Reflex Launches AI App Builder to Bypass Enterprise Frontend Stacks

Reflex: Vibe Code Production-grade Apps in Python With Reflex — Ai4

Ai4

Compiling natural language prompts directly into pure Python allows backend engineers to deliver full-stack enterprise applications without facing low-code scaling limits or dedicated frontend overhead.

AI-assisted summary of Reflex: Vibe Code Production-grade Apps in Python With Reflex — original source. Check the original for full context.

Source published:

Detailed notes

## Background Published by Ai4 on YouTube on August 25, 2025, this presentation features Reflex co-founder and CEO Nikhil Rao demonstrating Reflex's AI app builder. According to the description, Reflex enables Python developers to build and deploy production-grade, full-stack applications through natural language prompts and pure Python without requiring frontend experience or proprietary lock-in.

## TL;DR Nikhil Rao highlights the dilemma enterprise data and backend engineers face between lightweight prototyping tools that hit a 'graduation ceiling' and the complex overhead of React and JavaScript. He introduces Reflex's unified, open-source framework and outlines four essential pillars for enterprise-grade vibe coding: native data connectivity, transparent scalability without black boxes, uncapped customization, and out-of-the-box enterprise security. Through a live demonstration of building a global supplier management platform connected to production databases and custom Python logic, Rao shows how enterprise teams at organizations like Dell and Fastly dramatically cut development cycles without hiring dedicated frontend engineers.

## Atmosphere & style The presentation carries a crisp, pragmatic, and developer-centric tone, balancing high-level architectural analysis with hands-on technical demonstration. Rao communicates with calm, measured authority, stripping away generative AI hype to focus directly on real enterprise engineering bottlenecks. The pacing feels efficient and direct, creating an empowering atmosphere for backend and Python practitioners eager to ship full-stack software autonomously.

## Key takeaways - [00:03](https://www.youtube.com/watch?v=94pLMVTCvwA&t=3s) Data and Python engineers face a major bottleneck by repeatedly being pulled away from core logic to build and maintain user interfaces. - [00:03](https://www.youtube.com/watch?v=94pLMVTCvwA&t=3s) Traditional lightweight tools like Streamlit, Dash, and Tableau suffer from 'graduation risk' when internal apps require custom UI and production scalability. - [00:03](https://www.youtube.com/watch?v=94pLMVTCvwA&t=3s) Transitioning to traditional web stacks like React creates heavy overhead, forcing backend teams to learn frontend languages or rely on cross-functional teams. - [02:25](https://www.youtube.com/watch?v=94pLMVTCvwA&t=145s) Reflex enables a unified development lifecycle where engineers build full-stack apps from natural language prompts entirely in Python. - [02:25](https://www.youtube.com/watch?v=94pLMVTCvwA&t=145s) The open-source Reflex framework has gained significant community and enterprise traction, with over 25,000 GitHub stars and 300,000 built apps. - [03:32](https://www.youtube.com/watch?v=94pLMVTCvwA&t=212s) Enterprise vibe coding requires native data/API integration, non-black-box scalability, uncapped extensibility, and built-in security like SSO and RBAC. - [04:48](https://www.youtube.com/watch?v=94pLMVTCvwA&t=288s) Reflex goes beyond static dashboards by supporting full-stack CRUD capabilities, write-back forms, and embedded Python IDE logic editing. - [04:48](https://www.youtube.com/watch?v=94pLMVTCvwA&t=288s) All generated applications compile directly to standard Python code with no vendor lock-in, enabling single-click deployment to AWS, GCP, Azure, or Databricks. - [07:51](https://www.youtube.com/watch?v=94pLMVTCvwA&t=471s) Enterprises like Dell and Fastly use Reflex to compress internal tool development cycles from months to days without hiring dedicated frontend engineers.

## Detailed walkthrough ### [00:03](https://www.youtube.com/watch?v=94pLMVTCvwA&t=3s) The Internal App Dilemma and the Graduation Risk

Reflex co-founder and CEO Nikhil Rao introduces the platform's dual mission: providing an open-source framework for building full-stack web applications entirely in Python, paired with an AI-native app builder that takes projects from natural language prompts to production without proprietary black boxes. Rao highlights a fundamental bottleneck across enterprise engineering teams: data and Python engineers want to focus their energy on core business logic, agent workflows, and data infrastructure, but they are constantly pulled into building and maintaining user interfaces.

Drawing from his prior experience at Apple, Rao outlines the two flawed paths teams traditionally take when building internal applications. On one side are low-code and data visualization frameworks like Power BI, Tableau, Dash, and Streamlit. While these tools excel at spinning up quick proof-of-concept prototypes, dashboards, or basic forms, they suffer from severe "graduation risk." As soon as requirements expand to demand bespoke UI components, production-grade performance, or customer-facing reliability, these lightweight tools hit a ceiling. When teams inevitably graduate past these tools, they are forced into traditional web stacks like React and JavaScript. This transition creates massive overhead—either forcing backend Python engineers to learn an entirely separate frontend ecosystem or requiring cross-functional full-stack engineering teams, which introduces heavy coordination delays and communication friction.

``` Traditional Paths Reflex Unified Approach ───────────────── ─────────────────────── [Low-Code / Dashboards] [Natural Language & AI] (Streamlit, Dash, Tableau) │ │ (Graduation Risk) ▼ ▼ [Pure Python Framework] [Traditional Full-Stack] │ (React, JS, CSS Overhead) ▼ [One-Click Cloud Deploy] ```

---

[02:25](https://www.youtube.com/watch?v=94pLMVTCvwA&t=145s) Unified Python Development and Open-Source Traction

To resolve this dilemma, Reflex offers a unified development lifecycle where engineers build, integrate, and ship apps using only natural language and Python. The workflow is streamlined into three distinct steps: generating the application interface using AI and prompts, integrating directly with existing Python packages, APIs, and data sources, and deploying to any major cloud infrastructure with a single click—eliminating the need for dedicated frontend or infrastructure teams.

Reflex's developer-first approach to "vibe coding" has driven significant adoption across the software engineering community and enterprise organizations: - Over 25,000 GitHub stars on the open-source Reflex framework. - More than 300,000 applications built across the ecosystem since launch. - Enterprise adoption across Fortune 500 organizations and global institutions, including Autodesk, UNICEF, and Dell. - An average savings of over 60 hours per week for teams building internal tools.

---

[03:32](https://www.youtube.com/watch?v=94pLMVTCvwA&t=212s) The Four Pillars of Enterprise-Grade Vibe Coding

While generative AI models like ChatGPT can easily output basic UI snippets, Rao stresses that enterprise vibe coding requires a significantly more robust, transparent foundation to support mission-critical software. He outlines four strict architectural requirements that platforms must fulfill to succeed in production enterprise environments:

1. **Native Data and API Integration**: Seamless connectivity to production data layers, including PostgreSQL, MongoDB, enterprise SQL warehouses, and internal REST APIs. 2. **Production Scalability Without Black Boxes**: The ability to scale smoothly from a weekend prototype to an enterprise system, where developers can inspect and understand every query, state change, and underlying mechanism. 3. **Uncapped Extensibility and Customization**: Eliminating the risk of hitting a framework ceiling, ensuring any custom component, visual layout, or backend logic can be implemented directly. 4. **Out-of-the-Box Enterprise Security**: Built-in support for mandatory enterprise standards, including Single Sign-On (SSO) and role-based access control (RBAC).

---

[04:48](https://www.youtube.com/watch?v=94pLMVTCvwA&t=288s) Demo: Building a Global Supplier Management Platform

Rao demonstrates Reflex’s AI app builder through an enterprise customer case study: a global supplier operations dashboard. In raw form, the supplier alert data sat inside SQL data warehouses like Databricks and PostgreSQL, making it inaccessible to non-technical business operators. Using natural language inside the Reflex builder, the team generated a live interactive interface featuring a global supplier map, dynamic data tables, and multi-parameter filtering components wired straight to production databases.

