danny-avila

Architecture of a Multi-Tenant Private ChatGPT SaaS on LibreChat: Silo Tenants, Seat Billing and Token Caps This post is AI-assisted. ChatGPT Business costs $20 per user per month on annual billing and $25 on monthly billing. Claude Team's standard seat is the same $20 and $25. Microsoft 365 Copilot Business is $25.20 on monthly billing, on top of a Microsoft 365 license. That is a price umbrella, and a focused product can live under it: a private, branded company ChatGPT, connected to each customer's own tools, sold at $12 a seat. The open-source core already exists. LibreChat (MIT, about 46.6K GitHub stars) gives you the chat UI, agents, skills, MCP, multi-user auth, an admin panel and token balances. Shopify runs its internal AI on it, with thousands of custom agents and more than 30 internal MCP servers, according to ClickHouse's announcement when it acquired the project in November 2025. What LibreChat does not give you is the SaaS around it: signup, provisioning, isolation between customers, seat billing and cost control. This post is the architecture for that layer. !Two founders sketching a multi-tenant private ChatGPT architecture: control plane, tenants, Claude and MCP connectors Requirements Functional - A customer signs up, picks a plan and gets acme.yourapp.com within minutes - Their staff sign in with the customer's own identity provider (Google Workspace, Microsoft Entra ID, Okta) - Claude as the model, MCP connectors to the customer's tools, shared agents per customer - Billing per seat, updated automatically as people join and leave Non-functional - Hard isolation of chats, files and connector tokens between customers - Bounded model spend per user and per customer - Upgrades that can be rolled to one customer first - Target cost per seat of about $3 to keep a 75% gross margin at $12 Silo, not pool LibreChat is designed as one deployment per organization: one librechat.yaml, one .env, one user base, one admin panel. You could try to bolt tenant IDs onto every collection, but you would be fighting the codebase on every upgrade. So the data plane is one stack per customer (the silo model), and the shared parts live in a small control plane you write yourself. | Concern | Silo (one stack per tenant) | Pool (shared stack) | |---|---|---| | Data isolation | Separate database per tenant | Tenant ID on every query | | Per-tenant config, branding, MCP servers | Native: separate yaml and env | Needs a custom config layer | | Noisy neighbor | Contained | Shared blast radius | | Upgrades | Canary per tenant | All at once | | Infra cost per tenant | Higher | Lower | | Fit with upstream LibreChat | Clean | Fork pressure | Start silo. When you have hundreds of small tenants, pack several tenant containers per host, but keep separate databases. The architecture Control plane (yours) +================================================+ | Signup + plan Tenant registry (Postgres) | | Stripe webhooks Provisioner Seat sync job | | Usage + alerts Upgrade orchestrator | +================================================+ | creates / updates v Data plane: one stack per tenant (acme.yourapp.com) +================================================+ | LibreChat API + web (pinned image tag) | | MongoDB database: acme (users, chats, balance) | | Meilisearch index (chat search) | | RAG API + pgvector (file search) | | MCP servers for this tenant's tools | +================================================+ | | Anthropic API Customer IdP (OpenID) (one Workspace per tenant, Customer tools via MCP own key + spend limit) (per-user OAuth) Three decisions do most of the work: 1. One Anthropic Workspace per tenant. The Anthropic Console lets you group API keys into workspaces, with a monthly spend limit and rate limits per workspace. Each tenant stack gets its own workspace-scoped key. Spend reporting per customer comes for free, and a runaway tenant hits its own ceiling, not yours. 2. OpenID per tenant. Each tenant's .env points at the customer's identity provider. People leave the company, they lose access. No passwords for you to store. 3. Per-user OAuth for MCP. LibreChat supports stdio, sse, websocket and streamable-http MCP servers, and with requiresOAuth: true each person connects their own account. The assistant can only see what that person can see in the underlying tool. The tenant config template The provisioner renders this per tenant. The values that change are the tenant name, the spend cap and the MCP servers. version: 1.3.18 interface: customWelcome: 'Welcome to Acme AI. Ask about your clients, files and projects.' balance: enabled: true startBalance: 6000000 # 1,000,000 credits = $1, so $6 of tokens autoRefillEnabled: true refillIntervalValue: 30 refillIntervalUnit: 'days' refillAmount: 6000000 refillMode: 'reset' # replace the allowance each period, no rollover