Askli

What is Askli?

Askli is a code-free platform for enterprise knowledge bases and AI customer service robots; currently, its official website and pricing page are still hosted under the Chaindesk domain, while the entry points to the application are gradually using the Askli brand. Users can import company data and create custom AI agents that are capable of answering business-related questions.

Initially, this product was available under the names Databerry and Chaindesk; currently, the official website uses the name Askli. It is suitable for use in website customer service, lead collection, internal knowledge querying, and multi-channel automated responses, rather than as a regular personal chatbot.

Relationship between brand and product

The three names belong to the same product evolution chain, but the old code repository, the current hosting service, and the latest commercial features cannot be considered as identical software versions. The directory name can remain Askli, with the former name of Chaindesk indicated in the text.

Main functions

• Code-free Agent creation: Configure roles, response rules, and knowledge scope via a web interface.
• Corporate knowledge base: Organizes websites, documents, and third-party materials into searchable data sources.
• Enhanced retrieval Q&A: Relevant information is searched in the company’s data before providing an answer.
• Website chat component: Select a widget style and embed it on the corporate website.
• Multi-channel communication: Supports planning through various channels such as websites, WhatsApp, Slack, Telegram, etc.
• Manual takeover: Customer service staff can take over and continue handling robot conversations when necessary.
• Session monitoring: Provides a centralized view of issues, responses, and user interactions across various channels.
• Automatic synchronization and retraining: The paid plan allows for periodic synchronization of updated data sources.
• Multilingual Q&A: Website content, user questions, and answers can be in different languages.
• Team collaboration: The paid plans offer multiple seats and higher project limits.

Supported data sources

• Website pages: Capture public web pages and create a searchable database.
• Upload files: Import products, policies, help centers, and internal documents.
• Notion: Synchronize team knowledge pages and structured documents.
• Google Drive: Connects to cloud-based documents and is used for knowledge-based queries.
• Manual text: Add common questions, rules, and standard answers directly.
• Custom data sources: Corporate clients can arrange for the integration of additional systems or interfaces.

Distribution Channels and Integration

What use cases are suitable?

• Website customer service: Answers questions regarding products, prices, features, and accounts 24/7.
• Pre-sales consultation: Presenting solutions, identifying requirements, and gathering information on potential customers.
• E-commerce support: Explains delivery, returns and exchanges, product specifications, and common order processes.
• SaaS Help Center: Answers configuration and troubleshooting questions based on documentation.
• Internal knowledge base: Allows employees to access information on policies, procedures, training, and projects.
• Multilingual support: Respond to messages in different languages using the customer’s preferred language.
• Agent delivery: Creates separate agents and data storage for multiple clients.
• Automation of messaging channels: Extending the same knowledge and capabilities beyond the website.

Which users are it suitable for

• Small and medium-sized enterprises that do not have a dedicated AI engineering team.
• Customer service and operations teams that aim to reduce the number of repeated inquiries.
• Startups that need to quickly deploy Q&A components on their websites.
• Agencies and service providers that manage multiple customer robots.
• I would like to connect with internal knowledge teams using Notion or Google Drive.
• Enterprises that need to extend their workflows through APIs or custom integrations.

Tutorial on creating AI customer service

1. Determine which topics the robot should handle, and when it is necessary to transfer tasks to a human.
2. Organize the latest products, prices, policies, help documents, and standard responses.
3. Create a data storage and import websites, files, or cloud data.
4. Create an Agent and set its identity, tone, restrictions, and actions to take when it cannot answer.
5. Use real-world questions to test the retrieval of content, the scope of citations, and multi-turn responses.
6. Configure the colors, positions, welcome message, and lead fields of website components.
7. It is first opened to a small number of visitors, with errors and unresolved issues being monitored.
8. Continuously refine the documents and Agent rules, then gradually increase the traffic volume.

Knowledge base maintenance tutorial

1. Specify the business owner for each type of data and its update cycle.
2. Delete duplicate, expired, conflicting, and unauthorized documents.
3. For important policies, provide clear titles, dates, locations, and the target audience.
4. Enable the appropriate automatic synchronization and check whether each retraining is successful.
5. Create a set of fixed test questions to serve as the regression evaluation set.
6. Determine whether the incorrect answer is due to a problem with the document, the search process, the hints provided, or the model itself.
7. Record the answers before and after the modification as well as the time of deployment, to facilitate tracking.
8. Regularly export unresolved issues to update the help center and standard answers.

Comparison of package prices

Askli currently offers four monthly subscription tiers: Free, Growth, Pro, and Enterprise. The website states that opting for an annual payment can save 20%. There are also customized solutions available for specific projects.

