Talkative Features & Data

Below is a list of what each Talkative AI feature does, what data it processes, and how it works.

All AI features are optional and configurable. No customer data is shared with third parties outside of the sub-processors disclosed in Talkative’s standard Data Processing Agreement.

Our Data Processing Agreement can be found here: Data Processing Addendum

Feature Table

Feature	Description	Categories of Data Processed	PII	AI Models and Subprocessors	Data Locations	Duration of Storage With Subprocessor
Voice AI	Real-time communication with an interactive voice bot.	Conversation transcript, interaction data	Yes	Choose your LLM (see table 1), and STT and TTS providers (see table 2)	Refer to table 1 and 2	Refer to table 1 and 2
Chat AI	Real-time communication with an interactive chat bot via website/SMS/social.	Conversation transcript, interaction data	Yes	Choose your LLM	Refer to table 1	Refer to table 1
Email AI	Provides suggested responses to emails.	Conversation transcript (email contents)	Yes	Choose your LLM	Refer to table 1	Refer to table 1
AI Agent Assist (KB suggestions, Navi)	Provides real-time agent suggestions based on ongoing chat.	Conversation transcript, interaction data	Yes	Choose your LLM	Refer to table 1	Refer to table 1
AI Agent Rephrase	Offers rephrasing options for agent-written responses.	Agent input text only	No	Regional hosted LLAMA	Regional, same as Talkative AWS instance	0 days
AI Agent Training	Uses fictitious interactions to create scenarios for agent upskilling.	Fictitious conversation transcript	No	GPT4.1	US	0 days
AI Interaction Summaries	Generates summaries of customer interactions for internal review.	Conversation transcript	Yes	Regional hosted LLAMA	Regional, same as Talkative AWS instance	0 days
AI Interaction Data Capture	Automatically extracts key fields (e.g. names, issues).	Conversation transcript	Yes	GPT5 Mini	Regional, same as Talkative AWS instance	0 days
AI Analytics Assistant	Enables querying customer interaction trends using natural language.	Analytics reports	No	GPT4.1	US	0 days
AI Insights Report	Periodic reports highlighting trends, issues, and agent performance.	Conversation transcripts	Yes	Gemini	Regional, same as Talkative AWS instance	0 days
AI Sentiment Analysis	Evaluates emotional tone of text interactions.	Conversation transcript	Yes	AWS Comprehend	Regional, same as Talkative AWS instance	0 days

💡

Talkative is currently in the process of being able to offer EU-based LLMs from OpenAI.

Talkative LLM selection

Talkative allows users to select your LLM in a self-service configuration within the application.

Currently Provided LLM’s

LLM	Provider	Hosted by	Regions	Storage Duration
GPT4.1	OpenAI	OpenAI	US only*	0 days
GPT5 mini	OpenAI	OpenAI	US only*	0 days
Gemini Pro	Google	Google https://cloud.google.com/vertex-ai/generative-ai/docs/data-governance	London (europe-west2), US ( us-east5), Australia (europe-west2), mirroring the Talkative regional instances, except Australia that runs on the London instance	0 days (Google may log prompts to detect potential abuse and violations of its Acceptable Use Policy and Prohibited Use Policy as part of providing generative AI services to customers)
Gemini Flash	Google	Google https://cloud.googavalablele.com/vertex-ai/generative-ai/docs/data-governance	London (europe-west2), US ( us-east5), Australia (australia-southeast-1), mirroring the Talkative regional instances	0 days (Google may log prompts to detect potential abuse and violations of its Acceptable Use Policy and Prohibited Use Policy as part of providing generative AI services to customers)
Claude Sonnet	Anthropic	AWS Bedrock	London for EU and AU Talkative instances, US for US Talkative instance, mirroring the Talkative regional instances	0 days
Claude Haiku	Anthropic	AWS Bedrock	London for EU and AU Talkative instances, US for US Talkative instance, mirroring the Talkative regional instances	0 days
LLAMA 70B	Meta	AWS Bedrock	London for EU and AU Talkative instances, US for US Talkative instance, mirroring the Talkative regional instances	0 days
LLAMA 8B	Meta	AWS Bedrock	London for EU and AU Talkative instances, US for US Talkative instance, mirroring the Talkative regional instances	0 days
Talkative LLM**	OpenAI	Talkative AWS	Any AWS region	0 days

💡

Talkative is currently in the process of being able to offer EU-based LLMs from OpenAI.

Talkative LLM is an open weights LLM completely hosted and managed by Talkative as a dedicated resource. It is available as a paid add-on, contact us for details.

Voice AI Subprocessors

Voice AI uses three components:

The LLM (see table 1 above) - currently only OpenAI LLMs are supported, we are working to bring in Bedrock LLMs;

Text-to-speech (TTS) provider, OR;

Speech-to-text (STT) provider

The components are listed in the table below. Note that Talkative is actively evaluating alternative STT and TTS providers. Please contact us if you have specific requirements.

Type	Provider	Regions	Data Duration	Notes
TTS	ElevenLabs	United States, Netherlands, Singapore	0 days	https://elevenlabs.io/docs/resources/zero-retention-mode
STT	Deepgram	USA	Audio files not stored, logs stored for 90 days	https://developers.deepgram.com/docs/using-logs-usage
TTS	Google	US, Europe	0 days	https://cloud.google.com/speech-to-text/docs/data-usage-faq
STT	Google	US, Europe	0 days	https://cloud.google.com/speech-to-text/docs/data-usage-faq
S2S	OpenAI	US only*	0 days

Talkative Server Locations

Any data processed by the LLM will also be processed and stored within the main Talkative application, which is hosted in AWS. As of April 2025, the regions with Talkative instances are:

London (eu-west-2)

USA (us-east-2)

Australia (ap-southeast-2)

Germany (eu-central-1)

Where possible, the subprocessors used in the tables above are mapped to the corresponding AWS region. For example, LLAMA within AWS Bedrock is used from the UK region for all users within the London AWS environment.

Other AWS regions are planned. Please contact us for a timeline with these.