Follow new updates and improvements to Synthflow.
October 13th, 2024
More Flexible Agency Structures: Tailor workflows and features to fit your brand and unique requirements with improved customization options.
New Interface Themes & Brand Colors: Personalize the look and feel of your agency by choosing custom themes and brand colors to align with your identity.
Sleek Interface Redesign: Enjoy a more spacious and cleaner look with improved sidebar navigation for a better user experience.
Rearranged Assistant Tabs: The Configure and Prompt tabs are now easier to access, streamlining your assistant setup process.
Table View for Assistants: You can now manage assistants in a table format for better organization.
Simplified Setup: Building assistants is now faster with our intuitive new setup flow.
Our enterprise users can now integrate their own telephony systems with custom SIP trunking, providing greater flexibility and control over voice communications.
Note: SIP Trunking is only compatible with Orchestrator 2.0.
Our HubSpot integration just got a major upgrade, offering even more powerful tools for managing your contacts and workflows:
Automatic Contact Updates: After every call, Synthflow will automatically update your HubSpot Contacts, ensuring your data is always current and accurate without any manual effort.
Build Directly in HubSpot: You can now create and manage workflows seamlessly within HubSpot, making it easier to synchronize your voice interactions with your existing sales and marketing pipelines.
How to Set up HubSpot Integration
Also, check out the quick demo to see how Synthflow and HubSpot work together!
The import subscription feature is available through the update subscription (PUT) endpoint in the Synthflow REST API, allowing you to link an existing Stripe subscription with a subaccount for seamless integration.
Stripe Subscription ID (stripe_subscription_id): Provide the ID of the existing Stripe subscription that you wish to import. This is ideal when a subaccount already has an active Stripe subscription, allowing you to easily associate it with their Synthflow account.
Synthflow Price ID (synthflow_price_id): Specify the Synthflow pricing tier ID that matches the imported Stripe subscription, ensuring the subscription is correctly mapped to the equivalent plan within Synthflow.
By using this PUT endpoint, agencies can efficiently sync and manage existing Stripe subscriptions for subaccounts within Synthflow, streamlining the billing process without needing to recreate subscriptions.
Learn more in our API docs here
We’ve added new endpoints to help you manage subaccounts more efficiently:
Update Subaccount: Make changes to an existing subaccount.
Get Subaccount: Retrieve detailed information about a specific subaccount.
List Subaccounts: Fetch a list of all subaccounts associated with your account.
Stay informed with real-time system performance and reliability insights.
Check it out the status page here
Orchestrator 2.0: We’ve made significant improvements to address previous issues, ensuring smoother, more reliable, and responsive voice interactions.
Crosstalk Reduction: Significant improvements to call clarity and overall quality.
You can now choose GPT-4o Mini, offering a more cost-effective solution with faster performance, perfect for high-speed tasks without compromising quality.
August 25th, 2024
We are thrilled to announce the launch of Synthflow Voice Engine 2.0! This major update brings a host of new features and improvements designed to elevate your AI voice experience like never before.
Cartesia Voice Provider: Add crystal-clear, conversational voices that make every call feel more personal and engaging.
Adjustable Voice Speed: Tailor the voice speed to match customer preferences, making conversations more pleasant.
Warm Call Transfer: Seamlessly transfer calls to human agents without dropping the ball, enhancing customer.
Realistic Background Noises: Choose from office or cafe background sounds to add a layer of authenticity to your calls.
How to enable: Head to Settings → Experimental, then toggle on the Voice Engine 2.0 feature.
We've introduced a new API endpoint that allows you to create custom actions and assign them to your assistants via the API.
This provides greater flexibility and enables you to enhance functionality without leaving your development environment. Be sure to check out our documentation for more details.
We are thrilled to announce our official integration with Cal.com! Synthflow is now also available in Cal.com's app store, allowing your AI phone call agent to seamlessly schedule appointments directly into live calendars with instant notifications. This integration simplifies appointment setting and enhances customer interactions by making the scheduling process more efficient and user-friendly.
