Aissist.io

What is Aissist.io?

Aissist.io is an enterprise bot platform designed for customer service, sales, and back-office operations. It is not a chatbot that merely answers FAQs; rather, it integrates knowledge retrieval, process decision-making, system operations, manual handoff, and operational analysis into an end-to-end workflow.

The platform positions itself as an AI Operational Layer that operates on top of a company’s existing systems; it typically integrates with existing help desks, e-commerce platforms, logistics systems, and business interfaces, rather than requiring the company to replace its entire customer service system.

Three major product modules

Core functions

• Multi-agent task orchestration.
• Knowledge base retrieval and context understanding.
• Cross-system queries and data updates.
• Automation of customer service and sales processes.
• Route to the human support team according to the rules.
• Understanding of text, images, documents, audio, and video.
• Multi-channel deployment and unified processing.
• Dialogue and business insight analysis.
• Quality assessment of AI and human agents.
• Ongoing monitoring and process optimization.

AgentMesh automation engine

AgentMesh makes use of multiple specialized sub-agents to work together on a request, with each sub-agent responsible for different scenarios or steps. It can query orders, update records, handle returns, or collect sales leads in accordance with the company’s SOPs, and defer tasks to human operators when judgment is required.

• Identify user intentions and business scenarios.
• Select the corresponding sub-agent.
• Retrieve knowledge and real-time business data.
• Invoke authorized system actions.
• Check the execution results and proceed to the next step.
• Pause or transfer to a human agent according to the rules.
• Save the context for use in subsequent processes.

Digital agents and ordinary chatbots

Pulse Insights Engine

Pulse analyzes customer conversations, product issues, agent interactions, and business processes to help teams understand what is happening and why. Weekly insights are available free of charge with AgentMesh; advanced modules such as Ledger, Agent Insights, and Business 360 require contact with sales.

• Identify gaps in products and services.
• Aggregate user emotions and satisfaction levels.
• Analyze resolution rates and common issues.
• Evaluate the performance of AI and human agents.
• Extract structured information from the conversation.
• Synchronize the business records that need to be updated.
• A combination of public and internal data creates a business view.

Evolve continues to improve.

Evolve is responsible for detecting operational signals, analyzing the root causes, proposing improvement actions, and carrying out optimizations within the allowed limits. Basic AI Optimization is provided as part of AgentMesh; for features such as Signals, it is necessary to consult sales regarding the prices and available options.

Knowledge assets

Companies can add websites, documents, and other materials as Assets to the workspace, so that the agents retrieve information only when relevant questions arise. Global behavior rules should be placed in the Instructions section, while specific business processes are better handled by Sub Agents.

• Import content from public websites.
• Upload product and policy documents.
• Maintain FAQs and standard operating procedures.
• Organize knowledge by business area.
• Use a debugger to check the search results.
• Update expired content and conflict rules.

Sub Agents sub-agents

Sub Agents are used to define specific scenarios such as returns, order inquiries, account issues, or lead qualification. Teams can restrict agents to handling only the identified sub-processes, thereby reducing responses that go beyond the intended scope during the initial stages of deployment.

Actions business operations

An Action represents the actual business operations that an agent carries out during a conversation, such as querying orders, canceling orders, updating customer information, or calling custom interfaces. Each action should have designated permissions, input validation, failure handling mechanisms, and audit records.

Customer service automation

• Automatically answers questions about products and policies.
• Check order and shipping status.
• Handle cancellation, returns, and account processes.
• Gather the information needed for troubleshooting.
• Identify cases that require manual judgment.
• Leave the entire context to the agent.
• Analyze unresolved issues and improve knowledge.

Sales automation

• Answer pre-sales product questions.
• Collect information on potential customers.
• Assess the quality of clues based on conditions.
• Recommend suitable products or solutions.
• Pass on high-value leads to the sales team.
• Update records in the existing CRM.
• Analyze conversion barriers and user feedback.

E-commerce and order scenarios

The platform documentation provides connections to commercial systems such as Shopify, WooCommerce, and Adobe Commerce, and it supports operations related to orders, inventory, shipping, and returns. Before enabling write operations, it is necessary to verify permissions and error handling in a test store first.

Business systems that can be connected

Help desk integration status

Different pages on the official website show varying updates regarding the status of certain connectors; therefore, it is necessary to check both the current document and the relevant pages. Companies should not rely solely on marketing articles to determine whether Salesforce, Freshdesk, or Kustomer have enabled access to their accounts.

Gateway deployment channels

A Gateway determines in which management point or channel a particular workspace should be deployed. One Gateway can handle only one Workspace, while the same Workspace can be deployed across multiple Gateways. A project serves as the billing unit and can include multiple workspaces as well as multiple Gateways.

