Feishu Aily is an enterprise-level Agent platform launched by Feishu, which deeply calls Feishu cloud documents, multi-dimensional tables, task systems and other internal knowledge bases. Supports MCP protocol, knowledge retrieval and question answering, in-depth thinking, brand customization, and model privatization deployment.

飞书Aily

Feishu Aily’s core parameters and statistics

Feishu Aily is an enterprise-level AI agent platform launched by Feishu (ByteDance). It is officially positioned as "Easy, capable, trustworthy" - it is not an independent chat robot product, but an AI employee platform deeply embedded in Feishu's collaborative ecosystem. It can directly call the enterprise's cloud documents, multi-dimensional tables, task systems, project data and other internal assets.

Projects	Public Information
Product positioning	Enterprise-level AI Agent platform, deeply integrated with Feishu ecosystem
Core Competencies	Knowledge retrieval Q&A, in-depth thinking MCP access, enterprise customization, intelligent customer service AI quality inspection
Deployment method	SaaS (embedded in Feishu Enterprise Edition) + model private deployment
Integrated Ecology	Feishu Cloud documents, multi-dimensional tables, Feishu tasks, Feishu projects, Feishu knowledge base
Customer Cases	Bull Group, Meiyijia, Inovance Technology, Haichen Energy Storage, Taikang Group, Lantu Automobile, etc.
Developer	Feishu (ByteDance)
Latest version	2025-07 (continuous iteration)
Supported platforms	Web, Feishu client (desktop + mobile)
Supported languages	zh-CN
Place of Belonging	CN

Brief review in one sentence: Feishu Aily is not a general chatbot, but an enterprise-level AI employee who has grown up in Feishu. It can directly call your documents, project data and business systems, while providing enterprise-level AI application development platform capabilities.

Users and market recognition of Feishu Aily

Feishu Aily's market recognition mainly comes from the large-scale adoption by Feishu's enterprise customers, rather than the volume competition of independent products. Its core verification dimension comes from quantitative data in public cases of many leading companies.

Quantitative endorsement of customer cases:

Customer	Industry	Deployment size	Key figures
Bull Group	Manufacturing	Covering multi-channel customer service system	AI interception rate 85%, average monthly conversations 70,000+ times, question response reduced from 30 minutes to 22 seconds, user satisfaction 99.76%
Meiyijia	Chain retail	1,500+ instructors, 40,000+ stores	200,000+ cumulative consultations, 40,000+ monthly average, response time reduced from 33 seconds to 14 seconds, saving 94 man-days per month
Inovance Technology	Industrial Technology	Enterprise One-Stop Service Desk	Single consultation efficiency increased by 60%, AI interception rate increased by 21.38%, annual working hours saved over 10,000 hours, satisfaction rate 99.2%
Haichen Energy Storage	New Energy	Global Solution Center	The skill acquisition time for new employees is reduced from 2 years to 4 months, the response time is reduced from 480 seconds to 25 seconds, 83% of man-hours are released, and the accuracy rate exceeds 95%

From the perspective of customer structure, Aily's typical adopters are medium and large enterprises that have intensively used Feishu, covering 12+ industries such as advanced manufacturing, chain retail, new energy, finance and insurance, and policy consulting. Customer cases generally show three common characteristics: High-frequency repeated consultations are concentrated, the knowledge base asset inventory is sufficient, and the enterprise has strong compliance requirements for data security—this happens to be an internal enterprise scenario that is difficult for general-purpose AI chatbots to penetrate.

Enterprise-level ecological barriers: Compared with independent AI Agent platforms (such as Dify), Aily's market barrier does not lie in its leading technical parameters, but in the depth of integration of "Feishu native" - ​​it can read Feishu cloud documents, multi-dimensional tables, task lists, and project dashboards without additional docking. This constitutes a natural switching cost for enterprises that have already used Feishu. But this also means that Aily’s market ceiling is directly limited by Feishu’s penetration rate in the enterprise collaboration market.

Feishu Aily’s cost advantage

Feishu Aily's cost structure continues the "enterprise-level function packaging" model. It is not sold independently to individual users and is deeply tied to Feishu Enterprise Edition.