To get started, learn how to connect your Synthflow account with Cal.com by following our step-by-step tutorial 👇
Our AI assistant now supports backchanneling, allowing it to keep the conversation flowing naturally. When it hears interjections like "mm-hmm," "oh," "ah," "um," or "uh," it continues speaking, creating a more engaging and human-like dialogue.
To improve scalability and reliability, we've implemented new rate-limiting and queueing features. These updates ensure smoother operations during peak usage periods, helping you maintain a consistent, high-quality experience. The concurrency limits by tier are as follows:
Starter: 10 concurrent calls
Pro: 10 concurrent calls
Growth: 20 concurrent calls
Agency: 50 concurrent calls
Enterprise: 100+ concurrent calls (with priority queuing)
Please note that for non-Enterprise tiers, exceeding the limit will result in a 429 HTTP error, and excess calls will not be processed. This feature currently applies only to outbound calls. If you're using workflows with GHL, HubSpot, Zapier, or other automation tools, we recommend implementing a delay mechanism when batching calls to ensure optimal performance. For custom limits on the Enterprise tier, please contact our sales team.
You now have the option to enable an extra check for voicemails, employing two methods: a time-based detection that identifies voicemail based on call duration, and an ML-based detection that uses machine learning for even more precise results. To enable these features, simply visit the Experimental tab in your dashboard.
You now have the ability to control the assistant's dialogue by specifying words or characters that should not be spoken during conversations, such as special characters like asterisks or dashes.
This feature also helps ensure that all communications meet your brand and regulatory standards.
We are pleased to offer you the option to use your own OpenAI API key with your assistants. This provides greater customization and flexibility, allowing you to manage costs and control usage according to your specific needs.
We've expanded the capabilities of custom actions by adding two new methods: PATCH and PUT. This enhancement allows for more versatile API interactions, enabling you to update and replace resources efficiently within your custom actions.
We've significantly enhanced our HubSpot integration to allow for more precise control over the outcomes of AI phone calls. The Call Status (Busy, Failed, Completed, etc.) and End Call Reason (Voicemail, Human Goodbye, etc.) are now enumerated fields. This change makes it much easier and more reliable to set up workflows in HubSpot that branch based on these specific call outcomes.
With this improvement, you can effortlessly create conditional logic in your HubSpot workflows to trigger specific actions—like sending a follow-up email or scheduling a callback—based on how each call concludes. This level of granularity empowers you to tailor your customer engagement strategies more effectively.
Check out our tutorial on how to utilize this feature to effectively branch the outcomes of AI phone calls.
Import Custom Actions to Subaccounts: Agencies can now import their custom actions to their subaccounts. This highly requested feature allows for easier management and deployment of custom functionalities across all your client accounts, saving valuable time and ensuring consistency across your services.
Enable HubSpot Integration for Subaccounts: Agencies now have the option to enable the HubSpot integration for their subaccounts. This means you can provide a seamless CRM experience across all levels of your organization, ensuring that every team member has access to the tools they need for effective customer relationship management.
Our users often tell us how much they appreciate the quality and human-like feel of our AI voice assistants. We're thrilled to share that we've been featured in ElevenLabs' recent blog post! This deep dive explores how Synthflow AI has enabled businesses to achieve unbelievable ROI through our no-code AI voice solutions.
It's a testament to our close collaboration with ElevenLabs in delivering incredible results together. Check out the full article.
We’re excited to let you know that Synthflow Academy is now live! It’s your new hub for learning all about Synthflow’s AI voice technology. Whether you’re just starting out or looking to level up your skills, we’ve got you covered.
What You’ll Find at Synthflow Academy:
Learn at Your Own Pace: Our self-paced courses let you dive in at your own speed, from basics to advanced features.
Monthly Webinars: Join live sessions with our founding team for deeper insights, sneak peeks, and more.
Community Live Builds: Get hands-on with interactive sessions where you can build and deploy AI voice projects with fellow community members.
