Etles

Etles is a production-grade, open-source autonomous AI agent platform that runs an entire business end-to-end — from executive strategy and department management to sales, marketing, engineering, finance, customer success, and operations.

Read Docs · Features · Executive Layer · Model Providers · Deploy Your Own · Running locally

Features

Executive Command Layer

• C-Suite Agents: CEO, COO, CFO, Chief Revenue Officer, Chief Product Officer, Chief Risk Officer, and Chief People Officer — each with aggressive autonomous execution directives ("act first, report later").
• Cross-Departmental Orchestration: Executive leads synthesize metrics, enforce OKRs, resolve inter-departmental blocks, and present unified intelligence.
• Business Framework Seeding: On first run, the knowledge graph is auto-populated with a Business Model Canvas, KPI tree (CAC, LTV, NRR, ARR), OKR templates, and Weekly Business Review (WBR) scorecards — giving every agent a shared operating model from day one.

Autonomous Sub-Agents Framework

• 85+ specialized agents out of the box, organized into 15 departments:
  • Executive & Operations, Sales & Revenue, Marketing & Brand, Engineering & Infrastructure, Product & Design, Finance & Accounting, Customer Success & CX, HR & Talent, Data & Analytics, Research & Strategy, Legal & Compliance, Security & Trust, Content & Studio, Supply Chain & E-Commerce, and Partnerships & Alliances.
• Each department has a dedicated department lead that owns KPI dashboards, runs internal standups, escalates only exceptions to the C-suite, and reports weekly scorecards.
• Agents operate intelligently out-of-band to perform complex, multi-step actions and proactive delegations.

Agent-to-Agent (A2A) Collaboration

• Any agent can spawn child agents, wait for their results, and synthesize outputs into a unified answer.
• Supports parallel fan-out (spawn multiple agents simultaneously), sequential workflows (chain outputs), and conditional branching.
• Agents share memory via readDepartmentMemory / writeDepartmentMemory so specialist work stays aligned within departments.

Deep Triggers & Automations (via Composio)

• Connect to over 1000+ platforms (Gmail, Slack, Salesforce, Stripe, GitHub, Jira, HubSpot, Linear, and more).
• Multi-account support: Connect multiple accounts per toolkit (e.g., work + personal Gmail) with aliases for each connection.
• Configure active background triggers in real-time, enabling reactive agent workflows.

Advanced Agent Toolkit

• Memory: Agents save, recall, and update long-term user memories via Upstash Vector.
• Knowledge Graph: Structured business knowledge (entities, relations, facts) stored per-user in Redis. Agents can upsertKnowledgeEntity, searchKnowledgeGraph, and addKnowledgeRelation.
• Goals & OKRs: Create, update, track, and delete goals with progress logging.
• Planner: Create multi-step plans with task tracking.
• Scheduling: Fully conversational cron jobs and reminders powered by Upstash QStash.
• Missions: Multi-week autonomous campaigns (e.g., "get 50 beta users") with daily check-ins.
• File Storage: Store and retrieve files with Vercel Blob.
• Generative UI: Interactive components natively stream charts, documents, and real-time weather into the chat.
• Rich Artifacts: Persisted, editable Markdown reports, planners, dashboards, presentations, PDFs, and spreadsheets. Presentations support generated images, native charts, tables, speaker notes, varied slide layouts, and .pptx export; dashboards support filters, responsive charts, detail rows, and CSV export.
• Artifact Guidance: Built-in wiki playbooks teach the agent how to create high-quality presentations, planners, and dashboards without mock data or repetitive layouts.
• Sandboxed Code Execution: Secure Daytona sandboxes for code execution, Git operations, and browser automation.
• Persistent Browser: User-scoped Daytona browser profiles preserve Playwright storage state across sessions while enforcing strict sandbox ownership.
• BYOK Fallback: Environment credentials remain primary; encrypted per-user profile credentials are used only when deployment credentials are unavailable.
• Cloud Infrastructure Tools: AWS (S3, EC2, Lambda), GCP (Storage, Compute, Functions), Azure (Storage, VM, Functions).
• Database Tools: Direct query access to PostgreSQL, MySQL, and MongoDB.
• Legal Tools: Contract analysis and comparison.
• Communication: Twilio voice/SMS, Twilio WhatsApp, Telegram, and multi-platform chat SDK (Slack, Teams, Discord, GChat, GitHub, Linear).

Human-in-the-Loop (HITL) Approvals

• Mandatory safety gate for irreversible actions (emails, payments, social posts).
• Drafts are queued in Redis and sent to Telegram with Approve / Edit / Reject buttons.

Technology Stack

• Next.js App Router with RSC and Server Actions.
• AI SDK — unified API for text, structured objects, and tool calls. Supports OpenAI, Anthropic, Google, xAI, DeepSeek, and more via AI Gateway.
• shadcn/ui with Tailwind CSS and Radix UI primitives.
• Neon Serverless Postgres for chat history and user data.
• Vercel Blob for file storage.
• Upstash Redis for session state, knowledge graph, and agent status.
• Upstash Vector for semantic long-term memory.
• Upstash QStash for durable scheduling and webhook delivery.
• Upstash Workflow for durable multi-step agent workflows.
• Auth.js for secure authentication.
• Composio for 1000+ platform integrations.
• Daytona for secure code execution sandboxes.

Model Providers

This template uses the Vercel AI Gateway to access multiple AI models through a unified interface. The default model is openai/gpt-5.6-luna, with capabilities and availability defined in lib/ai/models.ts. The registry includes models from:

• Anthropic: Claude Haiku 4.5, Sonnet 4.5/4.8, Opus 4.8, and thinking variants
• OpenAI: GPT-4o, GPT-4.1, GPT-5 Mini/Nano, GPT OSS 120B
• Google: Gemini 3 Flash/Pro, Gemini 3.1/3.5 Flash, Gemma 4 26B
• DeepSeek: V3.1 and V4 Flash/Pro variants
• xAI: Grok 3, Grok Build 0.1
• Alibaba: Qwen 3.6 Max, Qwen 3 Coder Plus, Qwen 3.8 Max
• Moonshot AI: Kimi K2.5/K2.6/K2.7 Code
• MiniMax: M2.5/M2.7/M3
• ZAI: GLM-5/5.1/5.2/5V Turbo
• Perplexity: Sonar
• NVIDIA: Nemotron 3 Nano
• Inception: Mercury 2

AI Gateway Authentication

For Vercel deployments: Authentication is handled automatically via OIDC tokens.

For non-Vercel deployments: Set the AI_GATEWAY_API_KEY environment variable in your .env.local file.

Deploy Your Own

You can deploy your own version of Etles to Vercel with one click:

Running locally

You will need to use the environment variables defined in .env.example to run Etles. It's recommended you use Vercel Environment Variables for this, but a .env file is all that is necessary.

Note: You should not commit your .env file or it will expose secrets that will allow others to control access to your various AI and authentication provider accounts.

1. Install Vercel CLI: npm i -g vercel
2. Link local instance with Vercel and GitHub accounts (creates .vercel directory): vercel link
3. Download your environment variables: vercel env pull

pnpm install
pnpm db:migrate # Setup database or apply latest database changes
pnpm dev

Your app should now be running on localhost:3000.