``` ┌─────────────────────────────────────────────────────────────┐ │ Reflex App Builder │ │ ┌──────────────────────┐ ┌──────────────────────┐ │ │ │ Prompt: Map, Table, │ ──────> │ Live Interactive App │ │ │ │ Alert Write-Back Form│ │ (Map, Charts, Form) │ │ │ └──────────────────────┘ └──────────┬───────────┘ │ │ │ │ │ │ ▼ ▼ │ │ ┌──────────────────────┐ ┌──────────────────────┐ │ │ │ In-Browser Python IDE│ ──────> │ Custom Business Logic│ │ │ │ (Slack API, Scripts) │ │ (Alert Notification) │ │ │ └──────────────────────┘ └──────────────────────┘ │ └──────────────────────────────┬──────────────────────────────┘ │ Compiles to Open Source ▼ ┌─────────────────────────────────────────────────────────────┐ │ Enterprise Cloud Deployment: AWS / GCP / Azure / Databricks │ └─────────────────────────────────────────────────────────────┘ ```

The demo highlights how Reflex supports full-stack CRUD capabilities rather than static dashboards. With a follow-up prompt, Rao adds a complete data-entry form enabling suppliers to submit incident reports directly back into the warehouse. When custom business logic is required—such as triggering a Slack alert when a high-severity incident is logged—engineers can switch directly into the builder’s embedded Python IDE to call their existing APIs and custom scripts. Because both the frontend components and backend state compile directly into standard Python, engineers retain full visibility into all database queries and execution logic without lock-in, enabling single-click deployment to AWS, GCP, Azure, or Databricks.

---

[07:51](https://www.youtube.com/watch?v=94pLMVTCvwA&t=471s) Enterprise Case Studies and Rapid Delivery

Reflex is actively compressing enterprise development timelines from multi-month cycles with external contractor teams down to days or weeks handled by existing staff. Tech leaders and enterprise teams at Fastly, Amazon, and Dell are utilizing Reflex to build their internal business intelligence stacks and mission-critical tooling in pure Python.

At Dell, a principal backend data engineer was able to build an entire enterprise-wide server monitoring platform from scratch using a small internal support team, without needing to allocate frontend developers or contract outside resources. Rao wraps up by inviting attendees to test Reflex, scan the QR code to schedule a custom demo, and connect with the on-site engineering team.

## Notable quotes > Reflex is the only AI app builder built for Python developers. — Nikhil Rao

> Reflex gives you the speed of vibe coding with the control of traditional development, so you can build real apps, not just demos, no frontend experience required. — Nikhil Rao

> Go from prototype to production without switching tools, and deploy anywhere with a single command. — Nikhil Rao

> [00:03](https://www.youtube.com/watch?v=94pLMVTCvwA&t=3s) data and Python engineers want to focus their energy on core business logic, agent workflows, and data infrastructure, but they are constantly pulled into building and maintaining user interfaces. — Nikhil Rao

> [00:03](https://www.youtube.com/watch?v=94pLMVTCvwA&t=3s) As soon as requirements expand to demand bespoke UI components, production-grade performance, or customer-facing reliability, these lightweight tools hit a ceiling. — Nikhil Rao

> [03:32](https://www.youtube.com/watch?v=94pLMVTCvwA&t=212s) enterprise vibe coding requires a significantly more robust, transparent foundation to support mission-critical software. — Nikhil Rao

> [03:32](https://www.youtube.com/watch?v=94pLMVTCvwA&t=212s) The ability to scale smoothly from a weekend prototype to an enterprise system, where developers can inspect and understand every query, state change, and underlying mechanism. — Nikhil Rao

> [04:48](https://www.youtube.com/watch?v=94pLMVTCvwA&t=288s) Because both the frontend components and backend state compile directly into standard Python, engineers retain full visibility into all database queries and execution logic without lock-in — Nikhil Rao

> [07:51](https://www.youtube.com/watch?v=94pLMVTCvwA&t=471s) compressing enterprise development timelines from multi-month cycles with external contractor teams down to days or weeks handled by existing staff. — Nikhil Rao

## Mentioned Nikhil Rao, Reflex, Ai4, Apple, Power BI, Tableau, Dash, Streamlit, React, JavaScript, Python, ChatGPT, PostgreSQL, MongoDB, Databricks, Slack, AWS, GCP, Azure, Autodesk, UNICEF, Dell, Fastly, Amazon

## Video description Nikhil Rao, CEO, Reflex

Abstracts

Reflex is the only AI app builder built for Python developers. Build full-stack apps with prompting, integrate securely with your company’s data, and use any Python library or existing code, all in one unified framework. Go from prototype to production without switching tools, and deploy anywhere with a single command. Every app is powered by Reflex’s open source framework, already used by thousands of teams and enterprises to build internal tools, dashboards, and AI-powered apps, fast. Reflex gives you the speed of vibe coding with the control of traditional development, so you can build real apps, not just demos, no frontend experience required. Come see how Reflex is changing the way Python developers build.