C client/individual users: The independent version is not available for individuals. Aily is a functional component of Feishu Enterprise Edition, and individual users cannot subscribe separately. This is completely different from the free C-side strategy of general AI assistants (such as DeepSeek, Doubao) - Aily's value proposition is to improve internal efficiency of the enterprise, rather than as a personal daily assistant.

Enterprise/Team Cost: Aily is included in Feishu Enterprise Edition service. The specific pricing is subject to the real-time page of Feishu official website. You need to contact Feishu Business to get a quote. Enterprises can choose different deployment packages according to their needs:

Package level	Implementation threshold	Typical scenarios	Reference value
L1 entry level: department-level knowledge guide	Extremely low, online in 1 day, zero code	Scenarios with many rules, fast updates, and many new people, such as sales tactics and complex policy Q&A	Solve the problem of "finding answers" efficiency, new people can directly @ the robot to produce standard answers in seconds
L2 Intermediate: Internal business intelligent customer service	Medium, 1-2 weeks to go online, SOP needs to be sorted out	Cross-department cooperation, expert intervention is required, and scenarios with processing time requirements	Solve the efficiency of "cross-department services", SLA can be tracked, support efficiency dataization
L3 high-end: All-domain enterprise-level intelligent services	Higher level, requiring in-depth business knowledge compilation and API development	Facing external customers, needing to connect to core business systems, and large-scale data analysis	Not only a customer service tool, but a "data command cabin" that can automate quality inspection and customer insights

The design of the three-tier package is essentially ladder risk control - enterprises can first use L1 to quickly verify the effect in a single department, and then gradually expand to L2/L3. This "from point to surface" path lowers the decision-making threshold for enterprise AI trial and error.

API/Developer Cost: Aily does not offer standalone API sales. Developers access Aily's tool chain through the Feishu open platform and MCP protocol. Private deployment of models (such as access to third-party models such as DeepSeek R1) requires additional evaluation. The relevant costs depend on the selected model, deployment scale and computing power configuration, and are subject to the Feishu business plan.

Hidden Cost Tip:

• Data preparation cost: The quality of Aily's Q&A is highly dependent on the completeness and structure of the knowledge base. If the company's historical documents are scattered, have inconsistent formats, and are stagnant in updates, manpower needs to be invested in sorting out and annotating knowledge. This hidden cost is often underestimated in the early stages.
• Customized development costs: L3-level global deployment requires API docking and business system integration, and the investment of the IT team cannot be ignored.
• Model deployment cost: Private deployment options (such as DeepSeek R1, etc.) require additional GPU computing power and operation and maintenance capabilities, and this cost is not covered by Feishu standard subscription.

Main functions of Feishu Aily

Feishu Aily's functional system is divided into two layers: "personal work efficiency improvement" and "enterprise-level AI application development". The same platform supports both usage modes.

Personal Agent capability layer

• Personal Agent: Each Feishu user has an exclusive Aily work assistant, which can complete knowledge query, document summary, task reminder, schedule and other operations directly through conversations on the Feishu client. Supports natural language questions without learning specific command formats.
• Proactive Work: Aily can proactively push important information according to preset rules, such as project progress change reminders, pending approval notifications, and key document update reminders. The value of this capability is to upgrade AI from "passive response" to "active assistant" - users do not need to remember to ask, Aily will appear when needed.
• Personalized Memory: Aily can remember the user's working habits and preferences, and provide more accurate personalized recommendations after long-term use. For example, remember the knowledge areas that users often consult and proactively push relevant documents when they are updated.
• Skill Library: Provides preset AI skill templates that users can enable on demand without the need to configure them from scratch. Skills include document translation, meeting minutes generation, automatic writing of weekly reports, data analysis and other high-frequency tasks.