Ready to start? Join Synthflow Academy today and take your skills to the next level!
Recent fixes include:
Reset Password Flow: We've resolved an issue where the reset password flow didn't work in some cases for subaccount users. It now functions seamlessly for all users.
Improved Custom Action Prompt Editor: Some users reported that writing custom action prompts was laggy and slow. We've optimized the user experience to ensure that the cursor remains responsive, making the editing process smoother than ever.
Are you ready to shape the future of AI with cutting-edge innovations? At Synthflow, we don't just follow trends; we create them. We're on the hunt for passionate, creative, and forward-thinking individuals to join our dynamic team. If you're driven by innovation and thrive in a fast-paced, collaborative environment, we want you!
Ready to make a difference? Visit our Careers Page to view open positions and submit your application. Join Synthflow, and be part of a team that's transforming the tech landscape.
August 25th, 2024
We've added a new feature that allows you to enable background music during call transfers. You can now choose from a selection of music options or input your own .mp3 URL to play custom background music.
The transfer target (human agent) will hear the music along with a custom message that you can set before the music starts.
Our REST API has been upgraded with 12 new endpoints. These new endpoints include creating actions, updating actions, updating assistants and more. Moreover, you can now pass pricing tiers to attach to subaccounts.
To simplify your setup, we've updated the API docs and released new Postman collections for easy integration.
We're excited to announce that our Zapier app v3.0 is now available! This update includes over 20 new actions, giving you even more ways to automate and streamline your workflows.
Synthflow's official Zapier app.
Our HubSpot app is officially live in HubSpot’s marketplace! You can now seamlessly integrate Synthflow with HubSpot to enhance your CRM capabilities.
Synthflow's official Hubspot app.
We've made our app's actions logic more flexible. Actions can now be created and managed in a separate Actions tab without needing to attach them to an assistant. These actions can be reused across multiple assistants, saving you time and effort.
You can now specify the maximum duration of a call and set reminders for idle users. Customize reminders to keep the conversation flowing smoothly.
We've added reusable Calcom integration. Once authenticated, Calcom actions can be easily created on the platform or through our API.
We’ve upgraded to the new GPT-4.0 model as our default. Now, everyone can benefit from the fastest and smartest model in the market, ensuring maximum speed and highest intelligence than ever before.
August 1st, 2024
We are rolling out the closed Beta of our stage-based prompting approach. This feature allows you to easily import your company information from your website and define conversational stages. The assistant will recognize the stage of the conversation and adjust its behavior accordingly. If you’re interested in joining the early testers, let us know!
You can now set up custom actions with different CRMs at the click of a button. We’ve added two new Hubspot actions:
Get user preferences from Hubspot
Get product from Hubspot
More custom actions with additional CRMs are on the way.
Synthflow can now be easily connected with Hubspot. Our official app will soon be available, enabling users to trigger calls and more directly from Hubspot.
We have further improved the quick setup process, making it even easier to set up your voice assistant and connect them with a CRM in just a few clicks.
We’ve introduced multiple info extractor templates. Setting up this action is now just one click away.
You can now attach tags to contacts during creation. This includes:
Full CRUD (Create, Read, Update, Delete) capabilities for tags,
Filtering contacts by tag, such as for running batch processes.
When importing contacts from your CRM, tags will be imported as well.
You can now run multiple batch campaigns for any outbound assistant. These campaigns can be run directly on the contacts you’ve imported from your CRM or CSV.
Conversation logs for contacts imported from GHL will now automatically sync with GHL and be visible in your GHL portal.
By logging the calls, you ensure that every interaction is recorded, providing a comprehensive view of customer engagement in both Synthflow and GHL.
You can now define the body of POST API calls.
We have laid the groundwork to include more variables based on Contacts, distinguishing between those retrieved by the AI during conversations (formerly known as “user inputs”) and those populated behind the scenes (e.g., the user’s phone number).
Variables can now be easily included in query parameters (for GET API calls) and the body (for POST API calls) by typing “<”.