display: 'currency' mcpServers: documents: type: streamable-http url: https://mcp.acme.yourapp.com/documents/mcp requiresOAuth: true timeout: 60000 And the branding in the tenant .env: APP_TITLE="Acme AI" CUSTOM_FOOTER="Acme AI. Private assistant for Acme staff." ALLOW_REGISTRATION=false ALLOW_SOCIAL_LOGIN=true OPENID_ISSUER=https://login.acme-idp.example.com OPENID_CLIENT_ID=from-the-customer OPENID_BUTTON_LABEL="Sign in with Acme" ANTHROPIC_API_KEY=key-from-the-acme-workspace The refillMode: 'reset' line matters for a seat business. With the older add behavior, unused allowance rolls over and a quiet user can bank a large balance. Reset caps every person at the same monthly ceiling. Check that every replica runs a version that supports reset before you switch, as the example config warns. Provisioning flow // control-plane/provision.ts (sketch) export async function provisionTenant(t: Tenant) { const ws = await createAnthropicWorkspace(t.slug, { monthlyLimitUsd: t.seats * 6 }); await createMongoDatabase(t.slug); await writeConfig(t.slug, renderYaml(t), renderEnv(t, ws.apiKey)); await deployStack(t.slug, { image: `librechat:${PINNED_TAG}` }); // ECS service or compose on a host await createDnsRecord(`${t.slug}.yourapp.com`); await waitForHealthy(`https://${t.slug}.yourapp.com/health`); await registry.markActive(t.id); } createAnthropicWorkspace is a placeholder for however you manage workspaces and keys; whether you do it by API or once by hand in the Console, keep the key in a secrets manager and inject it at deploy time. Seat billing that follows reality Bill on seats provisioned, sync once a day, and let Stripe prorate. // control-plane/seat-sync.ts import Stripe from "stripe"; const stripe = new Stripe(process.env.STRIPE_SECRET_KEY!); export async function syncSeats(t: Tenant) { const users = await tenantDb(t.slug).collection("users").countDocuments({}); const seats = Math.max(users, t.minimumSeats); if (seats !== t.billedSeats) { await stripe.subscriptionItems.update(t.stripeSeatItemId, { quantity: seats, proration_behavior: "create_prorations", }); await registry.setBilledSeats(t.id, seats); } } Stripe Billing costs 0.7% of billing volume on pay as you go, plus the standard US card fee of 2.9% plus 30 cents. On a $600 monthly invoice for 50 seats that is about $0.44 per seat. Cost control in three fences 1. Per user: LibreChat balances, as configured above. When a user's allowance is spent, their requests stop until the next reset. 2. Per tenant: the Anthropic Workspace spend limit, set to roughly seats times the per-user cap. 3. Per model: route by task. Claude Haiku 5.5 is $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100K tokens. Claude Sonnet 5.5 is $2 and $10. Make Haiku the default for quick questions and send agents and long documents to Sonnet. The model mix is the biggest lever in the whole business. With 300 messages per seat per month, 4,000 input and 600 output tokens per message: | Mix | Token cost per seat per month | |---|---| | 100% Sonnet 5.5 | $4.20 | | 30% Sonnet, 70% Haiku | $1.41 | | 100% Haiku 5.5 | $0.21 | Add about $1.20 per seat for a $60 monthly hosting budget on a 50-seat stack (the t4g.medium that runs the app lists at about $24.53 a month on demand in us-east-1) and the $0.44 payments line. The routed mix lands at about $3.05 per seat, a 75% gross margin at $12. Prompt caching, which these numbers ignore, only improves it. Upgrades without fear - Pin the LibreChat image tag per tenant in the registry. - Keep one internal tenant on the newest tag, then one friendly customer, then everyone. - Render configs from templates so a new config key is one change, not 50. - Run a smoke test after each tenant upgrade: login, one Haiku message, one Sonnet message, one MCP tool call. What fails first - MCP OAuth tokens expire or get revoked. Make reconnecting a one-click path for users. - Long conversations grow context with every turn. Nudge people to start a new chat per task. - Slow MCP servers make the whole answer slow. Set per-server timeouts and alert on them. - Seat disputes. Show every customer admin the exact list of users being billed. Build it with us Axionry is an AI product development studio. We build private company assistants and multi-tenant AI products like this one, with Claude, MCP connectors, SSO and spend controls, and we ship them in weeks. The product page is at axionry.com, and you can book a call to talk through your version. Sources - ChatGPT Business pricing: openai.com - Claude Team pricing update (Jan 2026): claude.com - Microsoft 365 Copilot Business pricing: microsoft.com - ClickHouse acquires LibreChat, Shopify figures (4 Nov 2025): businesswire.com - LibreChat repo: danny-avila/LibreChat - LibreChat MCP server config: librechat.ai - LibreChat balance config: librechat.ai - Anthropic API pricing: platform.claude.com - Anthropic Console workspaces: claude.com - Stripe Billing pricing: stripe.com - AWS t4g.medium price (Holori): calculator.holori.com

