Everything You Need to Know About AI Agents

Download Whitepaper

Platform Capabilities

Explore the capabilities of our cognitive architecture and intelligent communication fabric – built to facilitate an agentic approach to automating workflows.

CONTACT US

Native Channel Integrations

Consistent and reliable communication across channels: Shared library with 700+ steps, including deep channel integrations so you can take advantage of native channel functionalities where it matters most.

Voice

Augment, iteratively rebuild, or replace legacy voice solutions in weeks, not years with our native voice stack, including Session Border Controllers and direct carrier connectivity with full support of SIP, WebRTC, and the PSTN.

Native Voice Stack Features:

  • Inbound Calling: Create self-service IVR portals. 
  • Outbound Calling: Single calls, bulk outreach, and scheduled campaigns.
  • Visual IVR: Visuals to supplement voice experiences. This can be done over web chat, mobile apps, SMS, and other messaging channels.
  • AI Powered IVR: Use customer data, contextual history, and advanced technologies together to create exceptional experiences.
  • Public and Private Routing: Run traffic over the PSTN or via SIP using the public internet or private networks.
  • Integrations with Telephony Providers: Natively integrate with your existing telephony stack using private networking, high quality codecs, and cheaper rates. Traffic can be passed via SIP refer, SIP transfer, PSTN transfer, etc.
  • Carrier Relationships: We have direct carrier relationships so we can provide PSTN connectivity out of the box.
  • Handle Background Noise: Our acoustic models are trained to perform in a variety of scenarios — including in noisy environments.
  • Barge In: Control for voice or DTMF if a user has to listen to a prompt before allowing for interaction.
  • Call Authentication: Detect fraudulent calls and identify phone numbers (mobile, landline, etc.)
  • Recording, Messages, and Voicemail: Record calls, specific call segments, and capture voicemails.
  • Phone Number Management: Buy, manage, and oversee phone numbers easily. Choose from International and/domestic numbers (Long code/DID, Short code, Toll-free) number, and have the option for number pooling.
  • Call Transcriptions: Turn voice into text for analysis, audibility, and summarization.
  • Call Analytics: Monitor connectivity, QoS, jitter, latency, deliverability, errors, performance, call flow, percent of calls per flow, mobile vs. landline recognition, CallerID lookup, call traffic, and outcomes.
  • Connect with Live Agent and Human in the Loop: Allow options to send calls to automation, voicemail or live agents to answer.

Text to Speech and Speech to Text

  • Automated Speech Recognition (ASR): Speech recognition enabled with Natural Language Understanding (NLU) to support natural conversations with users.
  • Generic and custom STT Models: We work with many vendorsto utilize and create custom STT models based on industry, domain, dialect, and function. Customizations can also be made to optimize extraction of different data types and dynamic grammars (names, passwords, etc.).
    • Amalgamate multiple STT solutions to get the best performing engine on an utterance by utterance basis
    • Use both cloud-based services (e.g. Google, Microsoft (MSFT), Nuance, etc.) as well as on-prem services (e.g. Nuance, Deepgram, Speechmatics, Lumenvox, etc.) which can help to ensure data privacy and support custom models
    • Live stream audio to STT vendors or the amalgamation engine, as opposed to relying on batch methods. This creates 1) reduced latency, 2) multi-threading (multiple STT vendors simultaneously), and 3) greater control.
    • STT control:  Native control over STT capabilities (as opposed to simply an API connection due to our native voice stack), which is where most of the STT capabilities are enhanced. For example: background noise reduction and interspeech timeouts. 
  • Text to speech (TTS): Create natural, high-quality custom voice prompts to use throughout interactions. Choose from 150+ languages, and dozens of voice options.
    • Custom TTS models via our partner network: These custom voices can be easily plugged into our platform and used just like any other standard voice integration, including full control over SSML as well as voice preview.  The TTS system used can vary by individual caller, topic, time of day, etc. This is often part of a hyper-personalized design approach.
    • TTS Caching: Our models store audio in cache from previous conversations, reducing the cost of processing with TTS vendors.
  • No code access to Speech Synthesis Markup Language (SSML): Control the speed, pitch, inflection, pronunciation, and add pauses to voice prompts. Test prompts as you write them in the builder.
  • Accents and Dialects: Advanced acoustic models and language models to account for voice variation across nationalities, regions, and dialects.

Consumer messaging

Consumer messaging features

  • Inbound and outbound messaging: Send and receive messages, including bulk sending and scheduled sending.
  • Deep Channel integrations: Robust support for any messaging platform utilizing the standard services and specific capabilities of each. OneReach.ai acts as the orchestration layer to each platform using each system’s unique capabilities: 
    • WhatsApp
    • FaceBook Messenger
    • Twitter 
    • Telegram
    • Viber & Viber Business
    • WeChat
    • Apple Business Chat
    • Rich Communication Services (RCS)
    • Google Business Messages
    • Kik
  • Text & Files types support: Send messages with different file types (e.g. images and videos) 
  • Random messaging versions and variant testing: Use message variations to seek and reveal optimizations. 
  • Timeouts and Error Handling: Easily account for errors and timeouts based on specific criteria. For more see “Conversational Designer and Dialog Management.”

Security and privacy

  • Voice Biometrics and cognitive integrations: Use a person’s voice as a uniquely identifying biological characteristic for authentication. 
  • Private Networking: Bypass the public internet to improve privacy and security.
  • Engage with End-users: Using the messaging apps they already know and love by using deep channel integrations for systems like WhatsApp, Facebook Messenger, Google Business Messages, Viber, email, web chat, etc.

Enterprise collaboration

Predefined channel connectors for popular enterprise systems (Slack, Teams, Zoom, Email, ServiceNow, Salesforce, etc.) as well as the ability to create your own connectors.

