Echobase

A one-sentence summary

Echobase is a code-free AI platform for knowledge management, designed for businesses and teams. Users can upload documents to create folders of information, and then develop custom AI agents that are capable of answering questions, generating text, and performing analyses.

Tool Introduction

Echobase is operated by Core Flow Intelligence, and its main purpose is to enable teams to utilize generative AI for working with internal documents. It brings document management, knowledge bases, Agent configurations, chat histories, and organizational permissions together in a single web-based workspace.

The official website uses the term “training an Agent,” but the actual process is more akin to indexing files into a searchable knowledge base, after which the model uses that information to provide answers. It is not the same as users training or fine-tuning a basic large model on their own.

Main functions

Document knowledge base

Users can create folders and upload local files; these folders represent the scope of knowledge accessible to the Agent. The official documentation confirms that DOCX, PDF, CSV, and TXT formats are supported, with a maximum file size of 50 MB per set.

• Upload local files by dragging and dropping, then wait for the indexing process to complete.
• Use different folders to isolate projects, departments, or business topics.
• When creating an Agent, select an existing dataset to serve as the knowledge base.
• Each package is also subject to a limit on the total amount of data on the page; it’s not just the 50MB limit per request that applies.
• Web scraping and connections to Google’s and Microsoft’s document systems are still indicated in the official documentation as features that will be released soon.

Knowledge quiz

Users can ask questions about the documents in a specified folder, and the Agent will search for relevant information to generate answers. This method is suitable for finding regulations, project documents, reports, and internal guidelines, but it is still necessary to verify the original files.

Content generated based on data

Echobase is not only used to find answers; it can also be used to create report paragraphs, proposals, emails, and other business-related content based on the knowledge base. The content generated may be similar to that produced by other users, and it is necessary to verify facts, copyright issues, and confidentiality requirements before making such content public or using it for commercial purposes.

Data aggregation and analysis

Agents can identify themes, compare information, or extract key points across different documents, making them useful for reviewing projects, conducting policy analysis, and preparing reports. When dealing with numbers, conclusions, or conflicting documents, it is necessary to review the original documents manually.

Predefined Agent Types

According to the official documentation, the platform offers more than 50 pre-defined agent types, covering areas such as consulting, operations, marketing, and finance. These pre-defined types enable quick setup of roles and tasks; however, users should still adjust the system prompts and advanced parameters according to their own needs.

Model selection

The current help documentation lists models such as GPT-4o and Claude Opus, but the model names, versions, and available packages may change depending on the underlying services. The Professional version specifies access to GPT-4, while the Enterprise version lists the latest AI models.

Channel display

When creating an Agent, it is possible to enable Show Sources, so that the responses indicate the source document from which the information comes. Displaying the source helps with tracking, but the model may still misinterpret the citation location and the meaning of the original text.

Suggestions for subsequent questions

Follow-up Prompts generate directions for further questions based on the current conversation, helping users who are not familiar with prompt wording to refine their tasks step by step. It is still up to the user to decide whether the questions align with the actual business objectives.

Voice input

Speech to Text allows users to enter questions using their microphone, making it suitable for quickly recording spoken requests. Speech recognition may misinterpret proper nouns, numbers, and names; therefore, the text should be checked before sending it.

Short-term dialogue memory

The agent retains the short-term context of the current conversation in order to understand pronouns and subsequent questions. In long conversations, the context may be lost or misinterpreted; a new chat session should be started when the topic changes significantly.

Chat history

Users can view the historical conversations of a specific Agent, access the full context, and delete sessions that are no longer needed. In an organizational environment, chat records may be shared; therefore, it is important to avoid entering information that should not be seen by other members in shared Agents.

Team and role permissions

The Enterprise plan allows for the creation of organizations, inviting members, and sharing Agents, folders, and chat history. The available roles are Owner, Admin, Member, and Read and Chat Only; each role comes with different permissions regarding creation, deletion, and chatting.

