← Sticker Wall AGENT · DETAIL

Dojo Predecessor Tessl Testing Agent Infrastructure Subscription: A 100k RMB/Month Enterprise Quality Mid-Office Business

Workflow: Receives change events and requirement documents from enterprise code repositories daily, automatically generates test p

AGENT

Key Fields

FIELD STAMPS

IndustrySaaS / Enterprise Software

RegionChina（中国大陆）

ScaleSME

ChannelOnline

🔧 Workflow

Receives change events and requirement documents from enterprise code repositories daily, automatically generates test plans and cases, and dispatches them to cloud sandboxes for execution. After regression testing is complete, automatically summarizes failed cases, uses large language models to locate root causes, and pushes fix recommendation reports to development chat groups. Inputs are code commits and requirement PRDs, and outputs are test reports and root cause analysis documents. Every morning, it automatically pulls the latest code and executes full regression, at noon pushes test reports to WeChat Work, in the afternoon automatically generates root cause analysis based on failed cases for developers to investigate, and in the evening iterates testing parameters based on feedback.

🛠 Setup Requirements

Requires 5+ years of development and testing experience, with familiarity in API testing and CI/CD pipelines. On the tech stack side, must know how to use AI testing platforms like Tessl or TestSprite, and possess the ability to interface with WeChat Work or Lark bots for report pushing. Going from understanding Tessl's Agent Enablement Platform philosophy to running the first enterprise pilot takes about 2-4 weeks. Needs familiarity with Tessl's Spec-driven Development philosophy, the ability to describe testing intent using natural language, and mastery of API integration methods for tools like TestSprite. It is recommended to run a small-scale validation in an internal project first before expanding to the full lifecycle.

🧰 Toolchain

🔧 Tessl
🔧 TestSprite
🔧 Postman
🔧 GitLab CI
🔧 Docker
🔧 Cursor

💰 Revenue

Charged based on enterprise scale. Small and medium-sized SaaS clients have an annual subscription fee of 50,000-100,000 RMB, along with a one-time setup fee of 20,000-30,000 RMB. Estimated based on serving 2-3 clients simultaneously, monthly revenue can reach around 100,000 RMB. Specific pricing reference: Micro teams (<20 people) 50k/year, medium teams (20-100 people) 80k/year, large enterprises customized on demand 100k-200k/year. Setup fee ranges from 20,000 to 30,000 RMB depending on process complexity.

💸 Cost

Tessl's free version has a basic quota; usage beyond that is billed by call volume. LLM API costs are estimated at 500-1500 RMB/month based on 50 calls per day, cloud sandbox execution fees are about 500-1500 RMB/month, totaling around 2000-3000 RMB per month for a single project. If self-built GPUs or higher frequency calls are used, costs will rise.

⏱ Time Investment

About 40 hours per week, of which client on-site interviews and requirement confirmation account for 20 hours, testing Agent tuning accounts for 15 hours, and report delivery and client communication account for 5 hours. Each client takes about 4-6 weeks from deployment to acceptance.

🚀 Getting Started

Step 1: Register a Tessl account, import a public GitHub repository, describe a business scenario using natural language (e.g., 'user login should return user info'), and observe the generated test cases and execution results. Step 2: Input a Swagger URL into TestSprite to generate an API test report. Step 3: Package the usage of both platforms into a service package, publish practical experience in tech communities to attract the first batch of SME clients interested in AI testing.

🔑 Keys to Success

✅ Focus on high-frequency, rigid-demand scenarios like API integration and regression testing, with quantifiable value
✅ Reports must include failure root cause localization, not just a list of test results
✅ Accumulate industry test case template libraries to form compoundable assets
✅ Deeply integrate with CI/CD pipelines, embedding into existing development processes rather than reinventing the wheel
✅ Maintain compatibility with clients' existing testing processes to lower team learning costs
✅ Establish a traceable audit log of test results to enhance enterprise compliance confidence

⚠️ 风险

⚠️ AI-generated test cases may produce a large number of false positives, requiring human engineers to act as the final judge
⚠️ LLM-generated test cases carry hallucination risks and may miss critical business logics, requiring manual spot-check fallback mechanisms
⚠️ The trust-building cycle for clients regarding AI testing is long; historical defect data must be used for retrospective validation to facilitate renewals
⚠️ AI testing agents may generate test cases that do not match business logics, requiring continuous fine-tuning of model prompts and business rule libraries

📌 Real Cases

📌 Formerly known as Modex, Tessl was valued at over $1 billion during its financing in 2025, transitioned to an Agent Enablement Platform in January 2026, has a makerstack rating of 7.3/10, and starts for free
📌 In domestic 2026 actual tests of TestSprite, inputting a Swagger URL for backend API testing automatically generates test cases and returns reports, with coverage improvements reflected in automatically supplementing boundary scenarios more comprehensively than human design
📌 In the AI-native test generation cases showcased at the 2026 Singularity Conference, when a natural language requirement such as 'after a user uploads a PDF, the system should return structured JSON within 3 seconds' is inputted into the testing Agent, it automatically completes the closed loop of understanding, mutation, verification, and feedback reinforcement

Sources