How to Build Your Own Team ChatGPT With LibreChat, Claude and MCP (Pay per Token, Not per Seat) Most companies buy AI as seats. ChatGPT Business is $20 per user per month on annual billing and $25 on monthly billing, and OpenAI says more than 9 million paying business users now rely on ChatGPT for work. For a 200-person company that is $48,000 to $60,000 a year, whether people use it or not. This guide builds the alternative that a lot of engineering teams are moving to: your own "team ChatGPT" on LibreChat (open source, MIT, about 46.6K GitHub stars), with Claude as the model, MCP connectors for Google Drive and Gmail, shared skills and agents, company SSO, and per-user spend caps. It runs in your own cloud account and you pay per token instead of per seat. !A team using a self-hosted team ChatGPT built on LibreChat and Claude Everything below comes from the current LibreChat example config (librechat.yaml version 1.3.18) and docs. Check the docs for your release before you copy anything into production. What you are building Browser (SSO login) | LibreChat (web + API) -> MongoDB (users, chats, balances) | -> Meilisearch (chat search) | -> RAG API + pgvector (file search) | +-> Anthropic API (Claude Sonnet 5.5 / Haiku 5.5) +-> MCP servers (Google Workspace: Drive, Gmail, Calendar) Prerequisites - A Linux VM with Docker and Docker Compose (start with 2 vCPU and 8 GB RAM for a pilot) - A domain and TLS (any reverse proxy or a cloud load balancer) - An Anthropic API key - Optional: an OpenID Connect app in Google Workspace, Okta or Microsoft Entra ID for SSO Step 1: Clone and configure git clone https://github.com/danny-avila/LibreChat.git cd LibreChat cp .env.example .env cp librechat.example.yaml librechat.yaml cp docker-compose.override.yml.example docker-compose.override.yml In docker-compose.override.yml, uncomment the block that mounts your config into the API container: services: api: volumes: - type: bind source: ./librechat.yaml target: /app/librechat.yaml Step 2: Point it at Claude In .env, set a company-owned key (the default user_provided asks each user for their own key, which is not what you want for a team): ANTHROPIC_API_KEY=sk-ant-your-company-key ANTHROPIC_MODELS=claude-sonnet-5-5,claude-haiku-5-5 Why two models: on Anthropic's current price list, Sonnet 5.5 is $2 / $10 per million input / output tokens and Haiku 5.5 is $0.10 / $0.50 for prompts up to 100K tokens. Offer Haiku for quick rewrites and summaries, and keep Sonnet for real reasoning. That one choice can cut the token bill roughly in half. Step 3: Turn on spend caps before anyone logs in Per-token pricing has one risk: a single power user running long agent sessions. LibreChat has a built-in token balance. In librechat.yaml: version: 1.3.18 balance: enabled: true startBalance: 15000000 # 1,000,000 credits = $1, so about $15 autoRefillEnabled: true refillIntervalValue: 30 refillIntervalUnit: 'days' refillAmount: 15000000 LibreChat counts 1,000 credits as $0.001, so 15,000,000 credits is roughly $15 of model usage per person per month, still under a $20 to $25 seat. Tune the numbers after the first two weeks of real usage. The point is that no single account can surprise you. Step 4: Company SSO Use OpenID Connect so people sign in with their work account and leave when they leave. In .env: ALLOW_SOCIAL_LOGIN=true OPENID_CLIENT_ID=your-client-id OPENID_CLIENT_SECRET=your-client-secret OPENID_ISSUER=https://your-idp.example.com OPENID_SESSION_SECRET=a-long-random-string OPENID_SCOPE="openid profile email" OPENID_CALLBACK_URL=/oauth/openid/callback Once SSO works, set ALLOW_REGISTRATION=false so nobody can create a local account around it. Step 5: MCP connectors for Drive and Gmail This is what turns a chat window into a work tool. LibreChat speaks the Model Context Protocol and supports stdio, sse, websocket and streamable-http servers. Each user's connection and OAuth tokens are kept per user and never shared. For Google Workspace, one option is the open-source google_workspace_mcp server (MIT), which covers Gmail, Drive, Calendar, Docs, Sheets and more, runs over streamable HTTP and supports OAuth 2.1 for multiple users. Run it next to LibreChat and register it: mcpServers: google-workspace: type: streamable-http url: http://workspace-mcp:8000/mcp requiresOAuth: true timeout: 60000 When someone first uses a Drive or Gmail tool, LibreChat asks them to sign in with their own Google account, so the assistant only sees what that person can already see. You can also pass the signed-in user to your own internal MCP servers with placeholders: internal-crm: type: streamable-http url: https://crm-mcp.internal.example.com/mcp headers: X-User-Email: '{{LIBRECHAT_USER_EMAIL}}' X-API-Key: '${CRM_MCP_KEY}' Step 6: Shared skills and agents Two features make the tool feel like it belongs to your company. Skills. LibreChat can mirror a GitHub repo of SKILL.md files, so your team keeps prompts and procedures (brand voice, proposal format, support macros) in version control: skillSync: github: enabled: true intervalMinutes: 60 runOnStartup: true sources: - id: company-skills owner: your-org repo: ai-skills ref: main paths: - skills skillDiscoveryDepth: 2 token: '${GITHUB_SKILLS_TOKEN}' Agents. In the Agent Builder, create agents for real jobs, add MCP tools from the Tool Library, and share them with a group. Three that pay off quickly: 1. Proposal drafter: reads the client folder in Drive, writes the first draft in your format. 2. Policy answerer: answers HR and IT questions from your handbook files, with sources. 3. Inbox triage: summarises unread threads and drafts replies for a human to send. Step 7: Start it and set roles docker compose up -d Use the admin panel to manage users, groups and roles: who can create agents, who can add MCP servers, and which groups see which agents. What it costs Assume 200 people, 25 messages each per working day, 4,000 input and 600 output tokens per message: | Setup | Monthly cost | |---|---| | ChatGPT Business, 200 seats | $4,000 to $5,000 | | LibreChat + Sonnet 5.5 only (+ ~$300 hosting) | about $1,770 | | LibreChat + half Haiku 5.5 (+ ~$300 hosting) | about $1,070 | These are estimates from the stated assumptions, not quotes. Run them with your own usage numbers. When not to do this - Under about 50 daily users, the seat plan is cheaper than the engineering time. - If nobody on the team can own upgrades, backups and connector issues, do not self-host. - If the only reason is cost, check the numbers twice. The strongest reasons are data in your own cloud, multiple model vendors, and connecting internal systems. Need it built? Axionry is an AI product development studio. We build private team assistants like this, with your SSO, your connectors and Claude or open-weights models, in a couple of weeks. Here is the product page: axionry.com. Sources - OpenAI, 9 million paying business users (27 Feb 2026): openai.com - ChatGPT Business pricing: openai.com - LibreChat repo: danny-avila/LibreChat - LibreChat MCP server config: librechat.ai - LibreChat balance config: librechat.ai - Anthropic pricing: platform.claude.com - Google Workspace MCP server: taylorwilsdon/google_workspace_mcp

எல்லா AI model-களும் ஒரே chat-ல். அதுதான் LibreChat. OpenAI, Claude, Gemini, Ollama எல்லாம் ஒரே இடத்தில். Docker-ல் இயக்குவது எளிது. முதலில் எதை இணைப்பீர்கள்? danny-avila/LibreChat Instagram-லும் பின்தொடருங்கள்: instagram.com

在5个AI聊天窗口间来回切，太累了。 LibreChat 把 ChatGPT、Claude 等塞进一个界面，还开源可自部署。 注意：免的是界面，不是API钱。 怎么用：docker compose 起来即可。 仓库：danny-avila/LibreChat t.co

LibreChat: 12,283 fingerprints, 13,265 titles, and a chat front end that stores the keys to every model LibreChat: 12,283 fingerprints, 13,265 titles, and a chat front end that stores the keys to every model A chat application is a friendly interface over a credential store. LibreChat's deployment documentation asks the operator to supply provider keys for the models it will call, and those keys live in the application's configuration, which means the application is a secret store with a text box on it. The measurements On 23 October 2026, ZoomEye returned 12,283 matches for app="LibreChat", 13,265 matches for title="LibreChat", and 7,588 matches for title="LibreChat" && port="3080". The fingerprint and title counts are within eight percent of each other, which is the pattern of a clean signature. The port-constrained count covers roughly fifty-seven percent of the title population, so the 3080 default is common but not dominant. A majority of deployments sit somewhere else, typically behind a proxy on a standard TLS port. What the application holds LibreChat stores conversation history per user, uploaded files used in conversations, and the configuration for each model endpoint it can call. Providers are configured with keys, and the administrator interface is where those keys are entered. The consequence is that a compromised deployment yields two distinct losses. The conversation history is a data loss, and its value depends on what users typed into it. The provider keys are a financial and access loss, and their value depends on what those keys can reach in the upstream account. A key that can create fine-tuned models, list files, or call the most expensive model in the catalogue is worth more than a key scoped to one endpoint. That distinction lives entirely on the provider side, and it is invisible in any external measurement. Reading the port split The 7,588 hosts on 3080 are deployments that kept the documented default. The remaining 5,677 