CC Bots · 从 Claude Code 到千行百业 | AI 编程共创 · 私有部署

Q: 为什么是 Claude Code？

Claude Code 是能自主完成“理解需求→规划→调工具→执行→自我验证→修正”闭环的工程代理。Anthropic 自己公布全公司 100% 员工每天都在用它工作；它完成的任务里有 27% 是人们没有这个工具时根本不会去尝试的。它让非专业团队也能在带教下把 AI 用进真实业务。

Q: 数据安全吗？要自己买硬件吗？

全程私有部署，数据不出企业。bot 主要由云端模型 API 驱动，本地机器负责编排与本地任务（文件 / OCR / 检索），一台高端工作站即可起步（整机参考价见硬件成本矩阵）。硬件由企业自购自有、模型 token 自付。

引言 · 真知晶球The vision · the palantír

一颗叫 palantír 的水晶球，被搬进了现实A crystal ball called palantír — made real

托尔金《魔戒》里那颗能看见千里之外的水晶球，七十年后被一家公司搬进现实，还直接以它命名——帕兰蒂尔 Palantir，把混乱的信息变成决策的能力，被华尔街称作“美国国运股”。而真正赢得这场竞争的，是 Anthropic 的模型 Claude。2026 年 6 月，Anthropic 估值首次反超 OpenAI，胜在“商业化纪律”：自第一天起，便将模型嵌入企业的核心生产流程。Tolkien's far-seeing crystal ball was made real seventy years later by a company that named itself after it — Palantir, turning chaos into decisions, dubbed an "American sovereignty stock." The weapon that actually won the war is Claude, by Anthropic. In June 2026 Anthropic's valuation overtook OpenAI's — winning on commercial discipline: embedding models into core production from day one.

5/5

维全反超 OpenAI
估值·收入·利润·产品·人才dims overtaken
value·rev·profit·product·talent

2×

企业级编码市场份额
vs OpenAIenterprise coding share
vs OpenAI

100%

Anthropic 员工
每天用 Claude Codeof staff use
Claude Code daily

✓五角大楼行动的背后，已出现模型的影子（据《华尔街日报》）Models now shadow Pentagon operations (per WSJ)

✓连最强国产对手 DeepSeek，也处处对标 Claude CodeEven DeepSeek, the strongest domestic rival, benchmarks against Claude Code

✓它完成的任务里，27% 是“没有它根本不会尝试”的27% of its completed tasks are ones people wouldn't have attempted without it

✓这件武器不在硅谷、不在华尔街——回家亲手就能装上This weapon isn't in Silicon Valley or Wall Street — you can install it yourself

完整故事见演讲 → Full story in the talk → 《从 Claude Code 到千行百业From Claude Code to Every Industry》 · 竞争数据据公开报道整理，用于趋势说明competitive figures compiled from public reports, for trend illustration

私有部署 · 硬件成本矩阵Private deployment · hardware cost matrix

算力部署于贵公司本地，投入成本清晰透明Compute deployed in-house — a clear, transparent cost structure

CC Bots 全私有部署，数据不出企业。这些 bot 主要由云端模型 API 驱动——本地机器只负责编排 + 本地任务（文件 / bash / OCR / 本地检索）。下面是一台高端工作站的整机参考价（京东，2026-06）。CC Bots is fully private — data never leaves the company. The bots are driven mainly by cloud model APIs; the local machine only handles orchestration + local tasks (files / bash / OCR / local retrieval). Below is a reference build for a high-end workstation (JD.com, Jun 2026).

性价比首选Best value

Intel Ultra7 270K

¥13,879 整机·不含显卡base · no GPU

20C/20T · PassMark 多核 ≈59,000 / 单核 4,650 · Z890 DDR5 64G · 2T NVMe · Xe 集显+NPU
看重单核响应 + 多开选它，多核性价比最高（4.25 分/元）20C/20T · PassMark multi ≈59,000 / single 4,650 · Z890 DDR5 64G · 2T NVMe · Xe iGPU+NPU
Pick for single-core response + many concurrent; best value (4.25 pts/¥)

性能旗舰Performance flagship

Intel Ultra9 285K

¥16,079 整机·不含显卡base · no GPU

24C/24T · PassMark 多核 67,271 / 单核 5,050 · Z890 DDR5 64G · 2T NVMe · Xe 集显+NPU
看重多核吞吐选它，多花 ¥2,200 换多核 +15%24C/24T · PassMark multi 67,271 / single 5,050 · Z890 DDR5 64G · 2T NVMe · Xe iGPU+NPU
Pick for multi-core throughput; +¥2,200 buys +15% multi-core

🔮不确定贵公司该选哪台？点击右侧蓝色晶球，输入部署规模（Bots 数量 + 数据量），自动定位到匹配机型。下方配置表如转盘般滚动，停在为你推荐的那一档。Not sure which build fits? Click the blue orb, enter your scale (number of bots + data volume), and the table below — turning like a reel — stops at the match for you.