Enterprise tools

  • Orchestration: Amplify the value of your existing investments in enterprise tools by using OneReach.ai as orchestration layer. 
  • Leverage integrations with tools like: Trello, ClickTime, Monday.com, Jira, ServiceNow, Confluence, Formstack, GitLab, Slack, EWPS, Authorize.net, Zoom, MS Azure, Okta, MS Outlook, Basecamp, Salesforce, Bonusly, Teamwork, Stripe, Toggl, Todoist, Wrike, Survey Monkey, Eventbrite, Meetup, Expensify, Harvest, SmartSheet, JotForm, Typeform, Yammer, Scoro, ActiveCampaign, BigCommerce, Envoy, Pivotal Tracker, Time Doctor, Airtable, Dropbox, Asana, Zendesk, WordPress, iCalendar, Hubstaff, Instagram, Google Spreadsheet, Alexa, IMAP, WhatsApp, Facebook, Google Actions, Telegram, Twitter, Viber, Calendly, MS Teams, and many more*
  • *While these are some of our pre-built connectors, we also make it easy for our customers to build custom Steps and APIs. See “Architecture and Integration” for more about our capabilities around custom APIs

Slack and teams

  • Inbound and outbound messaging: Automate and manage conversations in direct messages and channels (public and private), including use of threads. Update and delete messages.
  • Rich Messaging Formatting: Take advantage of Slack and Team’s full interaction capabilities by designing with: Interactive & form components in messages and app surfaces, slash commands, reactions (emojis), snippets, posts, and ephemeral messages.
  • Manage channels and members: Add, remove, view channel members.
  • Search: Search conversation history, users, and files.
  • Manage Files: Upload and download files, snippets, and posts.
  • HITL UI: Use Slack for Human In the Loop capability so that automation can extend to a Slack user.

Teams and slack synchronization

  • Synchronize Channels: Create a single chat session that can exist in Teams and Slack at the same time allowing each user to use their preferred system.

SMS

Direct carrier and provider relationships allow you to leverage the most popular messaging protocol in the world across all number types (10 digit long codes, short codes, toll-free numbers, and alphanumeric sender ID)

 

  • Inbound and outbound messaging: Send and receive messages, including bulk sending.
  • Text & Files types support, Unicode support: Send messages with different file types. This allows for MMS message types and allows you to successfully send documents and images using the same SMS client. You can also use high bit characters for international language support.
  • Global Commands: Set rules that adjust conversational flow. Handle any “at any time text…” messaging like STOP, HELP, etc.
  • Random messaging and A/B testing: Set message variations that send during certain percentages of interactions. You can also create random surveys or sampling.
  • Timeout and error handling: Prompts in conversational design for accounting for errors and timeouts. For more see “Conversational Designer and Dialog Management.”
  • Status tracking – delivery receipts: See success rates and traffic history for flows.
  • Identifiers management – includes number groups, shortcodes: Use different number types to increase success of delivery and decrease the risk of messages being blocked.
  • Opt-in/Opt-out handling: Automatically handle opt in and opt out ongoing as well as date and time for each occurrence.

Cognitive Services

Amalgamate the best cognitive services on the market and easily adopt new innovations as soon as they become available using the OneReach.ai cognitive architecture.

Generative AI (GPT-4, etc.)

  • Autogenerate Disambiguation Questions: If your AI agent is confused, it can generate questions to gain clarity and move forward. For low confident responses the AI agent can ask follow-on questions of the user to improve confidence level and then auto train based on end user acknowledgements.
  • Create Synonym Questions: Mimic natural language by asking the same question in different ways.
  • Generate Text: Generate training examples that extend NLU model. Greatly improve training datasets by generating many examples based on a few manually entered utterances.
  • Out-of-the-box Model Usage: Use ChatGPT Integrations and APIs to access and make use of ChatGPT as is
  • Use GPT-4 in conjunction with other Encoder NLU engines: Use both GPT-4 and other NLU Engines (Luis, Azure, Rasa, etc.) to see if there is agreement between GPT-4 and the other engine and determine confidence. GPT-4 does not provide confidence levels, so it is much safer to use when combined with other NLU engines.
  • Prompt Engineering Tools: Create, tune and evaluate prompt inputs and outputs. Basically, this is backend usage of GPT-4 to finetune knowledge models. For example, in our steps, you can access GPT-4 to create synonyms, antonyms, classification data, rephrasing, etc. Burden is on the designer to make sure that the data set (instructions to GPT-4) is narrowed and specific.
  • GPT-4 Toolkits: Use GPT-4 broken down into its component elements, so they can be used within fashioned constraints for enterprise solutions
    • Classify
    • Entity Extraction
    • Synonym phrase
    • Data transformations
    • Q&A
    • Tone rephrase
    • Confidence phrase
    • Summarize
    • Antonyms
    • Embed Question
    • Stay on Subject
    • Phrase to question
    • Sentiment
    • Simple Query
    • Analyze phone outcomes

NLU and Language Tools

Natural Language Understanding and Language Tools

  • Automatic Disambiguation: OneReach.ai offers many methods of disambiguation with granular control methods. With simple disambiguation between two or more intents with high confidence, OneReach.ai automatically asks clarifying questions to determine the user’s goal. We use every disambiguated response collected to train our model automatically, improving its NLU capabilities. This capability extends to entity clarification as well to ensure we fully confirm our understanding of a users needs.
  • Multiple-Intent Matching: OneReach.ai’s first-party NLU engine supports the detection of as many intents that exist in an utterance. We then allow the designer to determine how many intents to acknowledge at once and how to coordinate the task management that results from these intents. This can be done with our no-code builder and therefore without engineers or NLU experts required. It’s also important to note that these controls are granular based on individual scenarios and are not universal rules.
  • Antagonistic Training: Use multiple NLU engines to train each other based on the highest percentage answer.
  • Knowledge Model Training: Educating your AI agent with relevant information. Create phrase/responses pairs for your solution that allow your solution to understand your user’s needs. You can train with a variety of response types such as text, JSON or a Flow.
  • Suggested/Unrecognized Handling: Set up your flow to train and update your model easily based on confidence levels gathered from interactions with the user
  • Confidence Scores: Your AI agent reports how certain it’s understanding is. This allows the designer to create logic to allow the AI agent to answer based on a rating percentage or above otherwise forward to a human to ensure highest confidence in answers.
  • Custom Model Creation: Connect to an existing model that has already been created
  • Leverage other knowledge models: Use any models you have created using NLU engines like MS LUIS, Dialog Flow, AWS Lex, RASA, and others.
  • Import Knowledge Model: Upload mass amounts of information to educate your AI agent. Upload a CSV or JSON file of an existing model that will populate phrases, responses and context into a new knowledge model you can then use in your solution.
  • Native Ontologies – Graph DB: OneReach.ai utilizes Graph Databases to store, organize and manage ontologies (for example, you can use a graph database to make connections between different internal operations, Q&A pairs, customers and their preferences/history, open tickets, etc.). These are incredibly powerful tools for tracking data and predicting trends or behavior across the customer journey, uncovering meaningful relationships within customer data that previously would have been impossible to discover.

