Congratulations on making the decision to invest in Conversational Intelligence. Choosing to offer Artificial Intelligence (AI) driven collaboration capabilities within your product may have been an easy decision, but here comes the tricky part: understanding the best solution for your business.

While cost will likely be a factor in your decision making, there are other not-so-obvious questions you should ask before proceeding. These questions will help you understand whether a horizontal platform is right for you, what features you’ll want to prioritize, what capabilities you need to future proof your solution and more.

1. Do you need real-time transcription or closed captioning in your conversational intelligence solution?

Real-time transcription is a game-changer for many of our customers. Human recall isn’t perfect and attendance and focus can’t be guaranteed. You’ll want to confirm your solution is flexible if your transcription needs expand in the future. Symbl transcription supports 71 supported languages in 127 locales. Please note: while we transcribe all these languages, we do not translate.

2. Can you monetize and generate value from existing data that your customers have already generated in the form of call recordings or existing call transcripts?

Delivering added value to the post-meeting experience and making it easy for your users to navigate meeting recording content could be the differentiator for your business. With Symbl, you can add an x-ray like feature to your call or meeting recordings. Want an index of contextual topics along with the call recording? No problem. We offer extensible APIs that let you configure and extend the post-meeting summary and analysis for delivery to your user’s inbox for quick action, or other work tools like Slack. The call recording can be searched based on action items, follow ups, and questions asked. AI-powered conversational intelligence turns your recordings into action to keep the teams focused on what matters most. Long-term benefit: We surface topics as search hints to find other meetings where a particular topic was also raised. And simple custom code helps your customers navigate videos and transcripts in a systematic manner.

3. Do you need a finished product or do you want the flexibility to add capabilities to your own offering that analyzes employee and customer conversations and generates automatic, real-time actions?

We designed our APIs so a developer can quickly and easily embed the functionality into a company’s own offering without the complexity of building the tech infrastructure behind it. We provide sample code, step by step guides, a thriving developer community for questions, and how-tos to fully support your technical resources in your implementation. And with our stellar support team, your team does not troubleshoot alone. There is power in enhancing your own product with native capabilities of passive embedded intelligence all connected via the same experience.

4. What type of content do you need to analyze?

Symbl is a programmable platform with a comprehensive suite of APIs to analyze any natural voice or text conversation. Our conversational intelligence APIs unlock value using proprietary machine learning algorithms that analyze natural human conversation without the need to gather and annotate training data, and create classifiers from the ground up. Symbl’s APIs can analyze any emails, texts, or human conversation contextually and provide the ability to carry the context across conversations for aggregated analysis. The use cases we support:

  • Real-time streaming audio in applications built over WebRTC
  • Audio data from your telephony systems such as PSTN or SIP
  • Recorded files from call centers, podcasts, or meetings
  • Text data in the form of transcripts, emails, social conversations

We are not a document understanding system and recommend you pursue another solution if your needs extend beyond natural human conversations such as contractual review.

5. What should you consider when weighing whether to build versus buy?

We’ve all heard the horror stories of companies who try to build an in-house solution and years after project kickoff, abandon the project and buy. Unless you have an unallocated data science team and years before you need conversational intelligence, investing in a company with a proven track record gets your company value in days not years. We did the hard work so you can focus on identifying the best user experience for your customers, decide what business model you wish to address, and select what customer segment to use it with first to foster upsells and increase engagement for your applications.

6. Does the solution you are vetting start with training data?

Most AI systems start with gathering the right data set. Over millions of minutes of conversations and the validation across thousands of users, our proprietary algorithms recognize what insights matter most in a workplace conversation – summary topics, action items, questions, ideas, requests, and follow-ups. We built multiple models that look for specific patterns in the conversation data that align dynamically with any type of conversation. If your solution requires some other type of insight detection that does not fall under the out-of-the-box insights we provide, you might need to start with training data. How much training data is necessary will depend on the system you are building internally and the type of insight you are generating. You can book a call with our AI specialists to solve this use case for your business.

7. Do you want your solution to depend on a ‘wake’ word?

Many virtual assistants require a trigger word before activating such as ‘Alexa, Siri, OK Google‘ to act when commanded to do so. We provide passive and embedded intelligence in your existing digital tools which act as a recommendation system for insights and actions. We do not build chatbot-like interfaces but recommend products like Google Dialogflow or RASA if your needs include a chatbot.

8. How important is cost today and as you scale? Do you have the right ways to reduce cost as you scale for Automatic Speech Recognition ASR and beyond?

Optimizing infrastructure and resources is one way to ensure cost savings. If developing a conversational intelligence system is the mainstream IP for your company, you might want to think about which tool stack can be used off the shelf and which parts you need to develop in-house. Working with the right partners can enable you to save on cost above what you would be able to save independently due to the limit of your own volume needs.

9. Do you know the right ASR vendor for you and do you want to manage / benchmark ASR?

The right ASR partner will depend on the need for specific ontology, the type of audio data, CODEC, speaker events, and supported dialects, as well as the level of customization needed. We benchmark all existing providers on a monthly basis to determine the best fit with our default offerings and can suggest best practices based on your specific use cases.

10. Do you want to invest in creating value beyond transcription? 

There are multiple transcription APIs that you can integrate into a business, but the true value is the ability to use the text generated to understand and analyze conversations to generate value and save time. Symbl offers a comprehensive API to integrate over real-time and Async channels so you don’t need to build your own interfaces. We also offer our AI text correction layer that improves transcription accuracy and lets you enable additional AI-powered conversation insights without training and with plug and play pre-built experiences.

11. What kind of insights and intents do you want to support in your conversation workflow?

You can classify bulk calls based on sentiment and certain keywords. The accuracy of the resulting output depends on the ability of the system to understand the context. Most systems work on domain-specific training and have limitations around keywords. If accuracy is important, you’ll want to consider a solution with contextual insights that go beyond keyword identification.

12. What is the on-going conversational intelligence roadmap and how is it aligned to your other product offerings or other AI capabilities across the company?

Amidst competing business priorities, it is useful to decide if building a team of data scientists to solve the conversation intelligence problem is part of the long term strategy and how it fits in the overall product roadmap of the company. Some questions to answer are: is there more than one product that can leverage the intelligence, do you already have a team of data scientists, or do you want to focus on building the experience and domain intelligence for your users and use off the shelf APIs for the backend.

13. What is the best experience for your users – part of an existing workflow in product or as part of a new experience or a marketplace add-on?

This is important to consider if you are sensitive to your brand dilution and want to own a native and integrated experience for your users in enabling AI for conversations. Some businesses that do not have a solution for native integrations, end up opening a marketplace of providers for value-added services which end up costing more than the base subscriptions. Business should look at conversational intelligence as an opportunity to upsell and increase adoption and engagement with their applications which can only happen if you are enabling the intelligence as part of an existing workflow that users are already familiar with. It takes time to create a new habit which might not be the best place to start if you want to increase engagement. This is a wise principle to keep in mind with any AI/ML capabilities where user engagement and experience is the key to value generation. Digital transformation is hard. We’ve published a series of blog posts on some of the most common problems we help you address. Recorded Meetings Sales Enablement Pay for Performance Customer Care