[Source](https://www.youtube.com/watch?v=94pLMVTCvwA)

中文简报

Reflex推出AI应用构建器，绕开企业前端技术栈 将自然语言提示直接编译为纯Python，使后端工程师能够交付全栈企业应用，无需面对低代码扩展限制或专职前端开发开销。 Nikhil Rao指出了企业数据与后端工程师面临的困境：轻量级原型工具容易触及“毕业天花板”，而采用React和JavaScript又会带来繁琐复杂的开销。他介绍了Reflex的统一开源框架，并概述了企业级氛围编码（vibe coding）的四大核心支柱：原生数据连接、无黑盒的透明可扩展性、无上限的定制能力以及开箱即用的企业级安全保障。通过现场演示构建一个连接生产数据库和自定义Python逻辑的全球供应商管理平台，Rao展示了Dell和Fastly等组织的企业团队如何在无需聘请专职前端工程师的情况下大幅缩短开发周期。 Reflex 联合创始人兼 CEO Nikhil Rao 指出，数据与 Python 工程师常因构建和维护用户界面而陷入瓶颈；传统轻量工具（如 Streamlit、Dash）面临扩展受限的“毕业风险”，而转向 React 等传统全栈技术又会带来巨大的前端学习与跨团队协作成本。 (0:03) Reflex 提供全栈 Python 开发闭环，支持通过自然语言提示词生成交互应用并直连数据源；开源框架已斩获超过 25,000 颗 GitHub Star，生态内构建了逾 30 万个应用，服务于 Autodesk、Dell 等企业，平均每周为内部工具开发团队节省超 60 小时。 (2:25) 企业级 Vibe Coding 需满足原生数据/API 集成、透明非黑盒的架构扩展、无上限定制以及内置 SSO 与 RBAC 等安全能力，避免停留在简单 UI 原型层面。 (3:32) 现场演示了全球供应商管理平台的搭建过程，展示 Reflex 不仅支持静态看板，还涵盖完整 CRUD 回写表单，并在内置 Python IDE 中编写触发 Slack 告警等业务逻辑，全量应用可无锁编译并一键部署至 AWS、GCP、Azure 或 Databricks。 (4:48) Dell、Fastly 及 Amazon 等企业正借助 Reflex 将开发周期从数月压缩至数天或数周；Dell 工程师在未借调前端资源的情况下，由后端数据团队全自主构建了全企业级服务器监控平台。 (7:51)

Reading summary

Reflex co-founder and CEO Nikhil Rao argues that enterprise engineering teams face an inefficient trade-off between brittle low-code dashboards and complex JavaScript stacks for internal tools. He proposes an open-source, AI-native framework allowing data and Python engineers to generate, customize, and deploy production-grade web applications entirely in Python. The Graduation Risk in Internal App Development (0:03) Enterprise data and Python engineers are frequently diverted from core business logic, agent workflows, and data infrastructure to build and maintain user interfaces. Drawing on his experience at Apple, Rao outlines two common but flawed historical paths for internal tools. Lightweight dashboard and low-code frameworks like Streamlit, Dash, Tableau, and Power BI allow rapid prototyping but suffer from severe 'graduation risk,' hitting a ceiling whenever projects require custom user experiences or high performance. Graduating past lightweight tools traditionally forces teams into standard web stacks centered on React, JavaScript, and CSS. Rao notes this shift introduces heavy friction: it either compels backend Python engineers to learn an unfamiliar frontend ecosystem or requires cross-functional coordination with dedicated frontend developers, producing communication bottlenecks and delivery delays. Four Architectural Requirements for Enterprise Applications (3:32) While consumer generative AI tools can readily generate basic interface snippets, Rao emphasizes that enterprise-grade application building requires far more architectural rigor. He outlines four foundational pillars that any enterprise-ready platform must provide: First, applications must support native integration with production databases like PostgreSQL, MongoDB, and enterprise SQL warehouses. Second, systems must offer scalable performance without proprietary black boxes, ensuring developers can inspect underlying state changes and queries. Third, tools must provide uncapped extensibility so teams never encounter framework ceilings when designing complex UI layouts or backend scripts. Fourth, platforms require built-in enterprise governance, including Single Sign-On (SSO) and role-based access control (RBAC). Full-Stack Interactivity and Transparent Workflows (4:48) Reflex addresses application development through a three-stage workflow: generating user interfaces via natural language prompts, wiring components directly into existing Python codebases and databases, and deploying to cloud infrastructure with a single click. In a global supplier operations demonstration, raw incident data in PostgreSQL and Databricks was surfaced to non-technical users through an AI-generated interface featuring interactive maps, dynamic tables, and write-back CRUD forms. When applications require bespoke operational behavior, developers can edit the underlying source code in an in-browser Python environment rather than relying solely on automated generation. For example, engineers can wire incident submissions directly into Slack alerting routines. Because interface components and backend state compile to standard Python, teams maintain visibility into data flow and can deploy without vendor lock-in to AWS, Azure, GCP, or Databricks. Enterprise Adoption and Resource Efficiency (2:25) Reflex reports substantial ecosystem traction, citing over 25,000 GitHub stars and more than 300,000 created applications. Rao claims the framework saves engineering teams an average of over 60 hours per week and enables organizations to compress project delivery from multi-month contracting cycles into days or weeks executed by internal staff. The platform has been adopted across organizations such as Autodesk, UNICEF, Amazon, Fastly, and Dell. In one cited implementation at Dell, a principal backend data engineer reportedly delivered an enterprise-wide server monitoring system with a minimal internal team, bypassing the need to recruit frontend specialists or hire outside contractors. Reflex positions pure Python and transparent code generation as the solution to internal tool backlogs, though its enterprise success relies on whether Python-centric abstractions can seamlessly satisfy complex, evolving frontend demands without eventually requiring traditional web development specialists.