Entity features

  • Spelling check: Inspect end user responses and correct wrong language usage to ensure end user meaning to improve user experience.
  • Autogenerate Disambiguation Questions: If your AI agent is confused, it can generate questions to gain clarity and move forward. For low confident responses the AI agent can ask follow-on questions of the user to improve confidence level and then auto train based on end user acknowledgements.
  • Generate Text: Generate training examples that extend NLU model. Greatly improve training datasets by generating many examples based on a few manually entered utterances.
  • AI Enabled Langauge Translation: We leverage language detection and machine translation for NLU on a user’s input for both voice and text via our partner ecosystem and via Autoregressive Language Models (ALMs, like GPT-4). Best practice is to have individual models for individual languages, but as a kick-starter this can be helpful and OneReach.ai fully supports this for both voice and text.

Machine Learning (ML)

  • Collect unrecognized utterances for training: We collect any utterances that prompt disambiguation techniques simple GUI for further review. We have full control in choosing an unsupervised or supervised learning approach to using this data to retrain our NLU models. OneReach.ai unsupervised learning algorithms automatically create suggested training data based on live conversations to vastly accelerate AI agent performance. They also identify unrecognized user phrases and can proactively prompt business users in a channel of their choosing to the potential of additional use cases.
  • Reinforcement Learning: We also support the continuous training of models via reinforcement learning through simple methods such as a thumbs up or thumbs down from users, live agents grading suggested responses from our models, dynamic A/B testing that automatically reinforces a model’s behavior based on certain success criteria, etc.

Cognitive Architecture

  • Sentiment Analysis: We fully support sentiment analysis; Since we have access to multiple sentiment analysis engines, you will always have access to the best performing options. These can be combined with contextual awareness to further improve the analysis (e.g. knowing that someone has called in three times today to ask a question about their bill likely means they are not highly satisfied). We have also created a custom sentiment analysis model with GPT-4.
  • Tonal Analysis, Demographic Matching, Voice Stress Analysis, etc.: Similarly to sentiment analysis, we can support any additional analysis engines and amalgamate multiple.
  • Computer Vision: Derive information quickly and accurately from files (PDFs, PNGs, JPEGs, etc.), images, and other media

Information Retrieval

Augmenting your LLMs with Memory and Plug-ins for Scalable Enterprise wide Knowledge Management

Semantic Search

Use a combination of embeddings, vector databases, graph databases, and LLMs to create semantic understanding of your enterprise data at scale. OneReach.ai provides natively managed services across the entire Semantic Search Stack.

  • Indexing: Vector databases and graph databases can be used to index the content of documents with semantic similarity, leading to more semantically relevant results than traditional text-based indexing methods can produce.. 
  • Querying: Generate more natural and informative search queries in combination with a personalized biosketch and other available meta data. This can help users to find the information they are looking for more easily
  • Retrieval: Rank documents more accurately and relevantly. This can help users to find the most useful information more quickly.
  • Evaluation: Evaluate the effectiveness of information retrieval systems more accurately. This can help researchers to improve the design and implementation of these systems.

Traditional Search

We also offer traditional search capabilities, using keywords to create queries and find relevant results. This is a fast, cheap and reliable way to perform search that. depending on your use case, may be all you need to achieve strong information retrieval results.

Hybrid Search

Often, the best information retrieval results follow from a hybrid approach combining traditional keyword search and semantic search. With this hybrid approach, traditional search acts as the first layer to quickly and cheaply narrow the search to a localized domain. Semantic search is then applied over this smaller domain, resulting in an improved search outcome for the user.

User Interface

LLMs allow for an improved search experience, transforming search from a one way experience into a multi turn dialog. Users now have a friendly conversational interface that brings the most relevant information to them in a friendly and digestible format, with the ability to ask follow up questions, maintain context, spanning across modalities and channels, and leveraging in-context micro-UI components where they improve the experience. 

Feedback

Feedback from users, such as simple thumbs up or down, can be used to reinforce model performance or to provide more detailed feedback on the relevance of search results. LLMs can also be used to generate their own success metrics, such as the number of times a human is requested for support or the amount of time a user spends reading a document. This feedback can be used to improve the performance of the information retrieval system.

LLM Custom Prompt Engineering

OneReach’s flexible information retrieval capability extends to custom prompting methods. There are infinite ways to optimize retrieval through prompting, and new techniques are being discovered continuously. Here are a few examples:

  • Chain-of-Thought (CoT): These prompts ask the LLM to explain its reasoning step-by-step, which can help to identify important concepts and relationships in the query.
  • Hypothetical Document Embeddings (HyDE): These prompts ask the LLM to imagine a document that would answer the query, and then to generate a vector representation of that document. This vector can then be used to retrieve similar documents from a corpus.

Enterprise Integration

Leverage a shared library with 700+ pre-built Steps (or build your own using JavaScript) so you can amplify the value of your existing technology investments and create a unified engagement layer for your enterprise.

Identity

  • Deep feature integration with Identity services like Okta, SSO, Active Directory, JumpCloud, IBM, etc.
  • Automate Account creation, password updates, etc.

CRM

  • Deep feature integrations with CRM services like Salesforce, Oracle, monday.com, Zendesk, Pipedrive, HubSpot, etc.
  • Use OneReach.ai as an orchestration layer, automating communications, processes, and integrating between CRM and other Enterprise systems

ITSM & HR

  • Deep feature integrations with ITSM and HR tools like ServiceNow, Workday, Peoplesoft, UKG Dimensions, Gusto, etc.
  • Use OneReach.ai as an orchestration layer, automating communications, processes, and integrating between HR and IT systems and employees

Data & Workflow

  • Deep feature integrations with Data tools like Office365, Tableau, Power Apps, Snowflake, etc.
  • Use OneReach.ai as an orchestration layer, automating communications, processes, and integrating between data and workflow tools

Data & Identity Services

Graph DBs

Store, organize, and manage ontologies for tracking data and predicting trends and behavior.