You can reference exact custom action names in the assistant prompt by typing “#”, making agents more effective in using them.
The UI/UX of the custom actions editor has been revamped.
Fixed the English message during the appointment booking action. If the agent speaks German, the message will now be in German.
Automatically creates contacts from inbound calls.
Ability to edit contacts (except those imported via GHL).
Improved UX for transfer calls. Instead of silence assistant will say messages during the transfer like "I'm still trying to reach the human agent. Just a moment" and "Just a few more seconds."
The get availability api from GHL returning the slots from the next day together with the slots from the selected day.
Old custom variables in the prompt conflicting with custom action.
New modular actions logic
API endpoints for actions, subaccount pricing tiers, and more
Voice orchestration 2.0- higher reliability, new features such as background ambient sounds, fade-outs during interruptions and more.
More action templates and integrations
Synthflow Academy
In this video you'll learn how to extract key information from your AI voice calls using the Information Extractor module built into Synthflow.
Are you ready to shape the future of AI with cutting-edge innovations? At Synthflow, we don't just follow trends; we create them. We're on the hunt for passionate, creative, and forward-thinking individuals to join our dynamic team. If you're driven by innovation and thrive in a fast-paced, collaborative environment, we want you!
Ready to make a difference? Visit our Careers Page to view open positions and submit your application. Join Synthflow, and be part of a team that's transforming the tech landscape.
July 5th, 2024
We are excited to announce that we've successfully closed a $7.4M seed round led by Singular VC, with strong support from Atlantic Labs and new backer BigStory VC. This funding will help us innovate further and expand our offerings.
Join Our Team! We are growing and hiring for roles in Customer Success, Customer Support, Marketing, and Engineering. Explore our opportunities here.
Read More on TechCrunch Check out our latest feature on TechCrunch to learn more about our journey and vision: Synthflow picks up $7.4M for no-code voice assistance for SMEs.
Setting up your assistants has never been easier! You can now set up your assistants in just a few simple steps by providing the link to your website. Our system will scrape the information and construct the prompt for you. Currently, the Create AI Inbound Receptionist template is available, with more on the way.
We’ve introduced a comprehensive timeline for your contact events, including:
Call timestamps (start, end, and actions)
Actions like Booking, SMS, Call Transfer, and Voicemail
Call outcomes such as Busy, Voicemail, etc.
Action outcomes like Appointments Booked, Data Extracted, etc.
This timeline provides a clear view of all interactions and outcomes, enhancing your ability to manage and review call activities effectively.
We’ve made significant improvements to the custom actions UX:
Variables from custom actions can now be referenced in the assistant prompt.
Users can set up variables in the parameters section more intuitively.
Complex API outputs (e.g., lists) can now be used in prompts.
For more details, check out our latest webinar recording here.
Our AI assistants now have time awareness, including the current date, time, and timezone. This means they can inform callers about office hours and provide better assistance.
We’ve addressed several bugs to improve your experience:
Whitelabel Issues:
Agency logo display issue fixed.
Subaccount access issues resolved.
Reset password flow is now functional.
Resolved errors caused by ad blockers and certain browsers (Arc/Brave).
GHL Integration:
Fixed issue with fields containing spaces not being retrieved by custom actions.
In this tutorial you'll learn how to create an AI Voice Assistant that can take inbound customer calls for a Car Dealership. We'll run through the prompt, configuration, info extraction and call forwarding to a human! Enjoy.
June 1st, 2024
Centralized Contact Management: Manage all your contacts directly within Synthflow.
Conversation History: Easily view all conversations involving a specific contact, whether they made the call or received it.
Import from CRM: Quickly import contacts from GHL into Synthflow.
You can now choose to perform actions before a call starts, rather than during the call.
We've always let you create custom data types to format results better. Now, there's a new popup that shows examples of each data type to help you choose the right one.
We've reorganized the API call settings into four easy-to-navigate tabs: Endpoint, Headers, Query Parameters, and Authentication.
User inputs are automatically added to the “Query Parameters” tab.