精炼阅读摘要

Reflex联合创始人兼首席执行官Nikhil Rao指出，企业数据与Python工程师常受制于前端界面的构建和维护需求；为此，Reflex推出了纯Python全栈开源框架与AI原生构建器，旨在消除传统低代码工具的扩展瓶颈及跨语言全栈开发的协作摩擦。 内部应用开发困境与低代码的“毕业风险” (0:03) 结合此前在Apple的工作经历，Rao分析了企业团队构建内部应用时面临的核心瓶颈。数据工程师与Python开发者的本职工作应当聚焦于业务逻辑、Agent工作流及数据基础设施，但现实中却频繁被拉去构建和维护用户界面。传统的解决路径主要呈现两极分化：一类是依赖Tableau、Power BI、Dash以及Streamlit等低代码或数据可视化工具；另一类则是转向传统的全栈Web技术栈。 Rao指出，低代码和可视化工具虽能快速产出PoC原型、数据看板或简单表单，但普遍存在严重的“毕业风险”（Graduation Risk）。一旦业务需求扩展至定制化UI组件、生产级性能或面向客户的可靠性要求，这些轻量级框架就会触及能力上限。当团队被迫“毕业”并重构系统时，往往必须切换到React、JavaScript和CSS等传统Web栈。这种技术断层不仅迫使Python工程师去学习全新的前端生态，还会引入跨职能全栈团队的协调延迟与沟通摩擦。 企业级Vibe Coding的四大架构支柱 (3:32) 针对通用生成式AI仅能生成基础UI代码片段的局限，Rao提出企业级“氛围编码”（Vibe Coding）必须建立在透明且稳固的基础架构之上，并明确了四大核心支柱：其一，原生数据与API集成，支持与PostgreSQL、MongoDB、企业级SQL数据仓库及内部REST API无缝连接；其二，无黑盒的生产级可扩展性，系统可从周末原型平滑过渡至企业系统，开发者能够完全审查和理解底层查询、状态变化及运作机制；其三，无上限的扩展与定制能力，确保任意定制化组件、视觉布局或后端逻辑都能直接实现；其四，开箱即用的企业安全标准，内建单点登录（SSO）与基于角色的访问控制（RBAC）。 为了落实这一方案，Reflex提供了一个统一的生命周期，开发者只需使用自然语言与Python即可完成开发：利用AI和提示词生成界面、直接接入现有Python库与数据源，并实现向主流云平台的一键部署。据Rao介绍，该开源框架在GitHub上已获得超过25,000颗星，生态系统内累计构建了超过300,000个应用程序，并被Autodesk、UNICEF和Dell等财富500强企业与全球机构采用，为内部工具构建团队平均每周节省超过60工时。 全球供应商平台演示与生产级CRUD实现 (4:48) Rao通过一个全球供应商运营仪表盘的客户案例，演示了Reflex的AI构建流程。在该场景中，原始供应商警报数据原本存放在Databricks和PostgreSQL等SQL数据仓库内，非技术业务人员无法直接访问。通过在Reflex中使用自然语言提示词，团队快速生成了一个具备交互能力的界面，包含直接接入生产数据库的全球供应商地图、动态数据表格以及多参数筛选组件。 该演示重点展示了Reflex支持全栈CRUD（增删改查）而非局限于静态展示的能力。通过追加提示词，系统自动生成了数据录入表单，允许供应商将事件报告直接写回数据仓库。当需要添加复杂的自定义业务逻辑时（例如在记录高严重度事件时触发Slack警报通知），工程师可直接切换至内置的浏览器Python IDE，调用现有的Python脚本与API。由于前端组件与后端状态均直接编译为标准Python代码，开发者免受专有黑盒限制，不仅能掌控全部底层查询，还能一键部署至AWS、GCP、Azure或Databricks。 压缩交付周期与企业级落地实践 (7:51) 在交付效能方面，Rao指出Reflex正在将原本依赖外部承包团队、耗时数月的企业开发周期，大幅压缩至由现有内部员工主导的数天或数周时间。包括Fastly、Amazon以及Dell在内的科技公司与企业团队，正在使用纯Python来构建其内部商业智能技术栈和核心业务工具。 以Dell为例，该公司的首席后端数据工程师带领一个小型内部支持团队，从零构建了一套覆盖整个企业范围的服务器监控平台。在整个开发流程中，该团队无需调配专门的前端开发人员，也未雇佣任何外部承包资源，证明了数据工程团队凭借Python全栈能力独立交付生产级系统的可行性。 Rao展示了Reflex如何通过将自然语言提示词、纯Python开发以及透明代码架构相结合，使数据和后端工程师摆脱对前端团队的依赖；不过，该方案在面对更为极端的超大规模复杂并发系统或深度定制前端交互场景时，其长期维护成本与性能表现仍有待更多生产环境的持续检验。