  • Native Database – Graph DB: OneReach.ai utilizes Graph Databases to store, organize and manage ontologies
  • Use a graph database to make connections: Create ontologies linking between different internal operations, Q&A pairs, customers and their preferences/history, open tickets, etc. These are incredibly powerful tools for tracking data and predicting trends or behavior across the customer journey, uncovering meaningful relationships within customer data that previously would have been impossible to discover.

Biosketches

Conversations are full of unstructured information that is usually not stored in legacy enterprise systems that are built using relational databases. Use OneReach.ai to automatically extract, store, and access unstructured data from conversations to fuel personalization and benefit from using conversational memory.

  • Storing of unstructured data: Data storage in biosketches so that all information shared in conversations can be leveraged later on. For example, when someone says they are unavailable for an appointment on Tuesday afternoon since they have to pick their kids up from school is really helpful since we now know they are a parent. This could come in handy in several ways, including the ability to emphasize safety features in future conversations.
  • Channel and context tracking: Conversations can persist on any combination of multiple channels simultaneously, or be spread out over both time and channels. Cross channel context tracking and security protocols are built into the OneReach.ai platform, so that developers and business owners alike can create rich, journey specific experiences.

Tables and data logging

Leverage Redis, Dynamo, Graph DB, Elastic Search, SQL Server all from UIs and APIs native to the platform, without having to worry about the underlying infrastructure

  • Extensive logging: Since OneReach.ai is based on an event model, we log vast amounts of data from every conversation. 
    • This includes: Custom Tags, Time, Location, Context, Transcript history, Feedback, User information, Total Users, New Users, User Engagement, Sentiment, Session Length (including response times per each piece of the engagement), Completed Conversations, Uncompleted Conversations, Fallback Responses, Missed Intents, Repeat Engagement, Top Messages, Top paths taken by users, Fulfilled Conversations, Escalated Conversations, Unfulfilled Conversations, Time-based pattern recognition, User-based pattern recognition, Topic-based pattern recognition
    • And hundreds of other potential data points that are completely configurable.
    • Standard metric collection and reporting: We provide flows for standard metrics for conversational reporting. Our focus is on conversation-experience specific analytics. See the Reporting and Analytics section for more detailed features.

Compliance and governance

Restrict access to information based on identity and/or role, preserve audit trails, and restrict access to third-party system credentials.

  • Encrypted data at rest and in transit: All data whether in transit or at rest is encrypted. During transit we use TLS 1.2 and at rest we use AES-256 bit encryption. Data retention policies can be customized at the PDE level, so customers have full control and ability to customize data storage as fits their use cases (See W03 for information about PDEs).
  • Least access rights: For privacy compliance we make sure that only certain people have access to certain environments based on their role. Roles-based privilege can be determined differently on a per-customer basis.

Multimodality

Seamlessly use any combination of channels sequentially or in parallel with shared context that spans time and channels while preserving conversational state.

Conversational memory

Conversational data (structured and unstructured) automatically saved in biosketches to fuel hyperpersonalization.

  • Bio sketches: Unstructured data is automatically stored in bio sketches so that all information shared in conversations can be leveraged later on. For example, when someone says they are unavailable for an appointment on Tuesday afternoon since they have to pick their kids up from school is really helpful since we now know they are a parent. This could come in handy in several ways, including the ability to emphasize safety in future conversations.
  • Hyperpersonalization: When external data systems are combined with our supplementary data products (such as graph databases for managing unstructured data), conversational context and hyper personalization are much more feasible.

Biosketches

Conversations retain context across channels and time.

  • Share session variables, entities, and context: Users can share any session variables, entities, and conversational context across channels and time (makes omnichannel, multimodal and contextual awareness feasible).

Right channeling

Easily execute channel pivots, multimodality, and simultaneous channel usage to deliver an optimal experience for the user based on their profile, preferences, machine learned optimizations, and other attributes.

  • Channel Pivot: Easily shift from channel to channel, based on what will provide the best user experience. For example, during a phone call you can collect email via SMS to make it easier on the user.
  • Simultaneous channel usage/Synchronizer: Channels can compliment and work together better through OneReach.ai synchronizer capabilities, so that multiple channels can be used concurrently or in asynchronous ways (e.g. synchronizing channels between Microsoft Teams and Slack).

Compliance & Governance

Restrict access to information based on identity and/or role, preserve audit trails, and restrict access to third-party system credentials.

  • Encrypted data at rest and in transit: All data whether in transit or at rest is encrypted. During transit we use TLS 1.2 and at rest we use AES-256 bit encryption. Data retention policies can be customized at the PDE level, so customers have full control and ability to customize data storage as fits their use cases (See W03 for information about PDEs).
  • Least access rights: For privacy compliance we make sure that only certain people have access to certain environments based on their role. Roles-based privilege can be determined differently on a per-customer basis.

Hosting and runtime

Cloud-native infrastructure that uses a unique blend of containerized and virtualized services to provide a highly flexible, scalable and redundant platform.

Private Dedicated Environments

Containerized architecture gives each OneReach.ai customer granular control over infrastructure-level considerations such as data residency (geographic locations), data retention, access controls, and availability (DR and Failover).

  • Private Dedicated Environments: The OneReach.ai platform is implemented on top of cloud infrastructure and uses a unique blend of containerized and virtualized services to provide a highly flexible, scalable and redundant platform. Our architecture allows OneReach.ai to provide our customers with one or more Private Dedicated Environments (PDEs) which are unique instances of our platform in specific geographic regions unique to each customer, each of which is multitenant (meaning multiple accounts for a single enterprise which is how we can offer well integrated solutions that are federated across different teams and business units). 
  • Separate environments for Dev, QA, Staging and Production: This separation allows us to provide separate environments for Dev, QA, Staging and Production. In addition this also means that our customers can have completely redundant environments in separate regions and either load balance their solution traffic between regions or use one as a primary and the second as a failover region to ensure solution availability.
  • Full control over infrastructure and publishing: Each customer has a dedicated environment, and therefore full control over their infrastructure.