Supported channels

• Email.
• Online forms.
• Social media.
• Web chat.
• WhatsApp and text messages.
• Help desk session.
• Voice messages.
• Other channels that access via Gateway.

Limits of vocal capabilities

Aissist can handle voice messages, but the developers have clearly stated that it is not currently an automated voice platform designed for real-time incoming calls. If a company’s main need is to receive and make calls, it should opt for a product that gives priority to voice functionality or wait for this feature to be made available officially.

Multimodal input

The platform can process various types of input such as text, images, documents, audio, and video, making it suitable for customer service scenarios that involve screenshots, attachments, or voice messages. The model may still make mistakes in interpreting images and audio; therefore, important evidence should be examined manually by a human.

Manual transfer

• Specify the transfer team by issue type.
• Stop automatic replies at low confidence levels.
• It is handed over to human staff once specific conditions are met.
• It carries the conversation and business context.
• It is possible to pause the agent’s responses after forwarding.
• Verify routing rules using a simulator.

Simulation and debugging

Before going live officially, it is possible to use the Simulator to test the entire dialogue logic, the Asset Debugger to check knowledge retrieval, and the Action Debugger to verify action triggering and interface responses. The official recommendation is to start with one process and gradually increase the volume of real traffic.

Quick Deployment Tutorial

1. Create a project and the first Workspace.
2. Set global Instructions, tone, and security rules.
3. Import websites, documents, and other knowledge assets.
4. Connect to the help desk or business systems.
5. Create a Sub Agent for a specific scenario.
6. Test normal and abnormal issues in the Simulator.
7. Create a Gateway and first deploy a small amount of real traffic.

Knowledge base configuration tutorial

1. Organize the current valid policies and product information.
2. Remove duplicate, conflicting, and expired documents.
3. Write the global rules to Instructions.
4. Import the reference knowledge as Assets.
5. Use typical questions to check the search results.
6. Use the Asset Debugger to locate and resolve errors.
7. Establish content owners and a regular update mechanism.

Tutorial on configuring business actions

1. Confirm the external system that is to be read from or modified.
2. Create integrated credentials with the minimum required permissions.
3. Connect to e-commerce, logistics, or custom APIs.
4. Create separate Actions for querying and writing.
5. Set parameter checks, timeouts, and fallback options in case of failure.
6. Use the Action Debugger to test each action.
7. First, verify in a sandbox environment before enabling production writes.

Online evaluation tutorial

1. Prepare a set of real historical conversations as a test set.
2. Define the resolution rate, accuracy, CSAT, and escalation criteria.
3. Replay problems of different difficulty levels in the simulator.
4. Manually check responses, actions, and routing.
5. Only one process and a small volume of traffic are allowed.
6. Observe errors, costs, and user feedback.
7. After meeting the threshold, other use cases can be expanded gradually.

Which users are it suitable for

• Use the existing customer service team from the help desk.
• E-commerce companies that need to automatically query and update orders.
• A sales team that is capable of handling pre-sales inquiries automatically.
• The departments that need to convert SOPs into agent-based processes.
• Managers who wish to analyze all service sessions.
• Technical teams that need to connect to custom APIs.
• Small businesses that prefer to conduct a free verification first and then pay on a usage-based basis.

Typical use cases

• Automatically handles queries regarding order status.
• Complete the cancellation and return in accordance with the rules.
• Identify high-value sales leads.
• Answer questions regarding product and service policies.
• Extract information from screenshots and attachments.
• Pass complex cases along with their context to a human.
• Generate weekly insights on customers and products.
• Continuously evaluate the quality of AI and human agents.
• Optimization actions are proposed based on operational signals.

It’s not very suitable for which situations

• Only a team for the simplest FAQ routing is needed.
• Companies that wish to completely replace their help desk systems.
• The main requirement is automatic answering of calls in real time.
• Organizations that do not have clear business processes and designated owners for knowledge.
• Closed environments that do not provide API or system authorization.
• Companies that need the full open-source code for private deployment.
• Teams that cannot accept pricing based on interaction volume are not acceptable.

Free quota

The current pricing page states that 1,000 free tickets are provided per month, no credit card is required, and the basic engine is used. The help documentation indicates that a free account includes 3,000 AI interactions per month, with responses bearing the Aissist brand.

Work orders and AI interactions are measured using different units, and the public website does not specify a fixed conversion rate between them. Before enabling this feature, it is necessary to check on the account’s Billing page to determine the actual free quota for one’s project as well as the rules for handling any excess usage.

Growth billing

The cost for growth is calculated based on actual usage; no long-term contract is required. The base price is $0.09 per AI interaction, with a maximum charge of around $0.60 per issue resolved. The system automatically selects the lower amount between the cost calculated per interaction and the upper limit set for per-resolution charges.