Input and output

Working principle

1. Users register an account and choose a suitable free or paid plan.
2. Create folders in the Files area to organize documents by project, department, or business topic.
3. Upload DOCX, PDF, CSV, or TXT files; the platform will process and index their content.
4. Create a new Agent in the Dashboard, and select the model, preset type, and corresponding dataset.
5. Enable additional features such as channel display, follow-up questions, or voice input.
6. Set system prompts and advanced parameters to define the role, tone, and scope of tasks for the Agent.
7. Send representative questions to assess responses, channels, and performance on complex tasks.
8. The team plan involves reconfiguring member roles and access permissions, as well as continuously cleaning up files and chat histories.

Usage tutorial

Create the first knowledge base agent

1. Register for an Echobase account and log in to check the query limits, Agent limits, and page limits for the current plan.
2. Go to the Files page and create a folder with a clear name.
3. Drag the documents for which you have permission to use into the folder; each batch of files should not exceed 50MB.
4. Wait for the upload and indexing to complete, then check for failed files, duplicate files, and expired versions.
5. Return to the Dashboard and select Create New Agent.
6. To name the Agent, select the model, preset type, and the dataset that was just created.
7. Enable channel display, follow-up questions, and voice input as needed.
8. Configure system prompts to clarify that responses should be based solely on the available information, and to specify the procedure to follow when something cannot be confirmed.
9. Use at least 5 real-world questions for testing, and then have target users participate in a trial.

Organize team collaboration

1. Confirm that the Enterprise plan supporting Organizations is being used.
2. In Settings, go to Organization and create an organization.
3. Invite members’ email addresses based on job requirements, and do not grant excessive permissions at once.
4. Assign Admin to managers, and assign Member or a read-only chat role to regular users.
5. Create folders and agents by department to prevent all information from being concentrated in the same knowledge base.
6. Use non-sensitive examples to check which Agents, folders, and chat records each role can actually see.
7. Access is revoked immediately upon a member’s departure or the completion of a project, and any shared content that is no longer needed is deleted.

Improve the reliability of responses

1. Only upload the latest, clear files with clear source information; delete duplicate or outdated versions.
2. Split different topics into separate folders to reduce the interference from irrelevant search results.
3. The system message instructs the Agent to acknowledge that insufficient information is available and not to guess answers.
4. Enable channel display and allow verification of the original text for important answers.
5. Test the accuracy using questions with known answers, and then add cases with no answers or contradictory information.
6. Record failure issues and update files, prompts, or Agent settings.
7. Conclusions regarding high risk are reviewed by professionals before being used externally.

Prices and packages

As of August 22, 2026, the official pricing page lists four monthly subscription plans: Free, Individual, Professional, and Enterprise, all priced in US dollars. Users can change or cancel their plans; the final fees, quotas, and settlement rules are subject to those specified on the purchase page.

Pricing considerations

• The query quota is calculated on a monthly basis; different models may consume varying amounts of quota or be subject to different restrictions.
• The benefits of GPT-4 listed on the pricing page start with the Professional tier; free users and Individual users should not, by default, have access to the same model.
• Enterprise includes 5 seats; whether there is an additional charge for adding members should be confirmed before settlement.
• The page capacity is expressed in terms of the number of pages, and a single upload is limited to 50MB.
• Integration with Google Drive and Microsoft is still marked as upcoming and cannot be used as a current feature for purchasing Professional.
• The official website may change the models, quotas, and prices; it is necessary to review the settlement page again before making a purchase.

Which users are it suitable for

• Consulting team: Searches for information related to project materials, analyzes the topics, and prepares a draft report.
• Operations team: Provide agents with the policies, procedures, and operation manuals to answer daily questions.
• Marketing team: Creates draft content based on brand materials and compiles research materials.
• Financial and business analysts: Extract key points from CSV files and reports, but it is necessary to verify the numbers.
• Project team: Different projects are separated by folders, allowing team members to share the same knowledge Agent.
• Small and medium-sized enterprises: They wish to create internal document Q&A systems without coding, rather than developing RAG systems on their own.