You can now connect or disconnect your GHL account directly from the Authentication tab.
After you start an API call, you'll see all the outputs as variables that you can use in your prompts. For example, you can set up a prompt like: “If {product_price} is over 100 USD, ask if the user wants to pay with a credit card.” The assistant will use the {product_price} variable from the API to decide how to ask about the payment method.
We released a new official Bubble plugin that allows users to easily integrate AI-powered voice assistants into their Bubble apps. This plugin enables businesses to create intelligent, conversational interfaces for customer interactions, handling tasks like appointment booking, product inquiries, and general customer support.
Read blog article
Documentation
We launched an official integration with Make, allowing users to easily incorporate AI phone calling into your Make automation workflows.
You can now fine-tune the emotional tone and pacing of the speech of your AI Assistants:
Subtle control over voice intonation and emotion
Allows for more nuanced and context-appropriate voice responses
Customizable through text-based prompts (e.g., "he said, confused")
New native payments management system for subaccounts:
Automated monthly billing based on minute usage
New dashboard tab for monitoring subaccount usage and billing
A new Playground feature has been added for testing and refining AI assistants:
Text-based chat interface for simulating conversations
Ability to edit and refine AI responses for improved accuracy
Option to save and review multiple conversation outcomes
Integrated with Actions for comprehensive testing
We've upgraded our referral program:
Increased affiliate referral commission to 30%
Improved tracking and reporting of referrals
15-month limit on commissions
applicable to all tiers
We've expanded our language support to include:
Italian
Dutch
Error handling for timezone issues
Comprehensive actions analytics
Improvements in VoiceTech, reducing latency and enhancing call quality
Bugfixes
May 31st, 2024
Our AI assistants are now multilingual! They can fluently communicate in Spanish, German, Portuguese, and French, in addition to English.
This update allows you to engage with a global audience and provide personalized experiences in users' preferred languages.
We have also added corresponding voices for each language to ensure authentic and natural-sounding interactions.
Your AI assistants can now automatically trigger custom API actions during phone calls.
For instance, at the start of a call, it can fetch data about the incoming caller from an external API like GHL. This process occurs instantly and without manual intervention, providing relevant information such as contact details to the agent right away. As the call progresses, the system can further tailor the interaction based on continued API interactions, ensuring the agent has all necessary context, like a caller’s specific interest in properties in New York rather than Berlin.
The new KPIs dashboard offers advanced tracking and analysis features for monitoring AI assistant performance. It provides detailed statistics on production calls, including status breakdowns and usage time.
Additionally, it enables agency users to view aggregated feedback from subaccount users and offers a comprehensive analysis of call quality, helping to assess the assistant's effectiveness in interactions
Synthesizer upgrade to the newest ElevenLabs Turbo v2.5 model resulting significant improvements in latency and stability.
Crosstalk improvements and parallel processing, ensuring faster response times.
You can now easily manage, purchase or import phone numbers directly from the Synthflow Dashboard.
We have streamlined the compliance process for phone numbers within the dashboard, allowing you to complete all the necessary steps to meet local regulations.
Allocate numbers to different subaccounts effortlessly.
Set trigger conditions for RAG search to enable efficient source location and reduce latency.
For instance, configuring a trigger for queries such as "user asks about city property count" directs the search to focus on property-related databases, optimizing efficiency and speed.
Allows you to integrate Zoom with the Real Time Booking action. Once a booking is confirmed, it automatically schedules a corresponding Zoom meeting.
You now have control over how long the AI assistant will stay on the phone during periods of silence.
Auto-populate "Test Number & Name" for convenient testing.
Create sub-accounts via API with API key in response for seamless demo assistant creation.
May 18th, 2024
Allows multiple in-call transfers to other AI assistants using only one phone number. This offers a much smoother interface compared to traditional IVR systems, making it much easier for customers to navigate to the desired department by simply stating their request.
For example, they can say "I'm calling for the tech department" to be instantly connected to the tech assistant or "I want to talk to sales" to be seamlessly transferred to the sales assistant.