Enterprise-level AI application capability layer

• Smart Support: Build a 7×24-hour AI customer service robot based on the enterprise knowledge base, supporting multi-modal question and answer via text, voice, and pictures. Automatic quality inspection can be implemented through Webhooks, and AI responses can be marked and evaluated in batches. Supports multi-channel unified services (Feishu Group, WeChat public account IM tools, etc.), and can manage all service entrances in a single backend. Expert View: The core difference lies in the "Feishu native" channel opening - the customer service robot can be called @ within the Feishu group, without jumping to a third-party platform, and the conversation context can be directly associated with the Feishu CRM and work order system, realizing the relationship from consultation to resolution.
• Smart Coaching: For training and onboarding scenarios, AI automatically generates practice questions and simulated conversations based on the corporate knowledge base, so new employees can practice in real business situations. The case of Haichen Energy Storage shows that this capability shortens the time for new employees to master business skills from 2 years to 4 months.
• Smart Tickets: Automatically extract key information from conversations to generate tickets, supporting intelligent classification, label extraction and prioritization. Linked with Feishu's multi-dimensional forms, the creation, circulation, and tracking of work orders are fully automated. The "living water mechanism" is its unique design - AI automatically labels and categorizes historical work orders, automatically summarizes and reviews manual customer service conversations and stores them in the database, and continuously feeds back the knowledge base.
• AI Quality Inspection: Upgraded from manual random inspection to AI full quality inspection, covering every customer service conversation. AI can detect illegal speech, identify service attitude problems, and discover potential risk points. In the case of Bull Group, the quality inspection coverage was realized from 0 to 1, the problem interception rate was stable at 60%, and the overall service accuracy rate reached 86.64%.
• AI Marking and Inspection: Automatically mark and classify business content to achieve structured management of massive data. AI inspection can regularly check the accuracy of the knowledge base

and timeliness, automatically marking outdated content and notifying updates.

Hidden linkage: functional synergy

The real differentiation of Feishu Aily is not a single function, but the complete synergy between functions. A typical workflow is:

Customer service dialogue → Automatic quality inspection scoring → Refining high-frequency issues → Update knowledge base → Feed back customer service accuracy

Each customer conversation is not only a question and answer, but also completes the triple tasks of quality inspection, knowledge mining and knowledge base updating. In the traditional model, these procedures require customer service, quality inspection, and knowledge management teams to execute them separately. Aily compresses them into one automatic procedure. For enterprises, this means that the implicit benefits of reducing cross-departmental coordination costs may be more valuable than the explicit efficiency gains.

Feishu Aily’s model and version evolution

As an enterprise-level platform product, Feishu Aily's version iteration rhythm is different from the "large version number" model of consumer-level AI products. Instead, it continues to evolve following the function release rhythm of Feishu Enterprise Edition.

Version context

Time node	Version identification	Key milestones
~2024-01	Early preview version	Basic Agent capabilities and knowledge base integration, targeted testing for Feishu enterprise customers
~2024-Q3	Capability expansion period	Enhanced knowledge retrieval and question and answer capabilities, access to Feishu Cloud documents and multi-dimensional tables
~2025-01	Deep integration period	MCP protocol support goes online, opening up exclusive tool chains such as Feishu projects and Feishu tasks
~2025-07	Current	Support model privatization deployment (DeepSeek R1, etc.), corporate brand customization, accurate usage analysis, intelligent customer service/training/work order/quality inspection full scenario capabilities

Key development context:

• Phase 1 (~2024): Verify the basic usability of AI Agent within the Feishu ecosystem. The core goal is to "be able to answer questions."
• Phase 2 (~2025-01 to ~2025-07): Expand from question and answer to workflow automation, introduce the MCP protocol to achieve deep integration of the tool chain, and upgrade Aily from an "information query tool" to a "task execution platform".
• The third phase (~2025-07 to present): Implementation of enterprise-level customization capabilities - brand customization, model privatization deployment, usage analysis, etc., while launching four major enterprise solutions: intelligent customer service/training/work orders/quality inspection, transforming from "tools" to "industry solutions".

Version Note: The version number of Feishu Aily is based on the release timeline of Feishu Enterprise Edition. The above dates are milestone inferences based on public information. There is currently no official precise version number naming system. For specific version information, please refer to Feishu’s official announcement.

Openness of model access

Unlike closed AI platforms, Aily supports enterprises to access third-party models for private deployment (such as DeepSeek R1, etc.), which means that enterprises can flexibly choose the underlying model based on cost, performance, and compliance requirements without being locked into a single model supplier. This capability is particularly critical for industries such as finance and government affairs that have strict compliance requirements for model sources.

Feishu Aily’s technical advantages

Main Type Judgment: Agent / MCP / Automation Tool - Enterprise-level AI Agent platform, which has the dual attributes of Agent orchestration platform and business-end application.