整机配置（含内存）Full build (incl. RAM)

Ultra7 270K

Ultra9 285K

适用Fit

基础（Xe 集显 + NPU）Base (Xe iGPU + NPU)DDR5 64G · 2T NVMe

¥13,879

¥16,079

纯编排 + 轻本地任务Orchestration + light local

RTX 4090 24G ×1DDR5 64G

¥36,899

¥38,079

本地 OCR / 中小模型Local OCR / mid models

RTX 4090 24G ×1DDR5 96G

¥38,979

¥40,979

OCR · 更大检索内存OCR · more retrieval RAM

RTX 4090 48G ×1DDR5 64G

¥42,899

¥44,079

大显存单卡Big-VRAM single card

RTX 4090 48G ×1DDR5 96G

¥44,999

¥46,979

大显存 · 推荐起步Big-VRAM · recommended

RTX 5090 32G ×1DDR5 64G

¥48,899

¥50,079

新一代单卡Next-gen single card

RTX 5090 32G ×1DDR5 96G

¥50,979

¥52,979

新一代 · 大内存Next-gen · more RAM

RTX 4090 24G ×2DDR5 64G

¥59,899

¥61,579

多容器 + 并发推理Multi-container + concurrent

RTX 4090 24G ×2DDR5 96G

¥61,993

¥64,479

多容器并发 · 大内存Concurrent · more RAM

RTX 5090 32G ×2DDR5 64G

¥84,379

¥86,079

重负载本地大模型Heavy local LLM

RTX 5090 32G ×2DDR5 96G

¥86,499

¥88,979

重负载 · 大内存Heavy · more RAM

RTX PRO 6000 96G ×1DDR5 64G

¥111,879

¥113,079

旗舰大显存推理Flagship big-VRAM

RTX PRO 6000 96G ×1DDR5 96G

¥113,979

¥115,979

旗舰 · 大内存Flagship · more RAM

RTX 5090 32G ×2DDR5 128G

以京东实际报价为准Per JD live quote

超大内存 · 重负载Max RAM · heavy

RTX PRO 6000 96G ×1DDR5 128G

以京东实际报价为准Per JD live quote

超大内存 · 旗舰推理Max RAM · flagship

配置表持续滚动展示全部 15 档；点击蓝色晶球，按规模为你停在匹配机型。The reel cycles all 15 builds; click the blue orb to stop at the one matching your scale.

说明：配置表已含 DDR5 64G 与 96G 两档整机参考价；128G 暂以京东实际报价为准。多数部门日常活与编排无需独显，基础机即可起步；只有重负载本地大模型 / 大批量 OCR 才建议另配独立 GPU 主机。硬件由企业自购自有、长期常开更省电（Arrow Lake 带 NPU）；模型 token 按云端套餐 / 充值由客户自付，不计入上表。Note: the reel covers full-build prices at both DDR5 64G and 96G; 128G is quoted per JD's live price. Most departmental work and orchestration need no discrete GPU — the base build is enough; only heavy local LLMs / bulk OCR warrant a dedicated GPU host. Hardware is bought and owned by the company (Arrow Lake's NPU is power-efficient for always-on); model tokens are paid by the customer and not counted above.

三层台阶 · 落到你的企业Three steps · into your company

三个层次：从日常的活，到“造工具”的能力Three layers: from daily work to the power to build tools

同一套 Claude Code 底座，三层递进——从各部门的日常工作，到核心流程的智能体定制，再到自主构建工具的开发元能力。One Claude Code foundation, three ascending layers — from everyday departmental work, to custom agents for core processes, to the meta-capability of building tools independently.