Serverless

Each experience built is a standalone application and microservice that is deployed to your PDE. These can be expressed as HTTPs endpoints, enabling you to build your own APIs.

  • Serverless Architecture: The output of what you build on OneReach.ai are standalone applications and microservices that are deployed into your private dedicated environments hosted within AWS. Our services heavily utilize containers, Kubernetes, and serverless technologies. 
  • Microservice Architecture/Event Based Model: Every flow built in our platform is a microservice and is triggered by the centralized event-based system. This means any flow (and step within a flow) can subscribe to events, emit events, and those events can then trigger other flows to run. 
  • Event-based Architecture: Because of the event based architecture, you can create flows that can handle complex schedules, scheduling automation for specific dates like the third wednesday of the month, or for specific scenarios, like sending and email three days after a form is completed if a user has visited the website more than twice this week. 
  • Composability: ​​Using a microservices, event-based approach that incorporates autoscaling services (like Teraform), every aspect of the platform autoscales. This means when a customer deploys a new skill it is unnecessary to estimate traffic or involve IT to provision any services as the platform is fully elastic. The only consideration is soft limits put in place to prevent intentional or unintentional load. We’ve implemented a very granular capability for our customers to manage soft caps, down to flow concurrency and read-writes. Out-of-the-box if not otherwise configured, our entire solution automatically scales to hundreds of request per second and 2,000 concurrent calls (each of these are soft limits that can be increased on demand).
  • Build your own APIs: Each Flow can be expressed as an HTTPs endpoint, enabling you to build your own APIs.

High Availability, Redundancy, & Scalability

PDEs can be hosted in one or more AWS region(s) and configured to account for the highest levels of availability. Scalability is limited only by your imagination and the limits of AWS.

  • High Availability: Each PDE has it’s own SLA, so the level of availabilty of a solution is based on how many PDEs are utilized.
  • Multi-version Publishing Control: OneReach.ai is unique in the ability to publish new iterations of experiences while production traffic is running with no interruptions. This results in exponentially more public AI agent iterations, which we believe is critical to fine tuning user experience. (see Private Dedicated Environment above)

RESTful Architecture

Build your own endpoints, embed API calls into your worfklows, and design using our event-based architecture to ensure a fully open system that is future-proofed. Integration options and frameworks are so comprehensive that we are sometimes used as an Integration Platform as a Service.

  • Event Based Architecture: Every flow built in our platform is a microservice and is triggered by the centralized event-based system. This means any flow (and step within a flow) can subscribe to events, emit events, and those events can then trigger other flows to run. 
  • Build your own APIs and endpoints: Because of the Event Based architecture, you can create flows that can handle complex schedules, scheduling automation for specific dates like the third wednesday of the month, or for specific scenarios, like sending and email three days after a form is completed if a user has visited the website more than twice this week.
  • Embed API Calls into Workflows: Include integrations with external systems, enterprise wide systems, etc. into any workflow built on our platform. With deep integrations, our platform can listen and respond to literally thousands of different events that can alter and trigger automations.

Modern Building Experience

Consistent and reliable communication across channels: Shared library with 700+ steps, including deep channel integrations so you can take advantage of native channel functionalities where it matters most

No-code

700+ pre-built steps can be pulled from the Library to create sophisticated, custom AI agents to automate workflows, tasks and conversations with no-coding required.

  • Built for citizen developers: 80% of our platform’s users are non-developers.
  • Configure logic without coding: non-developers just need to sequence and configure the logic as their use case requires. We focus less on pre-created complete solutions (in fact, we don’t believe it is possible to offer complete solutions for complex scenarios; it is only possible to offer complete solutions for very basic ones), and focus more on starter packages to equip customers for building (teach a man to fish…).

Low-code

Crack open any Step if you’d like to tweak something by writing some Javascript, or view the logs to troubleshoot.

  • Integrated Developer Environment: Unlike most low-code tools that are a GUI facade for configuring a set of pre-defined capabilities, OneReach.ai offers an integrated development environment (IDE) that allows developers and non-developers to co-create AI agents, end-to-end conversational applications as well as tools and patterns for building them. This team-based approach facilitates creativity and imagination in ways that other platforms don’t.
  • Edit Library templates as needed to suite enterprise requirements: Crack open pre built components that facilitate an agentic approach to automating workflows, and gain full access to customize them.

Full-code

Write your own functions and build your own UI elements using common languages and SDKs (e.g. JavaScript, JSON, VUE.js, NPM packages, etc.)

  • Build reusable UI Elements: Developers can create steps to be reused by non-developers and by departments across the enterprise
  • Export and import code: Export and import sections of code or code for entire workflows.

Library

700+ templated, editable and reusable components.

  • Composable objects: Since many enterprises come to us with complex, custom requirements, we’ve found that our investment in composable objects has really paid off. Advanced tooling to kick-start automation makes a really big difference, and we’ve spent a lot of time to make these advanced tools feasible to use. Generally, we find that 90% of what our customers need is already available within our Library (with pre-built building blocks, deep channel integrations, and reusable components), leaving our customers with around 10% of their requirements requiring custom development.
  • Domain specific steps: Steps and flows in library include some that are specific to different Industries, Use-cases and Domains.
  • Channel integrations: The OneReach.ai platform supports native integration with every channel listed in this survey (see section O for specifics on which channels) and any others using our deep integration toolkits. OneReach.ai leads the market in channel flexibility because of our large library of existing integrations and because we have a UI that allows you to create custom API steps visually, without requiring coding knowledge. You can integrate within minutes without a development team, and this is one of our greatest strengths as a platform.
  • Deep Integration Toolkits: It is important to note that OneReach.ai utilizes “deep integrations” to provide native functionality across all channels, with best practices for each channel built in to our low/no code GUI designer. This is a major differentiator when compared to other platforms that don’t have native integrations (e.g. rely on a third party “voice gateway”) or have basic integrations that don’t allow for channel-specific functionality.  For example, our integrations toolkit with Slack allows users to take advantage of the fullness of Slack’s features to create better designed interactions and conversations (threads, buttons, etc.). They offer a variety of application surfaces (messages, modals, app home, uploads/posts), entry points or initiation actions (slash commands, shortcuts, chat messages, emoji, any other events), and a variety of formatted and interactive components (Block Kit, unfurling and basic formatting like mentions and Markdown). Take in combination, it lets you create better-designed conversations and interactions, using the right mixture of these tools for the user’s goal and the task, rather than being limited to vanilla back and forth interactions.