Architecture link

The technical architecture of Feishu Aily follows the two-layer design of "Feishu Ecosystem Embedded + MCP Protocol Extension":

User portal (Feishu client/Web)
        │
        ▼
┌────────────────────────────────┐
│ Feishu Aily Agent Platform │
│ ┌─────────┐ ┌─────────────┐ │
│ │ Personal Agent│ │ Enterprise AI Application │ │
│ │ (Q&A/ │ │ (Customer Service/Sparring/ │ │
│ │ Memory/ │ │ Work order/quality inspection) │ │
│ │ Active push)│ │ │ │
│ └────┬─────┘ └──────┬───────┘ │
│ │ │ │
│ ▼ ▼ │
│ ┌─────────────────────────┐ │
│ │ MCP protocol layer │ │
│ │ (Tool discovery/call/authentication) │ │
│ └────┬──────┬──────┬───────┘ │
└────────┼──────┼──────┼────────────┘
         │ │ │
         ▼ ▼ ▼
   ┌──────┐ ┌────┐ ┌──────┐
   │Feishu │ │Feishu │ │Feishu │
   │Cloud Documents │ │Projects │ │Multidimensional │
   │MCP │ │MCP │ │Table │
   └──────┘ └────┘ └──────┘
         │ │ │
         ▼ ▼ ▼
   ┌───────────────────────────┐
   │Enterprise knowledge base/business system │
   │ (Private

Deployment/cloud) │
   └───────────────────────────┘

Control flow: User instructions → Aily Agent platform (intent recognition/task disassembly) → MCP protocol layer (tool selection/parameter generation) → specific Feishu MCP tools (document reading/table query/task creation) → result aggregation and return.

Data backflow: The results of each interaction will flow back to the knowledge base and quality inspection system, forming a process of continuous self-optimization.

Tool open list (Tool behavior)

The specific tool behaviors that Feishu Aily exposes to large models through the MCP protocol include but are not limited to:

Tool name	Function description	Corresponding Feishu ability
feishu_doc_read	Read Feishu Cloud document content	Knowledge retrieval and Q&A
feishu_doc_search	Search cloud documents by keywords	Knowledge management and discovery
feishu_sheet_query	Query multi-dimensional table data	Data analysis and visualization
feishu_sheet_append	Append data to multidimensional tables	Data entry and summary
feishu_task_create	Create Feishu tasks/to-dos	Task execution
feishu_task_query	Query task list and status	Progress tracking
feishu_project_read	Read Feishu project information	Project management
feishu_calendar_query	Query schedule	Time management
feishu_knowledge_search	Enterprise knowledge base semantic search	Enterprise internal FAQ
feishu_message_send	Send Feishu messages/notifications	Active push and reminder
feishu_bot_reply	AI robot replies to user messages	Intelligent customer service
feishu_ticket_create	Create a ticket	Work order management
feishu_quality_check	Dialogue quality inspection scoring	AI quality inspection

Example included: The user asked "Help me check the project progress of Q2 and send a weekly report to the team members." Aily's Agent process may be: feishu_project_read (check project progress) → feishu_sheet_query (check multi-dimensional table data) → feishu_doc_search (check weekly report template) → content generation → feishu_message_send (send a message).

Engineering Pitfall Guide

Based on the operating characteristics of the enterprise-level Agent platform, the following are issues that need to be paid attention to in actual deployment and operation and maintenance:

1. The "garbage in, garbage out" trap of the knowledge base: The quality of Aily's answers directly depends on the quality of the knowledge base data. If corporate documents contain a large amount of out-of-date information, confusing formatting, and irregular naming, the AI's answer may seem "convincing" but contain incorrect content - this "illusion of confidence" is more damaging than being unable to answer. Solution: Complete the governance and quality review of the knowledge base before the official launch, and set up a regular inspection mechanism; use Aily's AI inspection capabilities to automatically mark suspicious content; set up manual review points (Human-in-the-loop) in key business scenarios.

2. Context drift in multiple rounds of dialogue: In a long customer service dialogue, Aily may gradually deviate from the initial focus of the question, or lose earlier key constraints after multiple rounds of questioning. This is especially prominent in complex business scenarios at the L2/L3 level. Solution: In the prompt design, explicitly require the Agent to summarize the key context at the beginning of each round of dialogue; set a maximum dialogue round limit (such as 20 rounds) and then automatically switch to manual; key decision-making nodes require user confirmation before execution to avoid "one-stop" blind operations.

