Symbl builds understanding and generative AI that comprehends human conversations, giving machines the ability to understand communications better than humans. Our platform, with state-of-the-art AI and generative AI models, enables businesses to analyze multi-party conversations with AI capabilities in both real-time and asynchronously.

The Symbl.ai platform provides generative AI capabilities for conversations with scale and security. Through the platform, you can extract actionable insights from unstructured human conversations across multiple formats – audio, video, and text. You can also perform generative tasks such as creating summary reports, drafting emails, identifying and recommending resolutions to customer issues among many others.

The Generative Artificial Intelligence technology is powered by Nebula, a large language model (LLM). By providing a conversation transcript along with an instruction or question as an input, Nebula can generate responses and perform tasks pertaining to the conversation as an output.

Depending on the instructions and questions used in prompts and the type of conversation, Symbl can be used for a variety of use cases. Common use cases include generating meeting notes, providing real time assist, ensuring compliance, evaluating call scores and much more.

Symbl has a wide range of users including AI / ML engineers, business decision makers, product teams, sales reps, support agents, and compliance experts and more. Symbl’s large language model and conversation understanding features can be applied across multiple domains and use cases wherever conversation is the primary mode of interaction.

We provide two (2) options to interact with the platform and Nebula – using the ‘Playground’ or through APIs. The Playground allows you to test the model and features without writing code, while the APIs enables you to interact with the model programmatically. Additionally, to use the Nebula embedding model, you can access the Embedding API. Learn more here.

Nebula is Symbl’s large language model for understanding human conversations. Nebula’s architecture and training are tailored for conversation-centric tasks.

Nebula can be used for a number of generative tasks such as creating summaries, follow-up questions, drafting emails, issues to review, qualify sales leads, identify and recommend resolutions to customer issues etc, among others.

Nebula is based on the Transformer architecture and its attention
mechanism, which allows it to weigh the importance of different words in the given
conversation.

Nebula has been pre-trained on a filtered, deduplicated high-quality dataset consisting of 1T tokens and fine-tuned on a dataset curated from 100K+ conversations from the business domain, representing a multitude of conversation tasks.

Nebula Embedding API is optimized for calculating vector representations of conversation data. Using this, you can build custom search and index experiences over your own business specific data.

You can interact with Nebula using the Model Playground or the Model API. The Playground allows you to test the model with conversations and tasks without writing code, while the Model API enables you to interact with the model programmatically. Sign up for access to the Nebula Playground here.

Symbl’s REST API portfolio contains ‘Generative’ APIs (e.g. Call Score, Summary) that can generate an output from an uploaded conversation and ‘Understanding’ APIs (e.g. Transcription, Entities, Sentiment, Redaction, Tracker etc.) that can provide specialized conversation intelligence.

You can send conversation data to Symbl.ai in real-time via Streaming API or Telephony API and asynchronously using Async API. After processing a conversation, you can use specialized conversation intelligence features such as ‘Speech to Text’, ‘Topics’, ‘Trackers’, and ‘Action Items’ or generate a customizable UI.

You can use our Text API for transcripts generated by your own ASR. We also work with a variety of ASR partners if you are looking for a customized solution. Find our complete list of partners here.

Evaluating the results from various speech recognition vendors is a challenging task. To help make this process easier, we put together a utility, open-sourced on GitHub that will enable you to evaluate speech recognition vendors and results faster and goes beyond just Word Error Rate (WER) which you might see elsewhere. The utility automatically performs the pre-processing or normalization of the text to remove further manual efforts for the evaluation process.

Symbl.ai has best in class speech-to-text available across all channels. If a company claims a precise answer for out of the box speech recognition accuracy, here are a few considerations to keep in mind: 1. The way companies measure accuracy differs; 2. Custom models outperform standard, so the question to ask is what level of accuracy do we need to be successful; 3. The key differentiator when it comes to transcription is the presence of an intelligence layer. Symbl.ai provides a proven intelligence layer that transforms transcription into key insights and next best actions.

Yes, Symbl does integrate with a whole range of enterprise applications. Full details are available here.

We are hosted in the public cloud and have the flexibility to scale as you grow.

Yes, Symbl supports various deployment options based on your specific requirements. To learn more, you can contact us to discuss further.

The default deployment option for Nebula is on the public cloud. However, on-prem deployment is available too.

We store transcripts, insights, and metadata about the conversation, which is persisted at Symbl.ai’s end. All data access is logged and audited regularly. Logs are maintained for a minimum of 3 months and encrypted at rest. By default, we do not store audio recordings. In cases where storing is required, audio recordings are securely stored and encrypted at rest using 2048 bit RSA. Please visit our security page to learn more.

Symbl.ai collects and processes only such Personal Data as is adequate, relevant, and limited in scope to the requirement and for a length of time that is necessary for the stated purposes of its use. Please visit our privacy policy to learn more.

All data on the Symbl.ai platform is encrypted and all connections are secured using 2048 bit AES encryption. Over the wire, data is encrypted using RSA 2048 bit keys. At rest, data is encrypted using AES-256. Please visit our security page to learn more.

No. Client conversation data is strictly confidential and not used to train Nebula.

Symbl places utmost importance on data security and privacy and is Certified SOC 2 Type II compliant (full trust report). For a complete list of data security and privacy compliance certifications, please refer to our security page.

By default ingest conversations and processed intelligence are retained with Symbl.ai. No audio recordings or voice streams are stored by Symbl.ai, and are automatically purged with processing. To be able to have control on the data retained, we provide a Delete API that enables developers to programmatically delete conversations by conversation ID. In the dedicated private hosting options, businesses can choose custom data retention options for which you can contact our sales.

Symbl provides tiered support to customers with different service level agreements (SLAs) based on the support plan that customers are enrolled in, For more information, please refer to our ‘Support’ page

Absolutely! We are happy to be of any help and supporting relevant organizations, with discounts and/or free credits. Contact us to learn more

We offer public deployment by default on cloud. We are hosted in GCP. However, for enterprise-scale deployments, we also support dedicated private tenants as well as VPC tunneling to secure and isolate the flow of data as well as to enable customer configurations and capabilities.

Concurrent connections applies to the numbers of concurrent streams that can be processed in real time or the numbers of jobs for asynchronous processing for batch mode. We currently support 5 concurrent connections in a free plan. If you are looking to increase concurrency for your implementations, please contact our sales team for high scale deployment options.