← Sticker Wall AGENT · DETAIL

Building a Sana-like Enterprise Knowledge Q&A Agent: Onboarding Support + Policy Retrieval, Monthly Subscription Model Generating 45k/Month

Workflow: Spend about 2 days per week onboarding a new client: collect employee handbooks, product documentation, and FAQs; clean

AGENT

Key Fields

FIELD STAMPS

IndustrySaaS / Enterprise Software

RegionChina

ScaleSME

ChannelOnline

🔧 Workflow

Spend about 2 days per week onboarding a new client: collect employee handbooks, product documentation, and FAQs; clean and import them into Dify or Coze knowledge bases; build a Q&A agent that cites sources; and integrate it with WeChat Work or Lark bots. When employees ask questions, the agent automatically retrieves and cites the original text. I review logs of unanswered questions daily to update the knowledge base and provide clients with weekly reports on Q&A coverage and new knowledge entries.

🛠 Setup Requirements

No programming required. Use Dify or Coze to build the knowledge base application, then configure WeChat Work or Lark bots for group access and permissions. Core skills include document structuring, basic prompt engineering, and knowledge segmentation optimization. From tool registration to first delivery takes about 2 weeks. You can start with just a laptop; it is recommended to practice with a company you know to refine the process before taking paid orders.

🧰 Toolchain

🔧 Dify Knowledge Base
🔧 WeChat Work Bot
🔧 Coze
🔧 Lark Base

💰 Revenue

① SME Knowledge Q&A Agent Monthly Subscription (Primary Revenue): Clients pay a monthly service fee. Small teams under 50 people pay around 1,500 RMB, while companies with ~200 employees pay up to 4,000 RMB. With 12 stable clients, monthly income reaches 30k–45k RMB (avg. 2,500–3,750 RMB per client), accounting for ~100% of revenue. ② Multi-Knowledge Base Expansion: For ~200-person companies, additional fees apply based on the number of knowledge bases. ③ Industry-Specific Template Licensing: Licensing templates to peers in catering, bookkeeping, or cross-border e-commerce; pricing and volume data are undisclosed. ④ Opportunity: Per-seat subscription benchmarking Sana ($30 per user/month): Targeting companies under 200 people without dedicated IT; revenue potential is currently unquantified.

💸 Cost

Professional subscriptions for Dify or Coze plus large model API usage cost 500–1,500 RMB per month. WeChat Work bot integration is free, and the free version of Lark Base is sufficient. Total costs remain under 10% of revenue.

⏱ Time Investment

New client onboarding takes ~2 days, including document cleaning, database setup, and debugging. Maintenance requires 1–2 hours daily for reviewing logs, updating knowledge, and responding to clients, plus half a day weekly for coverage reports.

🚀 Getting Started

First, build a free demo using an employee handbook from your own or a friend's company. Record a video showing the agent instantly answering 10 consecutive new-hire questions to demonstrate the 'before vs. after' effect. Post case studies in local entrepreneur communities, HR groups, or Maimai. Offer the first order at a discounted 999 RMB to secure real data and referrals. After 3 successful cases, raise prices and organize industry-specific templates.

🔑 Keys to Success

✅ Store knowledge bases by department and product line; answers must cite original sources and page numbers to reduce trust risks.
✅ Proactively report unanswered questions weekly and update the database; this is the core hook for subscription renewal.
✅ Engage HR managers and administrative leads as the decision-makers; send weekly reports directly to management rather than just serving frontline employees.
✅ Develop industry-specific templates (e.g., catering, bookkeeping) to allow partial reuse for new clients, reducing delivery costs over time.

⚠️ 风险

⚠️ Outdated knowledge bases causing incorrect answers could lead to client complaints or liability; establish a monthly update mechanism and include a disclaimer that the latest official documents take precedence.
⚠️ Internal IT or SaaS vendors developing similar features may capture large enterprise clients; focus on SMEs under 200 people without dedicated IT and secure annual contracts.
⚠️ Incorrect answers regarding sensitive topics like salary or leave could cause labor disputes; default to human HR intervention for compensation-related queries.

📌 Real Cases

📌 Sana Labs was founded by Joel Hellermark in Stockholm in 2016. It focuses on enterprise knowledge integration and AI Q&A. It raised $34M in Series B (2022), an additional $28M (2023), and $55M in Series C (2024), covering compliance training, onboarding, and customer education.
📌 Sana claims its platform connects all enterprise data without coding to build custom AI agents for search, Q&A, and automation, proving that low-code knowledge Q&A is a replicable technical path for individuals.
📌 Sana's AI assistant features enterprise search, meeting summaries, and Q&A, with ISO 27001 certification and single-tenant architecture. This highlights that enterprise clients have strict data security requirements; individual contractors should prioritize SMEs with lower data sensitivity.

Sources