Detailed notes — zh

## Background 本内容由 YouTube 频道 Ai4 于 2025 年 8 月 25 日发布，记录了 Reflex 联合创始人兼 CEO Nikhil Rao 的主题演讲。页面描述指出，Reflex 是专为 Python 开发者打造的 AI 应用构建器，支持通过自然语言提示词构建全栈应用并与企业数据安全集成，帮助开发者摆脱低代码工具的限制并实现一键部署。 ## TL;DR Nikhil Rao 深入分析了 Python 与数据工程师在开发内部应用时面临的“毕业风险”与全栈协作成本痛点，进而介绍了 Reflex 的开源框架与 AI App Builder。他阐述了企业级 Vibe Coding 的四大支柱：无缝数据集成、渐进式生产级扩展、无黑盒代码掌控以及开箱即用的安全体系。通过全球供应商仪表盘的实机演示，演讲展示了从自然语言生成 UI、双向 CRUD 回写到原生 Python IDE 编写 Slack 联动的全流程，并分享了 Dell、Amazon 等企业将数月开发周期缩短至数天的落地实践。 ## Atmosphere & style 整场演讲呈现出典型企业级技术发布会的务实与快节奏风格，主讲人表达干练、条理分明。通过将低代码工具的局限性与全栈开发的痛点进行鲜明对比，演讲在抛出痛点后迅速切入流畅的实机演示，呈现出从“简单 Prompt”到“复杂生产代码”无缝衔接的技术自信。现场氛围兼具前沿 AI 生产力工具的兴奋感与面向大企业基础设施的沉稳落地感。 ## Key takeaways - [00:03](https://www.youtube.com/watch?v=94pLMVTCvwA&t=3s) 数据与 Python 工程师在内部工具开发中常被迫耗费精力于前端 UI 构建，偏离了数据基建、AI 及业务逻辑等核心价值。 - [00:30](https://www.youtube.com/watch?v=94pLMVTCvwA&t=30s) 轻量级 Python 框架与低代码 BI 工具在 PoC 阶段便捷，但面对复杂交互和高并发时存在严重的“毕业风险”，无法平滑过渡到生产环境。 - [01:30](https://www.youtube.com/watch?v=94pLMVTCvwA&t=90s) 转向传统 React/JS 前端栈不仅学习成本高昂，且引入跨团队全栈协作会产生沉重的沟通开销与交付延误。 - [02:25](https://www.youtube.com/watch?v=94pLMVTCvwA&t=145s) Reflex 提供统一开发平台，允许开发者仅通过自然语言提示和 Python 完成从原型设计到生产级部署的全流程。 - [02:50](https://www.youtube.com/watch?v=94pLMVTCvwA&t=170s) Reflex 开源生态已积累超 2.5 万 GitHub Star 并构建超 30 万款应用，平均每周能为团队节省 60 多个工时。 - [03:20](https://www.youtube.com/watch?v=94pLMVTCvwA&t=200s) 企业级 Vibe Coding 必须满足四大标准：原生多源数据集成、透明非黑盒的生产扩展能力、全面的定制自由度，以及 SSO/RBAC 等企业级安全。 - [04:48](https://www.youtube.com/watch?v=94pLMVTCvwA&t=288s) Reflex AI App Builder 能通过提示词直接生成绑定 Databricks 或 PostgreSQL 实时数据的交互式地图、图表与仪表盘。 - [05:45](https://www.youtube.com/watch?v=94pLMVTCvwA&t=345s) 系统支持从单纯的数据可视化快速扩展为具备表单回写（Write-back）能力的完整全栈 CRUD 应用。 - [06:30](https://www.youtube.com/watch?v=94pLMVTCvwA&t=390s) 内置 IDE 支持开发者直接调用原生 Python 脚本和现有 API（如 Slack 警报），实现无厂商锁定的灵活扩展。 - [07:15](https://www.youtube.com/watch?v=94pLMVTCvwA&t=435s) 应用最终被统一编译为透明的 Python 代码，并支持一键部署至 AWS、GCP、Azure 和 Databricks 等云环境。 - [07:51](https://www.youtube.com/watch?v=94pLMVTCvwA&t=471s) Dell、Fastly 及 Amazon 等企业借此将数月的开发周期缩短至数天，Dell 工程师在无前端协助下独立交付了全公司级监控系统。 ## Detailed walkthrough ### [00:03](https://www.youtube.com/watch?v=94pLMVTCvwA&t=3s) 企业内部应用开发的困境与“毕业风险” Reflex 联合创始人兼 CEO Nikhil Rao 登台介绍了 Reflex 的两大核心构成：一个支持用纯 Python 开发 Web 应用的开源框架，以及一个通过自然语言和 Python 贯通原型到生产环境的 AI 应用构建器（AI App Builder）。Nikhil 指出，现代企业在开发内部应用程序时耗费了过多的人力、时间和资源。Python 工程师与数据工程师最核心的价值应当体现在数据基础设施、AI、智能体（Agents）以及业务逻辑的打磨上，但由于所有系统最终都必须交付可用的用户界面（UI），许多工程师被迫将大量精力耗费在前端构建等边角工作上，脱离了真正能为企业创造价值的核心业务。 谈及过往在 Apple 的工作经历，Nikhil 剖析了当前开发者面临的两难路径。第一条路径是使用 Power BI、Tableau 等低代码商业智能工具，或是 Dash、Streamlit 等轻量级 Python 框架。这类工具在快速搭建 PoC（概念验证）原型、简单仪表盘或基础表单时非常便捷，但普遍存在严重的“毕业风险”（graduation risk）——一旦应用复杂度上升，需要更复杂的用户交互、生产级性能表现，或者需要作为正式产品交付给最终客户与企业内部使用时，低代码工具便无法承载。此时团队不得不转向第二条路径，即采用 React、JavaScript 等传统前端 Web 栈。然而这对数据和 Python 工程师而言意味着巨大的时间与金钱成本：要么工程师必须耗费精力去学习与自身技能完全脱节的技术，要么企业必须引入专职的全栈开发团队，而跨团队沟通需求和协同流转又会带来沉重的沟通开销与周期延误。 - **核心矛盾**：数据/Python 工程师希望专注于数据基建、AI、Agent 与业务逻辑，却被 UI 开发分散精力。 - **低代码/轻量框架局限**：PoC 阶段快捷，但面临无法支撑复杂交互与生产性能的“毕业风险”。 - **传统 Web 开发痛点**：React/JS 学习曲线陡峭，引入全栈团队会产生高昂的需求沟通与时间成本。 ### [02:25](https://www.youtube.com/watch?v=94pLMVTCvwA&t=145s) Reflex 统一平台与企业级 Vibe Coding 的四大支柱 为了彻底打破两难困境，Reflex 构建了一个统一平台，让开发者仅凭自然语言提示和 Python 就能实现从原型到生产的全流程开发，不再依赖专职前端或运维（Infra）工程师。整个工作流精简为三步：第一，借助 AI 与自然语言生成应用；第二，无缝集成现有的 Python 库、API 与第三方依赖包；第三，就绪后一键部署到任意云端环境。Reflex 开源生态发展迅速，在 GitHub 上已收获超 25,000 颗 Star，平台累计构建应用超过 300,000 个，被 Autodesk、UNICEF（联合国儿童基金会）以及 Dell 等财富 500 强企业广泛采用，平均每周能为内部应用开发团队节省超过 60 个工时。 面对当下关于“Vibe Coding”（氛围编码/提示词驱动编码）的热潮，Nikhil 强调普通聊天机器人只能生成玩具级的 Demo，而真正的“企业级 Vibe Coding”必须满足四大硬性标准： 1. **数据源无缝集成**：必须能原生连接企业内部数据库（如 PostgreSQL、MongoDB 等）及各类现有 API； 2. **平滑迈向生产环境**：框架必须能随业务复杂度增长而扩展，且拒绝“黑盒”（no black boxes），开发者必须能完全掌控底层细节； 3. **全面的定制自由度**：打破低代码工具的能力上限，提供从原型到企业生产级应用的完整定制能力； 4. **开箱即用的企业级安全**：原生支持单点登录（SSO）、基于角色的访问控制（RBAC）等企业应用准入门槛。 ```mermaid flowchart LR accTitle: Reflex 开发与部署流程架构 subgraph Input ["输入与构建"] A["自然语言提示 / AI"] --> C["Reflex 统一引擎"] B["Python 代码 / API / 库"] --> C end subgraph Core ["核心能力"] C --> D["双向数据集成\n(SQL / Mongo / APIs)"] C --> E["企业级安全\n(SSO / RBAC)"] end subgraph Output ["全栈编译与部署"] D & E --> F["纯 Python 编译\n(全栈透明非黑盒)"] F --> G["一键多云部署\n(AWS / GCP / Azure / Databricks)"] end ``` ### [04:48](https://www.youtube.com/watch?v=94pLMVTCvwA&t=288s) 现场演示：从自然语言到生产级 CRUD 与 Slack 联动 Nikhil 通过一段大型企业客户的真实案例视频展示了 Reflex 的实机表现。