Typical use cases

• Convert the employee handbook and policy documents into an internal system Q&A agent.
• Search historical reports for relevant project experience and create a new proposal framework.
• Identify common themes, differences, and issues to be clarified across multiple research documents.
• Create training Q&A and onboarding materials based on the standard procedures.
• Allow members of the consulting project to share databases, tips, and conversation histories.
• Draft marketing copy based on the approved brand materials, which is then reviewed and released by the responsible person.
• Extract business insights from CSV files and documentation, and manually verify values and perform calculations.

Product advantages

• It is possible to create a dedicated knowledge Agent from a folder without having to write any code.
• It supports Q&A, content creation, and cross-document analysis, offering more than just keyword search.
• The channel display can help users return to the relevant documents in order to verify their answers.
• The predefined agent types reduce the barriers to configuring tasks related to consulting, operations, marketing, and finance.
• The team solution offers hierarchical roles that allow for distinguishing between management, creation, and read-only chat permissions.
• The prices for the four tiers are listed clearly, showing query capacity, agent capacity, and page capacity, which makes it easy to make a choice based on scale.
• The terms state that the user retains ownership of the input and receives a transfer from the platform of the rights to the output.

Usage restrictions and precautions

• The agent’s responses may contain errors, omissions, and incorrect references; important information must be verified by referring to the original document.
• The supported file types are currently limited to DOCX, PDF, CSV, and TXT; other formats have not been officially approved.
• Each batch of uploaded files can be up to 50MB in size, and the package also imposes limits on the total number of pages that can be processed.
• Long conversations may result in the loss of context; complex tasks should be broken down and new chats initiated when necessary.
• Web scraping, as well as synchronization with Google and Microsoft, are still planned features and cannot be purchased using the capabilities that are already available.
• The so-called training mainly involves retrieving information from a knowledge base; it does not result in the creation of new foundational models specific to individual users.
• The official model descriptions may be outdated compared to the upstream versions; the actual models available in the account should be taken as the definitive reference.
• The free plan offers a very limited amount of resources, making it suitable only for testing and not for maintaining stable business processes.
• No complete public developer API, SDK, or Webhook documentation was found; the system integration capabilities need to be confirmed with the official sources.

Privacy, security, and data processing

The terms of service state that uploaded files are stored in an AWS environment and used for extracting information and providing answers. The privacy policy also specifies that account communications, usage data, device information, logs, cookies, and analysis data are collected, and these data may be processed by the service provider.

• The official terms state that user content will not be used to improve the services, but it may be processed for the purpose of providing those services, ensuring compliance with the law, and enforcing policies.
• The official website states that external models accessed through API calls do not use the collected data to train their models; companies should still verify the actual sub-processors and the relevant contracts.
• The personal information of international users may be transmitted to facilities and servers in Australia.
• Some operations, folders, and chat histories within an organization may be visible to other members.
• When handling personal information covered by GDPR or CCPA, the regulations require contacting the relevant authorities to implement the data processing protocols.
• Before uploading, passwords, keys, irrelevant personal information, and unauthorized customer data should be removed.
• The privacy policy does not specify the exact retention period for various types of data; companies should obtain written confirmation prior to making a purchase.
• Users should establish procedures for member departures, document updates, session deletions, and permission reviews.

Content ownership and commercial use

Within the limits permitted by law, the user retains ownership of the content they enter; by complying with the terms, Echobase transfers to the user its rights regarding the generated output. This arrangement does not guarantee that the output will not infringe upon the rights of third parties.

• Users must ensure that the files they upload have legitimate permission for use.
• The generated results may be the same as or similar to those of other users, and absolute exclusivity cannot be claimed.
• Before commercial use, it is necessary to check the referenced materials, trademarks, personal information, and third-party copyrights.
• The terms prohibit presenting AI-generated content as if it were created by humans, and such content must be disclosed in accordance with the applicable rules.
• When professional opinions, client deliverables, or public reports are involved, records of manual review should be kept.