3. Permission Boundary Governance of MCP Tools: Aily’s Feishu tool chain that can be operated through the MCP protocol covers a wide range of areas (document reading and writing, task creation, message sending). If permissions are configured improperly, AI may perform operations beyond expectations (such as accidentally deleting documents and sending messages by mistake). Solution: Follow the principle of least privilege and allocate Aily's tool access scope according to job roles; set up a secondary confirmation barrier for irreversible operations (deletion, payment, publishing, batch messages); enable Aily's operation audit log in the Feishu management background, and perform regular playback to check whether the AI ​​behavior is within the expected range.

4. Response bottleneck under high-frequency concurrency: When all employees of an enterprise use Aily at the same time (such as HR Q&A during peak attendance periods, customer service consultation during major promotions), response delays or timeouts may occur. Solution: Communicate resource allocation with Feishu Business in advance based on the estimated concurrency peak; enable automatic elastic expansion in the L2/L3 deployment plan; set a degradation policy—automatically switch to the preset standard reply when Aily's response times out.

Technical barriers to "Feishu native"

The core difference between Feishu Aily and general agent platforms (such as Coze and Dify) lies in the deep binding of the data layer. The general Agent platform needs to manually connect to Feishu through API, and every step must deal with issues such as authentication, data format conversion, rate limit, etc.; while Aily runs directly on Feishu's data infrastructure, and documents, tables, and knowledge bases themselves are the Agent's "natural database". The actual difference this brings is that it may only take a few hours to create a new intelligent customer service in Aily from configuration to online, while it may take several days to complete the same Feishu integration on a common platform.

The opposite of this deep integration is technology lock-in: Aily's capability boundaries are Feishu's capability boundaries. If you need to connect to systems outside the Feishu ecosystem (such as external CRMs such as SAP and Salesforce), the integration complexity of Aily will increase significantly. In this case, the general Agent platform will have more advantages.

How to use Feishu Aily

The usage portal of Feishu Aily is deeply bound to Feishu Enterprise Edition. The getting started path is simple but requires early configuration by the administrator.

Quick activation process

Role	Steps	Description
Enterprise administrator	Log in to the Feishu management backend → Select "Application Management" → Enable "Aily Work Assistant"	You can configure Aily's access rights, knowledge base connection scope and brand image in the management backend
Administrator (optional)	Add Aily work assistant to Feishu client navigation bar	Convenient for employees to access directly in the sidebar
Corporate employees	Search "Aily Work Assistant" on the Feishu client → start a conversation	No need to install additional applications, it can be used within Feishu
Web access	Open aily.feishu.cn in the browser → Scan QR code to log in with Feishu	Synchronize with client functions

3 minutes to get started quickly (deployment side)

Enterprise developers can configure Aily's MCP connection through the Feishu open platform. The following is a typical MCP Server configuration fragment (taking claude_desktop_config.json as an example, applicable to the scenario of connecting the Aily tool chain through the MCP protocol):

{
  "mcpServers": {
    "feishu-aily": {
      "command": "npx",
      "args": [
        "@feishu/mcp-server-aily",
        "--app-id", "<YOUR_FEISHU_APP_ID>",
        "--app-secret", "<YOUR_FEISHU_APP_SECRET>"
      ],
      "env": {
        "FEISHU_AILY_BASE_URL": "https://aily.feishu.cn"
      }
    }
  }
}

Note: The above code is a typical mounting mode based on the Feishu open platform MCP protocol. For the actual available MCP Server package names and parameters, please refer to the Feishu developer documentation.

Summary of deployment methods

Deployment level	Implementation method	Required resources	Applicable stages
L1 department level	One-click activation of Feishu management backend and uploading knowledge documents	No development required, business colleagues can complete it by themselves	Quickly verify the effect of AI Q&A
L2 internal business	Configure workflow, set manual transfer rules, and connect Feishu multi-dimensional tables	Need to sort out internal SOP, business + IT cooperation	Expand to cross-department business processes
L3 full-domain service	API docking with external systems, privatized model deployment, data analysis configuration	Requires in-depth business knowledge + IT development	Full-link solution for external customers

Product pricing of Feishu Aily

Feishu Aily is not priced independently, but is provided with the main product as a functional module of Feishu Enterprise Edition.