各部门技能库构建Dept skill-library build

A · 日常业务DAILY WORK

AI FDE 核心业务AI FDE · core business

B · 核心流程CORE PROCESS

CC 开发能力CC dev capability

C · 开发元能力META-CAPABILITY

自上而下层层收束——各部门的日常工作，逐层汇聚为“用 AI 造工具”的开发元能力Converging top-down — everyday departmental work distilled, layer by layer, into the meta-capability of building tools with AI

A · 日常业务DAILY WORK

各部门日常技能库Departmental skill library

各职能部门的日常工作——沿“龙虾场景”的 5 个维度逐项延展，持续沉淀为贵公司的企业技能库，让每个部门都能以 AI 完成本职工作。Everyday work in any function — extended along the "lobster scenario" 5 dimensions and continually distilled into your enterprise skill library, so every team handles its own work with AI.

市场Marketing采购Purchasing产品Product客服Support财务Finance人力HR

各部门技能库系列培训 →Dept skill-library training →

B · 核心流程CORE PROCESS

AI 驻场工程师 FDEFDE on-site engineer

关键业务流程的定制开发——如帕兰蒂尔一般，由一位 AI 驻场工程师（FDE）将 AI 深度嵌入贵公司最关键、最不容差错的核心流程。驻场方案与报价按项目评估后提供。Custom development for critical processes — as Palantir does, a Forward Deployed Engineer (FDE) embeds AI into your most critical, error-intolerant core flows. The on-site plan and quotation are provided after a project assessment.

FDE 驻场FDE on-site数据不出企业Data stays in-house工作流智能体Workflow agents

FDE 驻场服务详情 →FDE on-site details →

C · 开发元能力META-CAPABILITY

用 AI 造工具的元能力The power to build tools

把“用 AI 造工具”这件事本身，长成你公司自己的 Claude Code 开发元能力。先以一个真实项目跑通，再把能力复制出去——外贸独立站、短视频内容生产线、研发知识库与推理、ERP AI Native 升级，都只是其中的场景。Grow "building tools with AI" itself into your company's own Claude Code meta-capability. Prove it on one real project, then replicate — an export site, a short-video production line, an R&D knowledge base or an ERP AI-Native upgrade are just scenarios.

外贸独立站Export site短视频内容生产线Video production line研发知识库与推理R&D knowledge baseERP AI NativeERP AI-Native你自己的工具链Your own toolchain

看落地场景 →See the scenarios →

层Layer	服务项Service	价格Price	与 CC Bots 的关系Relationship to CC Bots
A日常业务Daily work	各部门技能库系列培训Dept skill-library training	¥25,000 / 5 次·赠 6 工位/ 5 sessions · +6 seats	在 CC Bots 上沉淀部门技能库，含 CC Bots 工位Builds the dept skill library on CC Bots; includes CC Bots seats
B核心流程Core process	AI FDE 驻场定制智能体AI FDE custom agents	按项目评估By assessment	定制智能体运行于 CC Bots 底座，嵌入核心流程Custom agents run on CC Bots, embedded in core processes
C开发元能力Meta-capability	开发能力共创（外贸站 / 内容生产线 / 研发推理 / ERP 等场景）Dev-capability co-creation (export site / video line / R&D / ERP, etc.)	按场景评估By scenario · 外贸站 ¥20,000 起· export site from ¥20,000	掌握以 CC Bots（Claude Code）自主构建工具的能力Master building tools independently with CC Bots (Claude Code)
底座Base	CC Bots 多工位入口 SKUCC Bots multi-seat SKU	Free ¥0 / Pro ¥39 起 / Team ¥29 工位·月起Free ¥0 / Pro ¥39+ / Team ¥29/seat+	A·B 两层服务共同的运行底座（私有部署）The shared runtime foundation for layers A and B (private)

三项服务均以 CC Bots 为统一底座；硬件由企业自购自有、模型 token 自付。各层详情见解决方案与报价总览。All three services share CC Bots as their foundation; hardware is owned by the company and model tokens are paid by the customer. See Services and Pricing.

越往上走，从“以 AI 完成工作”到“以 AI 构建工具”——越是竞争对手难以复制的护城河。The higher you climb, from doing the work with AI to building tools with AI — the deeper a moat competitors can't replicate.

FAQ

常见问题FAQ

CC Bots 到底交付什么？What does CC Bots actually deliver?

