Question 1

What counts as a run?

Accepted Answer

A run is a single execution of a recipe or workflow. If a workflow has 5 recipe steps, that counts as 1 workflow run. Scheduled and webhook-triggered executions each count as runs.

Question 2

Can I change plans later?

Accepted Answer

Yes. You can upgrade or downgrade your plan at any time. When upgrading, you get immediate access to the new features. When downgrading, the change takes effect at the end of your billing cycle.

Question 3

Do I need my own API keys?

Accepted Answer

Yes. JieGou connects to LLM providers using your own API keys (BYOK). This gives you full control over costs and ensures your data stays between you and the provider. Separately, Pro plans and above can create SDK API keys for headless execution — these authenticate programmatic access to recipes, workflows, and the coding agent via the SDK endpoint, and are distinct from LLM provider keys.

Question 4

What LLM providers are supported?

Accepted Answer

All plans include Anthropic (Claude Sonnet, Haiku, Opus), OpenAI (GPT-5.x, o3, o4-mini), and Google (Gemini 2.5/3). Pro and Team plans add 4 first-class providers — Mistral AI (Large, Medium, Small), Groq (ultra-fast inference), xAI (Grok), and any OpenAI-compatible endpoint — plus a model recommendation engine. Team and Enterprise plans add AWS Bedrock (Converse API with SigV4 signing) and Azure OpenAI (deployment-based endpoints) for regulated environments. Enterprise also adds Ollama and vLLM auto-discovery for local inference and a certified open-source model registry (Llama 4, DeepSeek, Qwen, Mistral). Each provider has full BYOK support, per-provider circuit breakers, and key validation. You can choose a different model for each recipe or workflow step.

Question 5

What are Knowledge Bases and Knowledge Sources?

Accepted Answer

Knowledge Bases let you upload documents and import URLs that JieGou processes into searchable chunks for RAG (Retrieval-Augmented Generation). Knowledge Sources are external connectors — enterprise search (Coveo, Elasticsearch, Glean), vector databases (Pinecone, Vectara), workspaces (Confluence, Notion, Google Drive), and help desks (Zendesk, Guru) — that give your AI agents 30+ MCP tools for grounded responses. The Starter plan includes up to 3 knowledge source connectors; Pro and Team plans include unlimited connectors; Enterprise adds custom source integrations with priority support.

Question 6

What happens when I exceed my plan limits?

Accepted Answer

On the Starter plan (20 recipes, 500 runs/month, 1 department pack), hard caps prevent usage beyond your allocation. On Pro and Team plans, most resources have overage rates so you can keep working — for example, additional runs cost $0.01 each, extra team members cost $5/month each, and connector calls beyond your included allocation cost $0.02 each. You can set a spending cap to control maximum overage charges. Enterprise plans have custom limits negotiated for your needs.

Question 7

How does token pricing work?

Accepted Answer

JieGou uses a Bring Your Own Key (BYOK) model — you provide your own LLM API keys and pay providers directly for token usage. The Pro plan includes $25 of platform token value per month (covering orchestration overhead). Beyond that, a small margin is applied to cover infrastructure costs. You can monitor token usage and costs in the Billing dashboard.

Question 8

Is there a free trial for Pro?

Accepted Answer

Yes. The Pro plan comes with a 14-day free trial. No credit card required to start. You can downgrade to Starter at any time if it does not fit your needs.

Question 9

What is the difference between Pro and Team?

Accepted Answer

Pro is designed for teams up to 10 people who need advanced automation — including the Coding Agent, SDK API access, skills, conversation branching, bakeoffs, batch runs, and quality monitoring with Mistral, Groq, and xAI providers. Team is for mid-size teams of 10–25 who additionally need DAG workflow orchestration, custom webhook tools, lifecycle hooks, AWS Bedrock and Azure OpenAI providers, canary rollouts, advanced approval policies, screen sharing, follow mode, and workflow bakeoffs. If your team is growing past 10 people or you need extensibility and enterprise provider support, Team is the right choice.

Question 10

How does team collaboration work?

Accepted Answer

JieGou includes real-time presence indicators, contextual chat on runs and recipes, and platform-wide team chat. Pro plans get presence and chat. Team adds screen sharing, session recording, follow mode (co-browsing with cursor tracking), and advanced approval workflows. Enterprise adds SSO/SAML and full RBAC with user groups.

Question 11

What are bakeoffs?

Accepted Answer

Bakeoffs let you compare recipes, models, or entire workflows side by side. Run the same inputs through different configurations and use LLM-as-judge scoring to find the best option. You can also set up live A/B tests that route traffic between variants and auto-stop when a winner is statistically significant.

Question 12

Do knowledge bases use my token allowance?

Accepted Answer

Yes. When you create a knowledge base by crawling a website or uploading documents, JieGou processes the content using LLM calls (for summarization, chunk analysis, and table-of-contents detection). These calls count toward your token usage. If your token allowance is exhausted, new knowledge base creation will be paused until the next billing period or until you upgrade your plan.

资源	入门版	专业版	团队版	企业版
执行次数	500/月硬性上限	2,000/月超额： $0.01/次	10,000/月超额： $0.01/次	无限
配方	20 硬性上限	无限	无限	无限
工作流程	6 硬性上限	无限	无限	无限
团队成员	3 硬性上限	10 超额： $5/人/月	25 超额： $5/人/月	无限
部门套件	1 硬性上限	500 超额： $0.01/份	2,000 超额： $0.01/份	无限
文件	50 硬性上限	5 GB 超额： $0.50/GB/月	25 GB 超额： $0.50/GB/月	客制超额：可议
储存空间	100 MB 硬性上限	20 超额： $1/排程/月	100 超额： $1/排程/月	无限
排程	2 硬性上限	20 超额： $1/触发器/月	100 超额： $1/触发器/月	无限
触发器	2 硬性上限	1,000/月超额： $0.02/次	5,000/月超额： $0.02/次	50,000/月超额： $0.01/次
连接器调用	100/月硬性上限	Unlimited	Unlimited	Unlimited
Knowledge sources	3 硬性上限	—	—	—

越用越聪明的 AI 自动化

入门版

专业版

团队版

企业版

使用您自己的 API 金钥

方案比较

免费开始——比您预期的更大方

为企业级可靠性而生

断路器

死信伫列

分散式追踪

停滞工作流程看门狗

并行控制

Prometheus 指标

比较 AI 自动化成本

常见问题