title matches are deployments that changed the port or put a proxy in front, which usually indicates a deliberate production setup. Neither group is more secure by construction. A proxy terminates TLS and can add a source restriction, and it can also be configured to pass everything through without one. Port 3080 direct means the application answers without that intermediary, which removes a layer of policy but also removes a layer of complexity. The useful inference is narrower: a deployment that moved off the default port has an operator who made at least one deliberate decision. A deployment that kept it may not have. What an operator should check Confirm who can register. LibreChat supports multiple authentication strategies and the choice determines whether an outsider can join. An instance that accepts open registration is an instance that hands out access to whatever upstream keys the administrator configured. Then review the model endpoint configuration. A provider key entered once and never rotated is common, and rotation is a five-minute task that changes the blast radius of a leak. Then decide what conversation retention means for your organisation. A chat interface that keeps every conversation indefinitely is a data store that grows without a review. Retention settings and export controls are the levers, and both are configuration rather than code. What the numbers cannot tell you Twelve thousand matches describe a population, and the population includes test instances, personal deployments, and production installations with equal weight. Nothing in the query distinguishes them, and nothing in the query reveals a release version. The honest conclusion is about scale rather than defect. This is a product deployed widely enough that automated scanning against it is worthwhile for someone, and its role as a credential holder makes a compromise more expensive than the visible interface suggests. The pattern across this batch Every product measured here has the same structural property: a small interface in front of a large amount of trust. The counts differ by three orders of magnitude between the smallest and the largest, and the question an operator faces does not change with the number. What does this interface hold, who can reach it, and what would the holder of its credentials be able to do next. References - LibreChat documentation, librechat.ai - LibreChat project repository, danny-avila/LibreChat - ZoomEye AI query results collected 23 October 2026 for app="LibreChat", title="LibreChat", and title="LibreChat" && port="3080", sub-type all

LibreChat: Enhanced ChatGPT Clone with Agents, MCP, Skills, Code Interpreter and AI model switching (46k stars) What it does All text below is quoted directly from the project README. >Enhanced ChatGPT Clone: Features Agents, MCP, Skills, DeepSeek, Anthropic, AWS, OpenAI, Responses API, Azure, Groq, o1, GPT-5, Mistral, OpenRouter, Vertex AI, Gemini, Artifacts, AI model switching, message search, Code Interpreter, DALL-E, and more. Features: - UI & Experience inspired by ChatGPT with enhanced design and features - AI Model Selection: Anthropic (Claude), AWS Bedrock, OpenAI, Azure OpenAI, Google, Vertex AI, OpenAI Responses API (incl. Azure) - Code Interpreter API: Secure, Sandboxed Execution in Python, Node.js (JS/TS), Go, C/C++, Java, PHP, Rust, and Fortran - Agents & Tools Integration: No-Code Custom Assistants: Build specialized, AI-driven helpers - Web Search: Search the internet and retrieve relevant information to enhance your AI context - Generative UI with Code Artifacts: Code Artifacts create React, HTML, and Mermaid content directly in chat - Resumable Streams: Never lose a response: AI responses automatically reconnect and resume if your connection drops - Speech & Audio: Chat hands-free with Speech-to-Text and Text-to-Speech >Bundled with the Docker Compose stacks for one-command setup Repository LibreChat-AI/LibreChat Repository stats |Metric|Value| |:-|:-| |Stars|45,960| |Forks|9,453| |Last release|None| |Languages|TypeScript 84.8%, JavaScript 14.7%, CSS 0.2%| |Contributors|474| |AI contributor accounts (Claude / Codex / Cursor / Copilot)|Copilot: 18, claude: 64| |AI co-authored commits|Claude: 75, Cursor: 3, Copilot: 25| Main contributor |Field|Value| |:-|:-| |Name|Danny Avila| |GitHub|@danny-avila| |Commits to this repo|3,996 (top contributor)| |X / Twitter|@lgtm_hbu| |Followers|1350| |Public repos|38| |On GitHub since|2022-08-01| ────────────────────────────── 📌 How many great GitHub repos have you discovered and completely forgotten about? Don't let another useful tool disappear into your Reddit feed. Get the community's best repos, organized by category, plus key takeaways from real developer discussions, delivered to your inbox twice a week. 📬 Never lose track of great repos again. Subscribe free.