Pricing Structure:

Fee Type	Description	Reference Information
Feishu Enterprise Edition Subscription	Aily’s basic functions are included	The real-time pricing on Feishu’s official website shall prevail
L1 department-level deployment	No additional cost, configure the knowledge base to use	Zero additional cost
L2 internal business deployment	Value-added services that may involve workflow configuration	Business confirmation required
L3 global enterprise-level deployment	Model privatization deployment API integration, custom development	Feishu business quotation
Third-party model privatization	Such as DeepSeek R1, etc., which require additional GPU computing power	Subject to the quotation of the model supplier

Cost Assessment Framework: The actual expenditure of the enterprise should be composed of three parts: "Feishu Enterprise Edition subscription fee + value-added service fee (L2/L3) + model deployment fee (optional)". Among them, Feishu Enterprise Edition subscription is the basic threshold, and the latter two depend on the enterprise's deployment scale and customization depth.

Hidden cost considerations:

• Human investment in knowledge base governance: If the enterprise's historical documents are scattered and require a dedicated team to organize and structure the knowledge, this cost may exceed the software subscription fee.
• Ongoing operating costs: The effect of AI does not end once it is launched online. It requires continuous updating of the knowledge base, monitoring of answer quality, optimization of prompts, and the arrangement of dedicated personnel.
• Model upgrade cost: If the connected third-party model releases a major version, it may need to be retested and adapted.

Application scenarios of Feishu Aily

The application scenarios of Feishu Aily revolve around "knowledge-intensive and process-intensive tasks within the enterprise". The following four types of scenarios have been verified on a large scale.

Scenario 1: Intelligent customer service and knowledge center

This is Aily’s most in-depth scene yet. Enterprises connect internal knowledge bases (product manuals, policy documents, FAQs, and training materials) to Aily, and front-line employees or external customers can obtain accurate answers directly through dialogue.

• Typical uses: dealers query product parameters, store employees inquire about promotion rules, HR answers salary policies, IT handles network fault repair reports.
• Quantitative deduction of cost reduction and efficiency improvement: Taking the Bull Group case as a reference, the AI ​​interception rate of 85% means that only 15 of every 100 customer service requests require manual intervention; the response time is compressed from 30 minutes to 22 seconds. For a medium-sized enterprise (average 1,000 customer service requests per day), this is equivalent to releasing about 850 manual processing times per day, which is equivalent to the workload of about 5-8 full-time customer service personnel.
• Key points of verification: Whether the coverage and accuracy of the knowledge base meet business requirements; whether the rule of switching to manual work when AI cannot answer is reasonable; the quality of question and answer in multilingual scenarios.

Scenario 2: Employee Onboarding and Training

Using Aily's "intelligent coaching" capability, new employees can quickly become familiar with business knowledge through conversational learning instead of traditional centralized training.

• Typical uses: New salespeople learn product vocabulary, new customer service staff familiarize themselves with common problem handling procedures, and new operations learn platform operation manuals.
• Quantitative Reference: In the case of Haichen Energy Storage, the time for new employees to master basic business skills was shortened from 2 years to 4 months, and support working hours were reduced from 6 hours to 1 hour per day.
• Key points for verification: Whether the training content is customized based on the company's own business scenarios; whether there is a quantitative evaluation mechanism for learning effects; whether manual coaching is needed in complex skill scenarios.

Scenario 3: Work order management and process automation

Aily can automatically identify user needs and create work orders during conversations, and combines Feishu's multi-dimensional forms and task systems to complete the entire process from submission to tracking.

• Typical uses: IT fault repair reporting (automatically extracting equipment information, fault description, and urgency), HR onboarding process (automatically creating approvals, assigning workstations, and opening accounts), legal contract review (automatically extracting key terms and marking risk points).
• Quantitative deduction of cost reduction and efficiency improvement: Assume that the company has 200 internal service work orders every day. Each work order takes an average of 5 minutes for employees to fill in and 10 minutes for administrators to process. AI automatic form filling can reduce employees' form filling time by 80% (from 5 minutes to 1 minute), and intelligent routing can reduce the administrator's processing time by 30% (from 10 minutes to 7 minutes), saving a total of approximately 13 person hours per day.
• Key points of verification: accuracy of work order classification; completeness of automatically created work order information; depth of integration with existing business systems.