Platform support and integration

APIs, SDKs, and open-source status

The Echobase terms include APIs, software, and developer services within the scope of the services offered; it also states that data extraction is possible when the API permits it. However, the current help center does not provide any information regarding authentication, endpoints, request structures, or rate limits – that is, no details on how to access these services publicly.

Therefore, the catalog should be labeled as “Unconfirmed public API”, rather than being described as an API available for development. No official SDK, open-source code repository, or open-source license was found; as such, the Echobase platform should be regarded as a commercial service that is not open source.

Basic information

Recommendation score

The recommendation score is 4.0 out of 5. Echobase provides an intuitive way of combining folders, knowledge agents, channel displays, predefined roles, and team permissions; it is suitable for small and medium-sized enterprises as well as consulting teams that wish to use internal resources without having to write any code.

The main drawback is that the updates to the public documentation are not consistent; some integrations are indicated as being available soon, and the details regarding public APIs and data retention are unclear. Before making a formal purchase, it is necessary to check the models, quotas, permissions, and current status of functions using a trial account.

Frequently Asked Questions

Is Echobase free?

There is a free plan: the price shown on the current page is $0 per month, with 20 queries to the basic model and 20 pages of upload capacity. It is suitable for testing purposes, but not for long-term, frequent use.

How much is the lowest priced subscription plan?

The Individual plan costs $19 per month and includes 2,100 queries per month, 5 agents, and 10,000 pages of upload capacity. The actual price is as indicated on the billing page.

Which file formats are supported?

The official documentation confirms that DOCX, PDF, CSV, and TXT formats are supported. The maximum size of a single file set uploaded is 50 MB, and the total amount of data that can be uploaded is limited by the plan chosen.

Can it synchronize directly with Google Drive or Microsoft Files?

It should not be considered a formal feature at the moment. The official pricing page and documentation still list Google, Microsoft, SharePoint, and related synchronization tools as features that will be released soon.

Does Echobase really train a new model?

From the perspective of the public process, it primarily converts documents into a knowledge base that can be searched by agents, after which existing large models generate responses. Users do not receive an independently fine-tuned base model as a result.

Is Chinese supported?

The price comparison shows multi-language support, and the large models used are generally capable of handling Chinese as well. However, specialized Chinese terms, numbers in tables, and scanned PDF files still need to be tested using actual data.

Can it show the channels for answering?

It is possible to enable Show Sources, so that the answers indicate the relevant documents. The hints provided by these channels can help with verification, but they do not guarantee that the citations and conclusions are entirely accurate.

Are team permissions available?

Enterprise supports organizations with up to 5 seats, and offers roles such as Owner, Admin, Member, and Read and Chat Only. Folders and chat history can be shared within an organization.

Are APIs provided?

The terms mention API, but no complete documentation on public authentication, endpoints, and invocation methods is available; therefore, it cannot be confirmed that ordinary users can access it directly. It is necessary to verify the conditions for access with the official party before making a purchase.

Is Echobase open source?

It is not open source; no official source code repository or open-source license has been found, and the terms of service prohibit decompilation as well as the identification of the underlying components.

Will user-generated content be used for training?

The terms of service state that users’ content will not be used to improve the services, and the official website also indicates that external models accessed through interfaces are not trained using the data collected. Companies should still verify their current sub-processors, data retention periods, and data processing agreements.

Can the generated content be used for commercial purposes?

The terms transfer the platform’s rights over the output to compliant users, but the users are still required to ensure that the input materials are legal and to verify the copyright, trademarks, personal information, and accuracy of the content in the output.

Summary

Echobase is ideal for quickly turning corporate documents into knowledge agents that can be used for answering questions, generating text, and performing analyses. Folder isolation, channel display, pre-defined agents, and team roles enable non-development teams to create basic knowledge workflows as well.

Before deciding to pay, it is essential to verify the quality of responses from genuine documents, the current model in use, the page quota, the scope of visibility within the organization, as well as any integrations that are not yet available. When customer or employee data is involved, written confirmations regarding data retention, deletion, and third-party processors are also required.