Explanation of billing units

• An AI interaction is an automated cycle that goes from user input to AI response.
• A cycle can include steps such as marking, retrieval, responding, and system updates.
• A one-time resolution refers to a session in which the AI handles the request without any upgrade.
• A transfer carried out according to the predefined procedure after at least three interactions can also be considered a resolution.
• Casual conversations are not included in the definition of valid user inquiries.
• The number of interactions via email and chat can differ significantly.

Example of cost estimation

If there are an average of 4 interactions per session, the cost per interaction is $0.36, which is below the upper limit of $0.60. The actual cost also depends on the free credit allowance, session duration, resolution rate, channel, and corporate discounts.

Pulse and Evolve prices

Model suppliers

The help documentation states that its proprietary orchestration layer can connect to multiple large-model providers; OpenAI, Grok, Claude, and Gemini are mentioned as examples. Enterprises should determine which models to use in specific scenarios, as well as the data retention policies, regional considerations, and failover methods.

Safety and compliance

The security page states that the platform has obtained ISO 27001 certification and complies with GDPR; it also describes SOC 2 as being \"Aligned\" rather than certified. The procurement team should request valid certificates, scope details, audit reports, and data processing agreements, and should not treat \"Aligned\" as an equivalent to SOC 2 certification.

• Static data is encrypted using AES-256.
• Transmission uses TLS 1.2 or a higher version.
• The infrastructure makes use of AWS and Azure.
• The internal system applies the principle of minimum permissions.
• Multi-factor authentication is enabled for employee accounts.
• The agent provides PII identification and masking.
• The system provides access to and the ability to modify audit logs.
• The platform carries out vulnerability scanning and penetration testing.

AI governance

The platform enforces rules and Guardrails to regulate responses, and quality can be assessed through simulation, evaluation, and continuous monitoring. For automatic optimization, it is still necessary to define clear objectives, establish approval processes for changes, and have mechanisms in place for rolling back changes; the system should not be allowed to modify itself indefinitely in a production environment.

Product advantages

• Moving from answering questions to carrying out business actions.
• It can be added on top of the existing help desk.
• Multi-agent systems are suitable for complex, step-by-step processes.
• It supports e-commerce, logistics, and custom APIs.
• Simulators and specialized debugging tools are provided.
• The free quota facilitates testing on a small scale.
• Growth is charged based on usage rather than a fixed number of seats.
• It covers automation, insights, and optimization at the same time.

Usage restrictions

• The free quota uses different measurement criteria on various pages.
• Charging based on interaction varies with the length of the session.
• The status of some connectors may not be synchronized across different pages on the official website.
• It is not a complete alternative to a help desk.
• Real-time phone automation is not a focus at the moment.
• Complex write operations require strict permissions and testing.
• SOC 2 is currently referred to as “Aligned” rather than “certified”.
• The core code of the platform is not open source.

Is it open source?

Aissist.io is a commercial cloud platform; its official website does not provide an open-source repository for its core products. The availability of REST APIs, AWS Lambda, and integration options with third-party services does not mean that AgentMesh, Pulse, or Evolve are open source.

Basic information

Recommendation score

The comprehensive recommendation score is 4.3 out of 5 points. Aissist.io is suitable for companies that wish to achieve end-to-end automation within their existing customer service systems; however, it is necessary to standardize the billing methods before going live, verify the status of the integration, and conduct thorough testing of write operations.

Frequently Asked Questions

Is Aissist.io free?

A monthly free quota is provided without the need for a credit card; however, the pricing page and the help documents use ticket-based and AI-based metrics for interaction measurement, respectively, with the account status being the determining factor.

How is Growth charged?

The public price is $0.09 per AI interaction; the maximum cost for resolving a single issue is around $0.60, with settlement based on actual usage.

Will Aissist replace the existing help desk?

No, it is primarily integrated with existing systems such as Intercom, Zendesk, Front, Gorgias, and HubSpot.

Can you handle orders and refunds?

Actions can be carried out through authorized e-commerce platforms or custom interfaces, but it is necessary to configure permissions, rules, and fallback procedures in case of failures.

Is real-time phone robotics supported?

It is currently capable of processing voice messages, but the company states that real-time automated handling of incoming calls is not one of its main capabilities.

Which large models are used?

The document mentions that OpenAI, Grok, Claude, and Gemini can be used, with coordination provided by a proprietary orchestration layer.

Has Aissist obtained SOC 2 certification?

The security page currently states that it is SOC 2 Aligned; it should not be described as having passed SOC 2 certification.

Is Aissist open source?

It is not open source; it is a commercial cloud platform. The availability of public interfaces and connectors does not mean that the core source code is made available.

How to test before going live?

The Simulator, Asset Debugger, and Action Debugger should be used simultaneously, starting with a single process and a small amount of real traffic.