Scenario 4: AI quality inspection and compliance monitoring

Aily can conduct full AI quality inspection on customer service conversations, sales recordings, and service records, replacing the traditional manual random inspection mode.

• Typical uses: Identify illegal expressions in customer service speech, detect whether sales commitments exceed the scope of authorization, and discover potential risks in the service process.
• Quantitative Reference: In the case of Bull Group, the quality inspection was upgraded from manual inspection (2 hours/item) to full AI coverage (0.5 hour/200 items). The efficiency at the quality inspection level was improved by 800 times; the problem interception rate was stable at 60%.
• Key points of verification: The accuracy and false positive rate of the quality inspection model; the recognition ability in complex scenarios (dialects, industry terms, etc.); the integrated mechanism between quality inspection results and subsequent improvement measures.

Scene adaptation matrix

Dimensions	Best fit	Second best fit	Poor fit
Industry	Advanced manufacturing, chain retail, finance and insurance, policy consulting, Internet	Medical care (requires professional certification), education (personalized teaching)	Highly creative industries (design, advertising)
Scale	Medium/large enterprises with more than 200 employees	Growth enterprises with 50-200 employees	Small and micro enterprises with less than 50 employees
Feishu usage level	Fully using Feishu Office	Partially using Feishu	Not using Feishu
Data readiness	Existing structured knowledge base	Documents are scattered but can be organized	No digital knowledge assets
Core requirements	High-frequency repetitive Q&A, internal process automation	Simple knowledge query	In-depth original content generation, complex negotiations

Applicable groups of Feishu Aily

The positioning of Feishu Aily determines that its applicable group has clear prerequisites—enterprises have already used Feishu and have a certain digital foundation.

Applicable people

• Enterprise Knowledge Management Team (KM/Documentation Team): Responsible for structuring the implementation of enterprise knowledge assets (product manuals, policy documents, training materials, historical cases) into the Aily knowledge base, and maintaining the quality of answers through continuous updates. Aily's AI inspection and automatic marking capabilities can significantly reduce the human investment in daily operation and maintenance of knowledge bases - only one person is responsible for tuning and maintaining a million-word business corpus.
• IT System Administrator and Developer: Responsible for Aily's deployment configuration, MCP tool chain connection, enterprise-level model privatized deployment and system integration. L2/L3 level deployment requires in-depth participation of the IT team, including API docking, permission configuration, audit log management, etc.
• Customer Service and Operations Team Leader: Use Aily’s intelligent customer service, intelligent quality inspection, and intelligent work order capabilities to optimize the service process. The core benefit of these roles is not to be "replaced by AI", but to be freed from tedious repetitive work and focus on high-value tasks that require human judgment - such as complex customer complaint handling and service quality improvement planning.
• HR and Training Department: Use intelligent sparring capabilities to improve employee onboarding efficiency and training coverage. Aily can automatically generate exercises and simulated conversations to implement personalized learning paths based on actual business scenarios.
• Enterprise executives and business line leaders: Through Aily's usage analysis and quality inspection data dashboard, understand the team's service efficiency, high-frequency problem distribution and service quality trends, and provide data support for management decisions.

Not suitable for the crowd

• Non-Feishu users: Aily is deeply bound to the Feishu ecosystem and cannot be used independently. It is not applicable to enterprises or individual users who do not have Feishu.
• Individual users: Aily does not have a personal paid version, and is aimed at enterprise team collaboration scenarios. AI assistants used daily by individuals (such as DeepSeek, Kimi, Doubao) are more suitable for such needs.
• Enterprises in need of in-depth original content: Aily's knowledge retrieval and Q&A are based on existing knowledge bases, and are good at "retrieving existing information" rather than "creating new content". If AI is needed for brand copywriting creativity, long-form in-depth reporting, novel creation, etc., direct interaction with general large models is more suitable.
• Small and micro enterprises that are extremely sensitive to AI costs: The subscription fee of Feishu Enterprise Edition plus the deployment and operation costs of Aily may not be cost-effective for teams of less than 50 people. Such companies can start with free general-purpose AI tools and then evaluate Aily as they scale.