Comparison of package quotas

The official website indicates that the message quota is based on GPT-4o-mini. Different consumption rules may apply when other models, integrations, or additional features are selected; it is therefore necessary to check the current quota in the account before making a purchase.

Limits of the free version

• There is a limit of 200 messages per month, making it suitable for verification purposes rather than for handling high-volume customer service tasks on a regular basis.
• It contains only one Agent and one data storage.
• The maximum number of pages that can be scraped from a website is 25, which makes it difficult to cover large help centers.
• There are no team seats, so it is not suitable for multiple people to operate it.
• Free Agents and data storage are deleted after 14 days of inactivity.
• Automatic synchronization, brand removal, and certain channel functionalities are subject to the actual account permissions.

How to estimate messaging and storage needs

Manual takeover and session monitoring

• Define the conditions for routing to a human agent, such as complaints, refunds, sensitive information, and low-confidence responses.
• The user’s original question, the retrieved information, and the robot’s response are retained to help agents understand the context.
• Set a responsible person, priority, and deadline for unresolved sessions.
• It calculates the automatic resolution rate, the rate of cases forwarded to human agents, repeated contacts, and incorrect answers.
• Prevent robots from pretending to be human beings, and clearly indicate the AI’s identity on the interface.
• Regularly incorporate the manually verified correct answers back into the knowledge base.

Product advantages

• Code-free configuration reduces the barriers to deploying enterprise knowledge robots.
• A knowledge base can be created from sources such as websites, files, Notion, and Google Drive.
• It supports web pages and multiple messaging channels, facilitating the reuse of the same Agent.
• It offers manual intervention and session monitoring, not limited to fully automatic responses.
• The paid plans clearly list the messages, agents, storage capacity, and seat limits.
• Multilingual capabilities enable serving visitors from different regions.
• The same robot can be deployed on multiple websites.

Usage restrictions

• The names Askli and Chaindesk coexist, and the brand information on the official website and in the application interfaces has not yet been unified.
• Knowledge base robots may still experience hallucinations, miss certain conditions, or refer to outdated information.
• Free data is deleted after a long period of inactivity.
• Advanced models, automatic synchronization, brand removal, and integration capabilities are available only with paid plans.
• On the official price page, there are obvious unit errors in some of the text regarding file size, so that information cannot be trusted directly.
• The message quota is based on a specific model; usage with other models requires separate confirmation.
• Multi-channel deployment may also incur costs for external platforms such as WhatsApp.
• The old open-source repository no longer contains the complete source code for a deployable product.

Safety and privacy

• Only company documents and customer data for which access is permitted are imported.
• Public customer service knowledge and internal sensitive information are stored in separate data repositories.
• Use an account with the minimum required permissions when connecting Notion and Drive.
• Prevent the Agent from displaying unpublished prices, keys, and personal information directly.
• Set up processes for data retention, deletion, export, and handling of employee departures.
• Add authentication and manual confirmation when dealing with order or account operations.
• Before making a purchase, verify the model supplier, data region, sub-processor, and training policies.

Open source and GitHub status

The original Databerry repository attracted a lot of attention due to its open-source projects for creating custom LLM Agents, but currently it only contains a README file; there are no complete code files, installation instructions, or licenses available for deployment. The main resources provided by Chaindesk’s official team are currently blogs and help site branches.

Therefore, the current Askli hosting products cannot be labeled as open source or self-hostable. Historical Star and Fork counts reflect the popularity of older projects; they do not indicate that the same source code is still available today.

Basic information

Recommendation score

The comprehensive recommendation score is 4.1 out of 5 points. Askli is suitable for quickly setting up a corporate knowledge base for customer service, and it offers multiple channels for handling inquiries as well as the possibility of human intervention; however, special attention should be paid to brand migration, quota rules, abnormalities in the pricing information, and the status of existing open-source repositories.

Frequently Asked Questions

What is the relationship between Askli and Chaindesk?

Askli is the current brand name, while Chaindesk is the old name that is still used on the website’s domain and in some documents.

What is Databerry?

It was the name of an earlier open-source project; the current repository no longer provides the complete source code.

Can Askli be used for free?

Yes, the free version provides 200 messages per month, one Agent, and one data storage unit.

Will the free data be retained forever?

No, the official website states that free Agents and data storage will be deleted after 14 days of inactivity.

Which knowledge sources are supported?

It supports sources such as websites, files, Notion, Google Drive, and manual entries.

Can multiple websites be embedded?

Yes, the official website states that the same chatbot can be deployed on any number of websites.

Is it possible to have human customer service take over?

Yes, the team can monitor the session and take over when necessary.

How much is Growth?

The monthly cost is 49 dollars; the official website states that paying annually can save 20%.

Is Askli open source now?

Current commercial platforms are not open source, and the old repositories no longer contain the complete code needed for deployment.