You can now access advanced options and further fine-tune the voice output to better match your specific needs or preferences:
Synthesizer: The model to synthesize voice.
Stability: Modulate the stability of the voice to allow for a wider or narrower emotional range.
Similarity: Set how closely the AI-generated voice should resemble the target or original voice.
Style Exaggeration: Adjust the degree to which certain stylistic elements of the voice are emphasized, making the voice output more unique and suited to particular contexts or contents.
Seamless Transfer: Quickly reroute your current call to a specified number or directly to an internal extension with minimal interruption.
Extension Digits: Utilize extension digits to connect the call to a specific department or individual within larger organizations. This bypasses the need for a manual directory or additional operator assistance.
Customizable Experience: Set specific instructions and messages that the app will use during the transfer process. For example, you can customize messages that the assistant will say when trying to reach the transfer target, when the target answers, and if the target does not answer.
Controlled Timing: Decide how long the app should wait for the transfer target to answer before it ends the attempt. This helps in managing time more effectively during urgent situations.
SMS Confirmation: You can now configure SMS notifications post-booking, including customizable text that can incorporate variables like the appointment date and link.
Google Meet URL: If you use Google Calendar, the app can now retrieve and include the Google Meet URL directly in your calendar event.
More Webhook Information: Post-booking, the app will now provide additional details via webhook about the booked meeting.
Allows control over a delay in seconds before the assistant will answer back.
This setting allows you to customize the delay, in seconds, before the AI assistant responds during outbound calls. By setting this delay, you can ensure a more natural flow in conversations, giving you control over when the assistant begins interacting.
Subaccounts can now acquire their own phone numbers directly through Synthflow. Thus, they can use the the platform even if they don't have their own Twilio account. You can enable this feature in the agency -> permissions settings.
You can now use the “email address” parameter to invite users to a subaccount. When you fill in this optional parameter, a “sign in” link will be generated and sent back in the response. The invited user can use this link to sign in and set up their account. For more details, read the documentation.
May 9th, 2024
You can now customize your assistant's knowledge base according to your specific needs and domain by feeding it with various types of data, including raw text, PDFs, and specific web pages. This allows you to expand your assistant's knowledge base and provide more accurate and relevant responses to user queries.
Once the data is ingested, our RAG (Retrieval-Augmented Generation) feature enables your AI assistant to retrieve and utilize this information during conversations. By leveraging the ingested data, your assistant can provide more informed and context-aware responses.
Llama 3 is now available, offering lower latency (up to 3x faster) and compatibility with more powerful synthesizers.
The new synthesizer is more accurate, natural, stable, and supports multiple languages (we will introduce new languages with the next release).
Stability and Diversity: Excellent stability, language diversity, and accuracy in voice cloning and accents.
Default Synthesizer: This is now the default synthesizer for all assistants using Synthflow LLM and Llama 3.
Agencies can now resell new products - Assistants, Teams, and Flows
New Pricing Tier Editor
More Granular Subaccount Controls
LLM as judge: Our fine-tuned LLMs now generate high-quality feedback to assess the success of AI-customer interactions based on various criteria.
Why is it Useful?
● Great feedback for prompt optimization
● Monitor quality of customer calls
End Call Reason: The new end_call_reason parameter provides insight into why a call ended.
Echo & Noise Cancellation Improvements
Crosstalk Improvements
Speech-to-Text Improvements: Leveraging the latest Deepgram version.
Voicemail Detection Enhancements
Incomplete Sentences: Fixed the issue with assistant saying an incomplete sentence but providing the full one in transcript.
We’ve lowered our pay-as-you-go prices to just 8 cents per minute. You can now enjoy our product at a lower cost, and as an agency, you can now maximize your margins when reselling assistants. Plus, you can now add additional minutes while in trial mode.
Specify the number of seconds before the agent picks up.
Infrastructure upgrade for improved performance and scalability.
More assistant templates.
Improvements to calcom booking format
New integrations tab