Summary and Outlook

Feishu Aily represents the "deep integration route" for the implementation of enterprise-level AI Agents - instead of building a new AI platform outside the existing system, AI capabilities are embedded as native modules into the collaboration systems already used by enterprises on a daily basis. The advantages and limitations of this approach are clear.

Core Competitive Position: Aily's irreplaceability comes from the depth of data integration of "Feishu Native". If the knowledge retrieval, work order automation, customer service quality inspection, intelligent sparring and other functions it implements within the Feishu ecosystem are implemented manually through the general Agent platform + Feishu API, the integration cost and time will be at least an order of magnitude higher. For enterprises that have already made extensive use of Feishu, Aily provides the shortest path from "owning data" to "data-driven efficiency".

Current Core Advantages:

• Native integration of Feishu ecosystem, allowing you to call internal assets such as cloud documents, multi-dimensional tables, task systems, etc. with zero configuration
• Laddered deployment design (L1→L2→L3), enterprises can gradually promote AI implementation with low risk
• Quantifiable effects demonstrated in customer cases (interception rate 85%, response time shortened to 22 seconds, annual man-hour savings exceeding 10,000 hours)
• Support model privatization deployment to meet the data security requirements of high-compliance industries such as finance and government affairs
• MCP protocol support, expandable Feishu exclusive tool chain

Major Current Limitations:

• Completely bound to Feishu ecosystem, non-Feishu users cannot use it, the market ceiling is limited by Feishu penetration rate
• The quality of answers is highly dependent on the completeness and timely updating of the enterprise's knowledge base, which requires continuous operational investment.
• Integration capabilities for external systems outside the Feishu ecosystem (such as SAP and Salesforce) are weaker than those of the general Agent platform
• The cost structure of Enterprise Edition subscription fees and high-end deployment (L3) is not transparent enough and requires business communication
• The accuracy of AI quality inspection and marking in complex industry terms and multi-lingual mixed scenarios has not been fully verified in public cases.

Follow-up observation points:

• Whether Feishu Aily will launch an independent enterprise version subscription plan to lower the adoption threshold for companies that are not full users of Feishu
• Whether the MCP ecosystem will be open to third-party tools other than Feishu (such as Google Workspace, Microsoft 365, etc.)
• Whether customer cases will expand from current large enterprises to mid-sized markets to verify replicability in enterprises of different sizes

Procurement and Adoption Risk Assessment:

• For enterprises that are already using Feishu, the adoption risk of Aily is low - it can be piloted in one department starting from L1, without the need to purchase additional hardware or software, and the effects can be quantified before expansion. The recommended path is to select a department (such as IT support HR service desk) with a large amount of manual consultation and relatively complete knowledge assets as a pilot. After running for 2-4 weeks, evaluate the three core indicators of interception rate, response time and user satisfaction. After reaching the standards, it can be expanded to customer service, sales, training and other scenarios.
• For enterprises that are considering Feishu, Aily should be included in the evaluation of Feishu selection - it may become a key capability that drives enterprises to switch from free collaboration tools to Feishu. It is recommended to start Aily's PoC (proof of concept) simultaneously during Feishu's trial operation to verify the effect with real business data.
• For teams that already use competing AI Agent platforms (such as Coze and Dify) and have deeply integrated with Feishu, they need to evaluate the trade-off between "the ecological integration benefits brought by switching to Aily" and "the cost of migrating existing Agent platforms". It is recommended to select 3-5 core automation scenarios, implement them on Aily and the existing platform respectively, and compare the time cost and operating effect from configuration to launch.
• Before formal procurement, all enterprises should confirm: whether the SLA of Feishu Enterprise Edition covers Aily services; whether the technical requirements for model privatization deployment match the existing IT infrastructure; whether the commercial authorization scope and usage restrictions of knowledge base data are clear.

Related tools: crewai, langchain

Version Info

• FeishuAily current ：Supports knowledge retrieval, in-depth thinking, enterprise customized MCP, and model privatization deployment. （~2025-07）
• Feishu Aily preview ：Early preview version, basic Agent capabilities and knowledge base integration. （~2024-01）