该企业拥有遍布全球的供应商网络，需要一个“供应商仪表盘”（Suppliers Dashboard），一方面让供应商在工厂发生异常时上报警报，另一方面让管理层拥有全局监控视图。最初，这些数据沉淀在 Databricks、PostgreSQL 等 SQL 仓库中，除了数据工程师外其他人很难直接使用。在 Reflex 的 AI App Builder 界面中，用户只需连接数据源并输入提示词：“展示供应商地图、数据表格以及过滤功能”，AI 随即生成了集成动态地图、数据表、图表与交互组件的完整可视化应用，并直接绑定企业实时数据。 ```mermaid flowchart TD accTitle: 供应商仪表盘应用生成与扩展流程 subgraph Stage1 ["第一阶段：生成可视化界面"] P1["Prompt: 生成地图、表格与过滤器"] --> APP1["绑定 SQL 数据源\n渲染交互地图与数据看板"] end subgraph Stage2 ["第二阶段：扩展 CRUD 回写"] P2["Prompt: 增加供应商数据提交流单"] --> APP2["生成输入表单\n支持直接回写数据库"] end subgraph Stage3 ["第三阶段：IDE 原生代码集成"] CODE["写入现有 Python 脚本"] --> APP3["触发高等级警报时\n自动通过 API 推送 Slack"] end Stage1 --> Stage2 --> Stage3 ``` 随后，Nikhil 演示了如何将只读仪表盘进一步扩展为具备增删改查能力的完整全栈 CRUD 应用。通过追问提示词“增加一个供供应商回写数据的表单”，AI 迅速生成了数据录入表单，提交后数据直接回写至数据库。更关键的是，Reflex 内部集成了完整的 IDE，支持开发者编写任意原生 Python 代码。该客户希望在系统接收到高严重级别警报时自动推送到内部 Slack，开发者直接在 IDE 中调用企业预先存在的 Python API 即可轻松实现联动。Nikhil 特别强调，Reflex 的前端与后端最终全部编译为纯 Python，绝非不可控的黑盒代码，所有的查询、API 调用与交互逻辑一目了然，且不存在任何厂商锁定（No Vendor Lock-in），支持一键无缝部署至 AWS、GCP、Azure 或 Databricks。 - **动态组件生成**：通过自然语言直接生成交互式地图组件、数据分析图表、筛选器与数据表格。 - **双向数据回写**：不仅支持前端呈现，还支持生成表单直接执行数据库 Write-back 操作。 - **原生 IDE 与代码掌控**：在 AI 生成的基础上支持自由注入 Python 代码与内部 API（如 Slack 警报推送）。 - **云原生多目标部署**：一键将应用分发至 AWS、GCP、Azure 及 Databricks 等云环境。 ### [07:51](https://www.youtube.com/watch?v=94pLMVTCvwA&t=471s) 客户案例与落地成效 在演讲尾声，Nikhil 分享了 Reflex 在企业级客户中的落地成果。包括 Fastly、Amazon 在内的大型企业正使用 Reflex 构建其完整的 BI 分析体系与内部应用，将原本需要协调外部全栈团队、耗时数月的开发周期，大幅压缩到了数天或数周之内。 Nikhil 重点列举了来自 Dell 的实际案例：Dell 的一位后端首席数据工程师（Principal Engineer）在完全没有调动前端或外部全栈资源的情况下，仅依靠自身的小型支持团队，便使用 Reflex 独立搭建出了一款供全公司范围使用的服务器监控应用程序（Server Monitoring App）。最后，Nikhil 展示了现场预约产品演示的二维码，并邀请对 Reflex 感兴趣的参会者在会后与现场的 Reflex 团队成员展开深入交流。 - **开发效率飞跃**：将传统外包或跨团队协作耗时数月的项目缩短至数天或数周即可上线交付。 - **Dell 落地标杆**：后端首席数据工程师无需前端工程师协助，独立完成全公司级服务器监控应用。 - **覆盖代表企业**：Fastly、Amazon、Dell、Autodesk、UNICEF 等。 ## Notable quotes > [00:03](https://www.youtube.com/watch?v=94pLMVTCvwA&t=3s) Python 工程师与数据工程师最核心的价值应当体现在数据基础设施、AI、智能体（Agents）以及业务逻辑的打磨上，但由于所有系统最终都必须交付可用的用户界面（UI），许多工程师被迫将大量精力耗费在前端构建等边角工作上，脱离了真正能为企业创造价值的核心业务。 — Nikhil Rao > [00:03](https://www.youtube.com/watch?v=94pLMVTCvwA&t=3s) 低代码工具在快速搭建 PoC 原型时非常便捷，但普遍存在严重的‘毕业风险’（graduation risk）——一旦应用复杂度上升，需要更复杂的用户交互、生产级性能表现，低代码工具便无法承载。 — Nikhil Rao > [02:25](https://www.youtube.com/watch?v=94pLMVTCvwA&t=145s) 开发者仅凭自然语言提示和 Python 就能实现从原型到生产的全流程开发，不再依赖专职前端或运维工程师。 — Nikhil Rao > [02:25](https://www.youtube.com/watch?v=94pLMVTCvwA&t=145s) 真正的‘企业级 Vibe Coding’必须满足四大硬性标准：数据源无缝集成、平滑迈向生产环境且拒绝黑盒、全面的定制自由度，以及开箱即用的企业级安全。 — Nikhil Rao > [04:48](https://www.youtube.com/watch?v=94pLMVTCvwA&t=288s) Reflex 的前端与后端最终全部编译为纯 Python，绝非不可控的黑盒代码，所有的查询、API 调用与交互逻辑一目了然，且不存在任何厂商锁定。 — Nikhil Rao > [04:48](https://www.youtube.com/watch?v=94pLMVTCvwA&t=288s) 在 AI 生成的基础上支持自由注入 Python 代码与内部 API，将只读仪表盘进一步扩展为具备增删改查能力的完整全栈 CRUD 应用。 — Nikhil Rao > [07:51](https://www.youtube.com/watch?v=94pLMVTCvwA&t=471s) 包括 Fastly、Amazon 在内的大型企业正使用 Reflex 构建其完整的 BI 分析体系与内部应用，将原本需要协调外部全栈团队、耗时数月的开发周期，大幅压缩到了数天或数周之内。 — Nikhil Rao > [07:51](https://www.youtube.com/watch?v=94pLMVTCvwA&t=471s) Dell 的一位后端首席数据工程师在完全没有调动前端或外部全栈资源的情况下，仅依靠自身的小型支持团队，便使用 Reflex 独立搭建出了一款供全公司范围使用的服务器监控应用程序。 — Nikhil Rao ## Mentioned Nikhil Rao, Reflex, Apple, Power BI, Tableau, Dash, Streamlit, React, JavaScript, Python, GitHub, Autodesk, UNICEF, Dell, PostgreSQL, MongoDB, Databricks, Slack, AWS, GCP, Azure, Fastly, Amazon ## Video description Nikhil Rao, CEO, Reflex Abstracts Reflex is the only AI app builder built for Python developers. Build full-stack apps with prompting, integrate securely with your company’s data, and use any Python library or existing code, all in one unified framework. Go from prototype to production without switching tools, and deploy anywhere with a single command. Every app is powered by Reflex’s open source framework, already used by thousands of teams and enterprises to build internal tools, dashboards, and AI-powered apps, fast. Reflex gives you the speed of vibe coding with the control of traditional development, so you can build real apps, not just demos, no frontend experience required. Come see how Reflex is changing the way Python developers build. [Source](https://www.youtube.com/watch?v=94pLMVTCvwA)