Human-in-the-Loop/Live Agent

Use co-bots to supercharge your team. Knowledge workers and contact center associates can now assist with training IDWs, disambiguate intents, and have live conversations with users.

Co-bots and Agent Assist

Allow your team to instruct Intelligent Digital Workers to take action (e.g. dispatch an intelligent digital worker to capture payment information so the agent doesn’t have to be exposed to sensitive information), and partner with your IDWs to accelerate time to resolution.

  • Sentiment Analysis/Emotion Detection: We provide built-in sentiment analysis and emotion detection information to live agents using a multitude of different technologies. This includes traditional NLP methods, along with ALMs (Autoregressive Language Models, like GPT-4). 
  • Passing information to human agents: We provide any aspect from the conversation or prior customer records to better inform the agent. Information like conversation topic, issues involved, status of customer are often used, but you can go beyond that and add things like sentiment analysis or recommended next actions to the agent. We can pass all forms of data for screen pops, CTI or any other functionality you may use on our stand alone agent interface or your third party system via custom SIP headers, APIs or more legacy methods of data transfer. 
    • Conversational history: AI agents can provides context about the conversation, presenting customer history, or things like sentiment analysis, etc.
    • Industry recommendations and real-time analysis: AI agents can indicate how to best handle certain topics if they are recognized. For example, in a sales call the agent can be informed that the customer is 22% more likely to convert if you mention our free return policy within the next 30 seconds.
    • Auto generated responses: AI agents can introduce auto generated response suggestions based on conversation context
    • Whisper agent: The interface can also be configured to introduce a “whisper agent” to assist the human agent throughout a conversation, and even enable widgets like date collection to be used by human agents to collect information in the same GUI based way an AI agent might.
    • Human in the Loop: Human-in-the-Loop capability allows agents to monitor all live conversations and be automatically prompted by an AI agent when it needs assistance to clarify a customer’s goals. These prompts are entirely customizable and can be based on the context of a conversation. For example, if an NLU’s confidence score falls below a certain threshold (which is set at the granular utterance level or the macro global level),  an agent can disambiguate the intent/entity in real-time. The human agent can then hand the conversation back to the AI agent or take over the conversation. Either way, we use this data to improve our NLU models while providing a superior user experience.

Escalations

Fully customizable agent escalation, handoff, and transfer routing, inclusive of sending agents summaries of previous engagements.

  • Multilayer Exception Handling: Exception handling can be determined with any variables, and includes fully customizable fallback responses, suggested responses, or escalation/handoff. 
  • Agent Escalation and Handoff: Agents can interject, take over conversations fully, and hand back the conversation to either human or digital colleagues. Our philosophy is that you should not limit your design thinking to “handovers between automation and live agents”, but instead consider that humans and digital workers are colleagues that are able to assist each other. For example, one conversation might start over the phone with an AI agent, transition to an agent for chat, be handed back to an AI agent to securely collect payment information, return to a voice agent for a live conversation, and traverse to automation again for a survey on yet another channel. And based on the survey results it could trigger another live conversation with management.
  • Escalation and Handoff in any channel: Agent escalation and handoff can be done on any channel or combination of channels, regardless of where the user is contacting the AI agent.

Any Live Agent UI

Use the OneReach.ai Live Agent interface, a third-party UI (e.g. Salesforce, ServiceNow, Teams, Slack, etc.), or build your own.

  • OneReach.ai Native Live Agent UI: OneReach.ai provides a native Live Agent interface that includes a fully customizable/composable UI and advanced features like agent assist. However, we can also serve as a headless orchestrator of third-party agent interfaces (e.g. Genesys, Salesforce) in which case we are able to interrogate every message from each party to assist with next best actions, agent coaching, etc.
  • Integrate with any third party live agent UI: Connect with any third party Live Agent UI.
  • Human-in-the-Loop and Synchronizer Capabilities super-charge Live Agent UI: Human-in-the-loop and synchronizer capabilities enable agents and customers to converse naturally across any number of channels in the same conversation. See T05 for a full description of Human-in-the-Loop and C02 for more information about Synchronizer. Human agents can watch AI agent conversations live via any channel and interject when it’s required, automatically be pulled into conversations based on topic, missed intents, or other attributes, and train the AI agents to ensure they are up to date. Any sequence of logic from the Flow can prompt an agent hand-over, and the hand-over can also be routed to particular agents or groups of agents based on real-time evaluation of availability and statistical optimization of outcome. It is important to note that customers can seamlessly maintain conversation between ai agents and human agents on the same UI regardless of the original channel of the user.
  • Live Agent Transfer Routing: Live Agent transfer logic is fully customizable via our no-code flow builder. This includes first-in-first-out, skill based routing, advanced queuing, bullseye routing, etc. See T04 below for context and transcript abilities.

Compliance

Redact information, protect PII, and mute recordings on an utterance by utterance basis.

  • Least access rights: For privacy compliance we make sure that only certain people have access to certain environments based on their role. Roles-based privilege can be determined differently on a per-customer basis. 
  • Masking PII: We can redact PII in live conversations and generated transcripts using traditional redaction NLP, regex, and ALMs (Autoregressive Language Models). We can also mute recordings and their corresponding transcripts on an utterance by utterance basis via our native voice stack.
  • Battle Tested: Our live agent interface has been used in production for 10+ years by thousands of agents. We have successfully deployed incredibly complex live agent configurations across the world, and industries. One customer currently uses our agent system for 4000+ agents across tens of call centers across the globe.
  • External Live Agent Connectors: The OneReach.ai Live Agent system can seamlessly integrate with third party live agent systems (e.g. Salesforce, ServiceNow, etc.) and can pass all forms of data for screen pops, CTI or any other functionality via custom SIP headers, APIs or more legacy methods of data transfer.
  • Detailed Agent Metrics: We log detailed agent performance metrics to enable granular details on every agent interaction. OneReach.ai can store data for you or stream data in real time to a third party BI tool/database.