不是外包一个项目，而是带你的团队用 Claude Code 把 AI 嵌入企业的三层：A 各职能部门日常活沉淀成企业技能库；B 由 FDE 驻场为核心流程定制智能体；C 把“用 AI 造工具”长成公司自己的开发元能力。最终留在公司的是能持续进化的能力，不是一份交付物。Not an outsourced project — we bring your team to embed AI into three layers with Claude Code: A daily departmental work into an enterprise skill library; B FDE on-site custom agents for core processes; C growing "building tools with AI" into your own meta-capability. What stays is an evolving capability, not a deliverable.

为什么是 Claude Code？Why Claude Code?

它是能自主完成“理解需求→规划→调工具→执行→自我验证→修正”闭环的工程代理。Anthropic 自己公布全公司 100% 员工每天都在用它；它完成的任务里有 27% 是人们没有这个工具时根本不会去尝试的。详见 Claude Code 能力页。It's an engineering agent that runs the full understand→plan→tool→execute→self-verify→fix loop. Anthropic says 100% of staff use it daily; 27% of its completed tasks wouldn't have been attempted without it. See the capabilities page.

FDE 驻场怎么报价？How is FDE priced?

FDE 驻场的方案与报价按项目评估后提供：我们先深入了解贵公司的核心流程、数据与目标，评估驻场方案与人天，再出具正式报价方案。详见 FDE 驻场服务。The on-site plan and quotation are provided after a project assessment: we first study your core processes, data and objectives, size the plan and person-days, then issue a formal quotation. See FDE on-site.

数据安全吗？要自己买硬件吗？Is data safe? Do I buy hardware?

全程私有部署，数据不出企业。bot 主要由云端模型 API 驱动，本地机器负责编排与本地任务，一台高端工作站即可起步（见上方硬件成本矩阵）。硬件由企业自购自有、模型 token 自付。Fully private — data never leaves the company. Bots are driven by cloud model APIs; the local machine handles orchestration and local tasks, and one high-end workstation gets you started (see the hardware matrix). Hardware is owned by the company; tokens are paid by the customer.

开始Get started

结合贵公司的行业场景，将 AI 系统化嵌入每一层Embed AI, systematically, into every layer of your operations

请提供贵公司的部门、核心流程与目标，我们将提供从 A 到 C 的定制化共创与报价方案。Share your departments, core processes and objectives, and we'll prepare a tailored A-to-C co-creation plan and quotation.

查看报价总览See pricing 了解解决方案Solutions

CC Bots 开发者介绍About the CC Bots developer

大卫 — AI 解决方案架构师David — AI Solution Architect

毕业于厦门大学Xiamen University

20亿+

Claude Code tokens
深度实战Claude Code tokens
hands-on depth

3300+

工程会话
真实项目打磨engineering sessions
real projects

✓企业落地：CC Bots 智能体已落地多家企业，私有部署、关键数据不出企业，微信就能用Enterprise: CC Bots agents deployed across companies — private deployment, data stays in-house, usable right inside WeChat

✓科研赋能：助力国家级研究团队，目标项目成功获批Research: supported a national-level research team; the target project was successfully approved

扫二维码，添加我为朋友Scan the QR code to add me on WeChat

联系人：Contact: 大卫David 电话：Tel: 18965123511 邮箱：Email: david@timebank.space

从 Claude Code 到千行百业
把“用 AI 造工具”长成你公司的护城河 From Claude Code to every industry
grow "building tools with AI" into your moat

一颗叫 palantír 的水晶球，被搬进了现实A crystal ball called palantír — made real

算力部署于贵公司本地，投入成本清晰透明Compute deployed in-house — a clear, transparent cost structure

三个层次：从日常的活，到“造工具”的能力Three layers: from daily work to the power to build tools

各部门日常技能库Departmental skill library

AI 驻场工程师 FDEFDE on-site engineer

用 AI 造工具的元能力The power to build tools

引擎、定价与解决方案，完整公开透明The engine, pricing and solutions — fully transparent

Claude Code 能力与性能 →Claude Code capabilities →

报价总览 · 工位与场景 →Pricing overview →

解决方案 · A·B·C →Solutions →

常见问题FAQ

结合贵公司的行业场景，将 AI 系统化嵌入每一层Embed AI, systematically, into every layer of your operations

大卫 — AI 解决方案架构师David — AI Solution Architect

按部署规模匹配机型Match a build to your scale