Source quotes

[0:03] Nikhil Rao: Python 工程师与数据工程师最核心的价值应当体现在数据基础设施、AI、智能体（Agents）以及业务逻辑的打磨上，但由于所有系统最终都必须交付可用的用户界面（UI），许多工程师被迫将大量精力耗费在前端构建等边角工作上，脱离了真正能为企业创造价值的核心业务。 [0:03] Nikhil Rao: 低代码工具在快速搭建 PoC 原型时非常便捷，但普遍存在严重的‘毕业风险’（graduation risk）——一旦应用复杂度上升，需要更复杂的用户交互、生产级性能表现，低代码工具便无法承载。 [2:25] Nikhil Rao: 开发者仅凭自然语言提示和 Python 就能实现从原型到生产的全流程开发，不再依赖专职前端或运维工程师。 [2:25] Nikhil Rao: 真正的‘企业级 Vibe Coding’必须满足四大硬性标准：数据源无缝集成、平滑迈向生产环境且拒绝黑盒、全面的定制自由度，以及开箱即用的企业级安全。 [4:48] Nikhil Rao: Reflex 的前端与后端最终全部编译为纯 Python，绝非不可控的黑盒代码，所有的查询、API 调用与交互逻辑一目了然，且不存在任何厂商锁定。 [4:48] Nikhil Rao: 在 AI 生成的基础上支持自由注入 Python 代码与内部 API，将只读仪表盘进一步扩展为具备增删改查能力的完整全栈 CRUD 应用。 [7:51] Nikhil Rao: 包括 Fastly、Amazon 在内的大型企业正使用 Reflex 构建其完整的 BI 分析体系与内部应用，将原本需要协调外部全栈团队、耗时数月的开发周期，大幅压缩到了数天或数周之内。 [7:51] Nikhil Rao: Dell 的一位后端首席数据工程师在完全没有调动前端或外部全栈资源的情况下，仅依靠自身的小型支持团队，便使用 Reflex 独立搭建出了一款供全公司范围使用的服务器监控应用程序。

Source quotes

Nikhil Rao: Reflex is the only AI app builder built for Python developers. Nikhil Rao: Reflex gives you the speed of vibe coding with the control of traditional development, so you can build real apps, not just demos, no frontend experience required. Nikhil Rao: Go from prototype to production without switching tools, and deploy anywhere with a single command. [0:03] Nikhil Rao: data and Python engineers want to focus their energy on core business logic, agent workflows, and data infrastructure, but they are constantly pulled into building and maintaining user interfaces. [0:03] Nikhil Rao: As soon as requirements expand to demand bespoke UI components, production-grade performance, or customer-facing reliability, these lightweight tools hit a ceiling. [3:32] Nikhil Rao: enterprise vibe coding requires a significantly more robust, transparent foundation to support mission-critical software. [3:32] Nikhil Rao: The ability to scale smoothly from a weekend prototype to an enterprise system, where developers can inspect and understand every query, state change, and underlying mechanism. [4:48] Nikhil Rao: Because both the frontend components and backend state compile directly into standard Python, engineers retain full visibility into all database queries and execution logic without lock-in [7:51] Nikhil Rao: compressing enterprise development timelines from multi-month cycles with external contractor teams down to days or weeks handled by existing staff.