Reporting & Analytics

Standard and custom reports available for conversation-experience specific analytics.

Standard Reports

47 out-of-the-box predefined metrics for conversational evaluation including charts and datasets for paths, session stats, graphs, dendrograms, logging, etc.

  • Full Customization of Reports: Customers can customize reporting views, dashboards and widgets that tie into their conversational experiences and any automated solutions. They can create triggers for automated tasks and real time analysis and even automated adaptations driven by data. We equip customers for making real time adaptations to the experiences their solutions offer, based on analysis of data, and triggered events in custom reports. OneReach.ai has the ability to store and report on usage and performance metrics. 
  • Robust Analytical Tools: The reports are built using the Flow builder (no-code designer) to ensure anyone can build them, not just data engineers. OneReach.ai can store data for you or stream data to a third party BI tool/database in real time/latent methods. You can download analytics anytime manually as well.
  • Reporting events embedded in each step of our builder interface: Flow Builders can use the Reporting Events feature embedded in each Step to decide what information should be logged. You can also write directly to the logs (Cloudwatch) which can also be accessed from your other systems for further integration and analysis. Furthermore, you can design data Flows that feed reporting Cards and Views that can visualize high-level stats such as trends, missed intents, channel utilization, response times, completion rates, etc. Analytics examples have been provided in a separate file and live demos are available upon request.

Custom reports

Quickly and easily customize existing report templates or create them from scratch. If you can drag-and-drop, you can build a custom report.

  • Proprietary Analytics Solution (Cards, Views and Dashboards provide customizable, live and historic data reporting): Our platform allows you to create complex, ultra-specific analytics dashboards to test content, ai agent personality, or even entire processes that are tuned based on previous conversation history, user attributes, cohort analysis, and real time user interactions. While we offer dozens of premade analytics solutions (tables, charts, graphs, etc.) that can be used with any solution, most enterprises come to us with specific, custom reporting dashboards in mind, and we have focused on making this possible. 
  • Library Cards – Predefined Metrics: We have a full library of 47 out-of-the-box predefined metrics or conversational evaluation. Customers are able to edit the templates we have, or create advanced custom metrics and add these metrics to their library independently to grow their analytics capabilities over time. 
  • Library Views – Gather Metrics in one place to view and share: Cards are organized into what we call Views. Often our customers will create custom views for different roles in their organization. The views can be shared with users in your account, or via a sharable link.
  • Dashboards – Collections of views: Customers custom create dashboards using Library Cards and Views, which can be combined to create views in minutes.

Conversational data

Tired of charts and graphs? Have a conversation with your data using your preferred interface (e.g. ask in Slack how the sales pipeline looks this quarter).

  • Enrich Chat Logs with custom Metadata: Transcript logs can be made more helpful and easy to read with the use of custom metadata. This can be configured and then placed into Views and Dashboards. Any step in any flow allows the builder to add custom tags. These tags are collected in real time and sent to a real time logging database so that they can be used not only for the collection of data for the purpose of analytics but also to emit events and trigger other actions.
  • Traffic and Paths Reports: In addition to overlaying custom meta data on top of transcripts, we also have strong capabilities in visualizing the entire chat flow on the flow builder itself (along with most common traffic patterns broken down by percentage). This enables our builders to quickly troubleshoot any issues and gain insight into common paths take by users which are often labeled by desired results. We can analyze the relationships between orchestrated flows and AI agents to via graph capabilities and overlay the chat journeys on top of all of this analysis.

Access Controls

Expose the right information to the right person at the right time. Create dashboards for user cohorts based on their interests and privileges.
  • Least access rights: For privacy compliance we make sure that only certain people have access to certain environments based on their role. Roles-based privilege can be determined differently on a per-customer basis. 
  • Hyper-customized access control: Solution access can be configured in multiple ways based on the solution that is designed, configured and implemented on the OneReach.ai platform. We have supported everything from ID/Password, multiple SSO implementations, PIN/Passcodes, biometrics, etc.
  • GUI Admin Controls: OneReach.ai has a built in advanced admin controls GUI for all aspects of the platform. User roles and configurations are all easily managed and governed from simple GUIs within the “Action Desk” module of the OneReach.ai platform. As an additional access control, each flow offers unique password protection in addition to standard role assignment.

Accounts, Admin, and Federation

We have customers with over a thousand users and provide granular control to system administrators.

Control User Permissions

Customized access controls based on user type.
  • Least access rights: For privacy compliance we make sure that only certain people have access to certain environments based on their role. Roles-based privilege can be determined differently on a per-customer basis. 
  • Hyper-customized access control: Solution access can be configured in multiple ways based on the solution that is designed, configured and implemented on the OneReach.ai platform. We have supported everything from ID/Password, multiple SSO implementations, PIN/Passcodes, biometrics, etc.
  • GUI Admin Controls: OneReach.ai has a built in advanced admin controls GUI for all aspects of the platform. User roles and configurations are all easily managed and governed from simple GUIs within the “Action Desk” module of the OneReach.ai platform. As an additional access control, each flow offers unique password protection in addition to standard role assignment.

Add users

Multi-user functionality which grants access rights to independent PDEs and accounts.

  • Multi-user functionality: Many of our enterprise clients have scaled to a significant number of PDEs and accounts within those PDEs. For example, one Fortune 100 company has over 135 independent accounts with thousands of users. To provide seamless user management at scale, we have enabled multi-user functionality which grants access rights to independent PDEs and accounts with the same enterprise email. 
  • Assign unique user roles: Each account and PDE assigns a unique user role to the same multi-user identifier. 
  • Least access rights: For privacy compliance we make sure that only certain people have access to certain environments based on their role. Roles-based privilege can be determined differently on a per-customer basis.
  • Department Based Controls: Assign access controls to users based on department.

Versioning and recovery

Store historical versions of solutions and revert back as needed.

  • Advanced versioning and recovery features: We provide advanced versioning and recovery features that enable you to revert your dialogue to any version since its inception. Even better, when your new version of dialogue is undesirable, you can easily roll back the hands of time to edit from a previous version with a simple GUI. Complete versioning management includes the editor’s user name, date, time, version number, and version change logs.
  • Change logs: Each version of an agent comes with unique change logs that are easily accessed and compared across versions. This helps with maintaining fast iteration on production solutions.

