Built-in SSML tags list
Use all common SSML tags (e.g. <audio> or <say-as>) and also platform specific effects & layers to optimize the sound of your answers.
- Contextual variables
Create individual responses, using platform context and user input to maximize both consistency and variety of speech.
Because automation does not mean standardization:
Use (conditional) text snippets.
We invented the Text Snippets for maintaining important and reusable text modules in a separate place.
This allows you to customize texts for the entire dialog with just a few clicks.
Talk to your customers based on conditional text entries, reference Text Snippets in other texts, and chose the right selection strategy (random, ordered, etc.) – because your customers deserve the best possible automated experience!
To minimize the complexity in the Conversational Flow, you will find Text Snippets in a separate view.
Type in a word from your dialog and find what you wrote in no time thanks to keyword search.
Let Parloa chose your texts randomly, ordered or cyclical in order to improve your customer’s experience.
Create important or recurring text modules once and reuse them in other Text Snippets. For example, you can easily reuse your company name in other Text Snippets – and change it with only one click.
The Speech Model Editor
Manage your NLU without a line of code.
Define all intents, train them with utterances, create your dictionary for synonyms and receive variable user inputs in slots.
Decide for each intent on which channel it is needed and wether it should be triggered on events, or set custom confidence thresholds.
Teach the NLU about possible words and phrases that can be said or written for this Intent. Of course you can also upload your existing training data via Bulk Edit.
Define Dictionary entries for synonyms and reuse it everywhere in Parloa.
Use predefined slots (e.g. ) or rely on machine learning and give the NLU hints in your utterances what to expect.
- Add Intents in the blink of an eye
Enable and disable specific intents for your platforms with one click, define different confidence levels, or trigger them only for specific events.
- Write Utterances or use Bulk Edit to upload your existing training data
Easily use your existing training data via Bulk Edit to train your NLU – or write it yourself.
- Benefit from your own dictionary
We invented the Parloa Dictionary to save you time when creating Utterances.
Learn from past conversations
Training makes the dream work.
Sustainable automation in customer communication starts with collecting training data.
And with Parloa you can not only keep an eye on past conversations but also train your NLU directly by adding unmatched utterances to your intents from the unique Parloa NLU Training Interface.
Filter by confidence levels, specific intents, or time periods to improve your NLU focused.
Just select the right intent for misunderstood utterances. The rest is magic.
If a misunderstood utterance is ambiguous, get context through the conversation history.