Original transcript

[0:03] Hey everyone, it's it's great to be [0:05] here. Uh so today I'm going to be [0:07] talking about Reflex. Uh I'm co-founder [0:10] CEO at Reflex and at Reflex we're [0:12] building two things. One, an open- [0:14] source library to build web apps in pure [0:16] Python. And two, an AI app builder to go [0:19] from prototype to production all through [0:22] natural language and Python, no black [0:24] boxes, so you can understand fully [0:26] what's happening in your vioded [0:28] applications. [0:30] So the core problem we've seen is [0:33] enterprises spend too much time, energy, [0:35] and resources developing internal [0:38] applications. So your Python and data [0:41] engineers, they want to be spending time [0:42] working on the data infrastructure, the [0:44] AI, the agents, the business logic. But [0:47] at the end of the day, these apps need [0:49] some sort of UI and user interface. And [0:51] what happens is a lot of people end up [0:54] spending time developing the final app [0:56] rather than focusing on the actual [0:58] business logic. that's providing value [0:59] to their company. So right now when [1:02] people want to build these apps, there's [1:04] kind of two approaches that that they [1:06] can take. And I've seen this at my [1:08] previous work at Apple where on one hand [1:11] you have these low code frameworks. So I [1:14] don't know if people here have used [1:16] tools like PowerBI, Tableau or even [1:18] Python tools like Dash and Streamlet. So [1:21] there's this class of tools makes it [1:23] really easy to get a proof of concept [1:25] prototype out if you want a simple [1:27] dashboard or a simple form. They're [1:29] they're really great for that. But what [1:31] we've often seen is they have a huge [1:33] graduation risk. So as your app gets [1:36] more complex either in terms of user [1:38] interfaces, you want production uh [1:41] performance on it. When you want a final [1:43] app that you want to ship to an end [1:44] customer or even within your own [1:46] company, these tools don't really cut [1:48] it. And what happens at that point is uh [1:51] teams need to go to these traditional [1:53] web frameworks. So these are things like [1:56] React, JavaScript, all the classic tools [1:58] used to build normal web apps. But for [2:01] especially your data and Python [2:02] engineers, there's a huge overhead in [2:04] terms of time and cost if they want to [2:07] use these type of tools. On the one [2:09] hand, if they're upskilling, they need [2:11] to learn something completely irrelevant [2:13] to their skill set. And if you want to [2:15] bring in a full stack team to work on [2:17] these, there's a huge cost in terms of [2:19] time and overhead in communicating your [2:21] app requirements to this team. [2:25] So we built Reflex as a unified platform [2:27] to go from prototype to production using [2:30] natural language and Python. So our goal [2:34] is to really simplify the entire process [2:36] of app development from building the app [2:39] to going to production and then all the [2:41] way to deploying your app. So taking [2:43] away the need for having front-end [2:44] engineers or infra engineers to host [2:46] your app, we have three simple steps. [2:49] One, build your app through natural [2:51] language and AI. Two, integrated with [2:54] your existing Python libraries, APIs, [2:57] and third party Python packages. And [2:59] three, when you're ready to host and [3:01] deploy, deploy with a single click to [3:03] any cloud. [3:06] So we since we started Reflex, we've [3:08] seen tremendous traction and interest in [3:11] this kind of vibe coded development. Our [3:13] open source project has over 25,000 [3:16] GitHub stars. And since launch, we've [3:18] had over 300,000 apps built, including [3:20] at large Fortune 500 companies like [3:22] Autodesk, UNICEF, and Dell. We save [3:26] teams that use us for internal apps over [3:28] 60 hours a week on average. [3:32] So there's a lot of talk about vibe [3:34] coding app generation and what I wanted [3:37] to talk about is what it means to be an [3:39] enterprise vibe coding. So even chatbt [3:42] and other websites can you know give you [3:44] a simple user interface but to take it [3:46] all the way to the end there's four key [3:48] requirements that you want in your [3:50] platform. One does it easily integrate [3:52] with all your company data sources. for [3:55] example, your your databases, your [3:57] Postgress, your MongoDB, as well as any [4:00] APIs that you're using. It needs to be [4:02] really easy to integrate with your the [4:03] tools you're using. Number two, can we [4:06] scale beyond a little toy prototype to [4:08] go all the way to production? What this [4:11] means is as your app grows in [4:12] complexity, can the framework grow with [4:14] you with while you still understand [4:16] what's happening under the hood? So, [4:18] it's not a black box. [4:20] Number three, having full customization. [4:22] So you never hit that point where you [4:24] see you find the framework can't support [4:26] you anymore. The framework should take [4:28] you all the way from your prototype to a [4:30] production grade app. And coming to [4:32] point number four, which is really [4:34] important for enterprise apps, is the [4:36] security aspect. So can this integrate [4:38] with my SSO? Can this have role-based [4:40] authorization? And all these [4:42] requirements that are table stakes for [4:43] these enterprise apps, they should come [4:45] out of the box. [4:48] So I'll walk through a video demo to [4:50] show how our platform works in action. [4:52] Um just to give some context, this is a [4:54] large enterprise customer we've been [4:56] working with and their goal was to have [4:58] a suppliers dashboard. They have many [5:00] suppliers all over the world and they [5:02] wanted a nice way for suppliers to uh [5:04] say when alerts are happening in their [5:06] factory as well as for management to [5:08] have a holistic view of everything [5:09] that's happening. So when your data [5:12] starts, you have your data in some sort [5:13] of SQL warehouse whether data bricks or [5:15] postgress etc. But in that format, it's [5:17] not really usable to anyone except the [5:19] data engineers. So what companies want [5:21] to do is build apps around this. So [5:23] anyone technical or non-technical can [5:25] actually see what's happening with their [5:27] data. So this is the interface for our [5:29] AI AI app builder. You can see you can [5:31] connect with your data sources. And in [5:33] this case, they gave it a prompt of show [5:35] me a map of the suppliers as well as a [5:37] table and a way to filter. So you can [5:39] see with a simple prompt you can easily [5:41] create these full applications that have [5:43] complex components like maps, tables, [5:46] charts, chat and this is all hooked up [5:49] to your live company data. So you can [5:51] see when you want to filter different [5:53] factories when you want to see what's [5:55] happening now you actually have this in [5:56] a visual interface rather than having to [5:59] go through SQL or other sources. So this [6:02] is a lot of companies use us as an [6:04] upgraded dashboarding uh tool, but we go [6:06] far beyond that to build full stack CRUD [6:08] applications. So you can see you can [6:10] keep iterating with Reflex by doing [6:12] follow-up prompts. So in the follow-up [6:14] prompt, they said I actually want a a [6:16] form where the suppliers can insert back [6:18] into the database. So with that simple [6:21] prompt, the AI will now build build a [6:24] form where you can go in, fill in the [6:26] data, and once you submit it, it will [6:28] write back to the database. So the goal [6:30] here is to make it make simple uh make [6:33] it very simple to build these extensive [6:35] UIs around your application. [6:38] Um one thing that we haven't seen yet [6:40] right now this was all built purely [6:42] through AI but the key part of reflex is [6:45] integrating with your Python code. So as [6:47] we'll see here part of reflex is within [6:50] our app builder is you have a full IDE [6:52] to put any Python code you want. So in [6:55] this case, the customer wanted to say [6:56] when a high severity alert comes in, it [6:58] should ping to our Slack and they [7:00] already have all these APIs that that [7:02] they had pre-existing to do all this and [7:05] we make it very simple in reflex to [7:06] integrate this with your code. So in in [7:09] addition to working with natural [7:10] language, you can actually come in and [7:12] put any Python code you want and sim [7:15] similar similarly our AI is the only one [7:18] built for Python. The front end and back [7:20] end all compile down to Python, which [7:22] means it's not a black box. you can [7:24] understand exactly what's happening [7:25] under the hood with the queries, the API [7:27] calls, etc. And once you're ready to [7:31] deploy your app, we don't lock you in at [7:33] all. So, you can use our app builder to [7:35] build, but it compiles down to our open [7:37] source framework. And with a simple [7:39] click of a button, you can deploy to any [7:41] cloud that you that you use, AWS, GCP, [7:44] Azure, data bricks. We you can easily [7:47] hook this up and deploy to any of these. [7:51] Um, so we've been accelerating [7:53] development at large companies including [7:57] Fastly, Amazon and large enterprises [8:00] that are building their entire BI stacks [8:02] in Reflex and they're able to work in [8:05] weeks or days, what used to take them [8:07] several months and an external team. So [8:10] just just one case study um from our [8:13] customers at Dell is a principal [8:15] engineer there who is a back-end data [8:17] engineer was able to build an entire [8:19] server monitoring app that's used [8:20] throughout the company without invoking [8:22] external resources and doing it all with [8:25] their small support team. [8:28] So part of some some of our teammates [8:31] are here. This QR code will link if you [8:33] want to pull out your phone, it will [8:35] link to book a demo with us. And we're [8:37] happy to show how Reflex can show can [8:39] accelerate uh workflows within your own [8:42] companies. After the the talk, feel free [8:44] to uh come up to me or some of our [8:46] teammates and we're we're happy to [8:47] discuss more about how Reflex can help [8:50] you and your