Admin

User roles and configurations are all easily managed and governed from simple GUIs.

  • GUI Admin Controls: OneReach.ai has a built in advanced admin controls GUI for all aspects of the platform. User roles and configurations are all easily managed and governed from simple GUIs within the “Action Desk” module of the OneReach.ai platform. 
  • Password Protection: As an additional access control, each flow offers unique password protection in addition to standard role assignment.
  • Customized Lifecycle Controls: customized life-cycle services which include things such as Account and User management, Solution Deployments and Automated testing. For instance:
    • Create/Manage Communication Studio Accounts
    • Create/Manage User Accounts
    • Copy Solution components from target location to destination component
    • Perform destination specific configurations (Endpoint, data, user account, variables, etc.)
    • Perform validation tests post deployment.
    • Perform rollback to previous versions

Privacy and security

Privacy, security, and roles-based privilege can be configured based on your specific enterprise requirements.

Privacy

Redact PII in live conversations and generated transcripts using traditional redaction NLP, regex, and ALMs (Autoregressive Language Models).

  • Encrypted data at rest and in transit: All data whether in transit or at rest is encrypted. During transit we use TLS 1.2 and at rest we use AES-256 bit encryption. Data retention policies can be customized at the PDE level, so customers have full control and ability to customize data storage as fits their use cases (See W03 for information about PDEs). 
  • Least access rights: For privacy compliance we make sure that only certain people have access to certain environments based on their role. Roles-based privilege can be determined differently on a per-customer basis.
  • Provider Flows: these are flows that contain credentials to third party systems. Regular Flows that call these systems (e.g. Salesforce) have an authentication step that does not reveal any system-level authentication credentials; those are obfuscated by the Provider Flows, thus preventing the wrong people from having access to that information.
  • Masking PPI: We can redact PII in live conversations and generated transcripts using traditional redaction NLP, regex, and ALMs (Autoregressive Language Models). We can also mute recordings and their corresponding transcripts on an utterance by utterance basis via our native voice stack.

Security

OneReach.ai works with customers with the strictest data security and compliance requirements including those from the medical, insurance, and government industries. If there are specific security or compliance needs we can customize the account to match your specifications. 

  • Certifications: GDPR, HIPAA, CCPA, PCI DSS, SOX, SHIELD, LGPD, FedRAMP, SOC 1, and SOC 2
  • Security Documentation: Since OneReach.ai offers Private Dedicated Environments with hyper-personalized client security, we have both robust security documentation and security professionals on staff to support any security needs the client may have.

Compliance

Customize access controls that can be configured based on enterprise standards, including shared authentication and multi-level permissioning.

  • Least access rights: For privacy compliance we make sure that only certain people have access to certain environments based on their role. Roles-based privilege can be determined differently on a per-customer basis. 
  • Hyper-customized access control: Solution access can be configured in multiple ways based on the solution that is designed, configured and implemented on the OneReach.ai platform. We have supported everything from ID/Password, multiple SSO implementations, PIN/Passcodes, biometrics, etc.
  • GUI Admin Controls: OneReach.ai has a built in advanced admin controls GUI for all aspects of the platform. User roles and configurations are all easily managed and governed from simple GUIs within the “Action Desk” module of the OneReach.ai platform. As an additional access control, each flow offers unique password protection in addition to standard role assignment.

Access controls

Customize access controls that can be configured based on enterprise standards, including shared authentication and multi-level permissioning.

  • Least access rights: For privacy compliance we make sure that only certain people have access to certain environments based on their role. Roles-based privilege can be determined differently on a per-customer basis. 
  • Hyper-customized access control: Solution access can be configured in multiple ways based on the solution that is designed, configured and implemented on the OneReach.ai platform. We have supported everything from ID/Password, multiple SSO implementations, PIN/Passcodes, biometrics, etc.
  • GUI Admin Controls: OneReach.ai has a built in advanced admin controls GUI for all aspects of the platform. User roles and configurations are all easily managed and governed from simple GUIs within the “Action Desk” module of the OneReach.ai platform. As an additional access control, each flow offers unique password protection in addition to standard role assignment.

Languages Supported

Common Languages (in production)
English: US, Great Britain, Canada, India, Australia Chinese: Cantonese, Simplified, Traditional Arabic: Peninsular Group, Mesopotamian Group, Levantine Group, Egypto-Sudanic Group, Maghrebi Group Spanish: Latin America, Castillian Spain, Catalan Spain Turkish
Czech Slovakian Russian Ukrainian Latvian
Romanian Italian French: France, Canada German
Other Languages Supported
Afrikaans Croatian Ido Malay Romanian
Albanian Czech Indonesian Malayalam Romansh
Amharic Danish Irish Maltese Russian
Aragonese Dhivehi (Divehi/Maldivian) Italian Manipuri Sanskrit
Armenian Dutch Japanese Manx Sardinian
Assamese Esperanto Javanese Marathi Scottish Gaelic
Azerbaijani Estonian Kannada Minangkabau Serbian
Bashkir Filipino (Tagalog) Kazakh Mangolian Serbo-Croatian
Basque Finnish Khmer (Central Khmer) Nepali Sindhi
Bengali (Bangla) Galician Kirghiz (Kyrgyz) Norwegian: Bokmal, Nynorsk Sinhala (Sinhalese)
Bihari Languages Georgian Korean Occitan Slovak
Bosnian German Kurdish (Kurmanji) Odia (Oriya) Slovenian
Breton Greek Latin Ossetian (Ossetic) Somail
Bulgarian Gujarati Latvian Pashto Sotho
Burmese Haitian Creole Limburgan (Limburger/Limburgish) Persian Sundanese
Cebuano Hebrew Lithuanian Polish Swahili
Chechen Hindi Luxembourgish (Letzeburgesch) Portugese: Brazil, Portugal Swedish
Chuvash Hungarian Macedonian Punjabi Tajik
Corsican Icelandic Malagasy Quechua Tamil
Tatar Telugu Thai Tibetan Turkish
Turkmen Urdu Uyghur Uzbek Vietnamese
Walloon Welsh West Frisian

Enter contact data

To try this demo we'll need some information about you

loader
✕

Contact Us

loader
✕