🔥 Host Your Own ChatGPT & Claude With This Open-Source Tool! #github #coding #automatedsolutions #ai "Tired of paying for multiple AI subscriptions or having your data tracked by commercial providers? 🛑 Meet LibreChat—the ultimate open-source, self-hosted AI chat UI that brings ChatGPT, Claude, Gemini, and local models together under one gorgeous, customizable dashboard! 💻🚀 🔗 Get the tool here: danny-avila/LibreChat 👇 Don't forget to LIKE, SUBSCRIBE, and hit the bell for more cutting-edge open-source tech and developer secrets!" #AI #OpenSource #GitHub #ChatGPT #Claude #TechShorts #ArtificialIntelligence #developer

GIAO DIỆN CHATBOT ĐA MÔ HÌNH ĐỈNH CAO THAY THẾ CHATGPT WEB: LIBRECHAT TÍCH HỢP AGENT, MCP VÀ CODE ARTIFACTS Sử dụng trực tiếp giao diện web của ChatGPT hoặc Claude khiến bạn bị giới hạn trong hệ sinh thái của một nhà cung cấp duy nhất và phải trả phí định kỳ cho từng tài khoản. Trong khi đó, việc tự host một giao diện trò chuyện linh hoạt giúp bạn tận dụng tối đa API key, chia sẻ cho cả công ty dùng chung và bảo mật lịch sử hội thoại nội bộ. LibreChat là giao diện web trò chuyện AI mã nguồn mở tiên tiến nhất hiện nay, cung cấp trải nghiệm sử dụng tinh tế hệt như ChatGPT Plus nhưng hỗ trợ kết nối đồng thời với tất cả các mô hình AI trên thế giới. Tech Stack: Xây dựng trên nền tảng Node.js, Express, React và Tailwind CSS, sử dụng MongoDB để lưu trữ dữ liệu hội thoại và PostgreSQL/pgvector cho tính năng RAG. Tại sao nên dùng? - Kết nối không giới hạn mọi nhà cung cấp AI: Hỗ trợ OpenAI, Anthropic Claude, Google Gemini, Mistral, Perplexity, Groq cùng các mô hình chạy cục bộ thông qua Ollama và vLLM. - Hỗ trợ giao thức MCP và AI Agents: Tích hợp đầy đủ Model Context Protocol (MCP) của Anthropic, cho phép người dùng tự tạo các Custom GPTs có khả năng gọi tool ngoài, duyệt web và thực thi code. - Tính năng Artifacts và Code Execution: Hiển thị trực quan các đoạn mã HTML/React và SVG tương tác ngay trong khung chat tương tự như giao diện của Claude.ai. Dự án sở hữu hơn 22.000 stars trên GitHub và là sự lựa chọn số một cho các tổ chức muốn xây dựng cổng giao tiếp AI tập trung cho toàn thể nhân viên. Link tham khảo: danny-avila/LibreChat #OpenSourceVN #VNOSS #opensourcevn #vnoss #DevCommunity #LibreChat #ChatGPTAlternative #AIAgent #MCP #React #NodeJS