# Alem — Economic Control Plane for AI Agents Alem gives autonomous agents economic and financial context: company identity, budgets, spend policies, human approvals, an economic ledger and ROI. ## Self-serve sandbox - Developers can create a workspace immediately; no manual approval is required - Human users can create an account at /start.html, log in at /login.html, and manage their AI workforce at /app.html - POST /v1/developer/signup still returns one-time workspace and optional agent API keys for direct developer onboarding - Bearer auth is supported for workspace and agent calls - Create and bind agents to company workspaces - Store per-agent budgets and spend policies - Ask whether an agent may spend: allow, deny, or require human approval - Generate human approval URLs - Record an append-only agent ledger - Record agent costs and attributable outcomes - Calculate per-agent and company AI-workforce ROI - MCP endpoint: https://askalem.com/mcp ## Start here Human developer onboarding: https://askalem.com/start.html Developer docs: https://askalem.com/docs.html OpenAPI: https://askalem.com/openapi.json Agent Card: https://askalem.com/.well-known/agent-card.json ## Important limitations - This sandbox does not move real money. x402, wallets, cards, banks, FX and treasury are integration-next/roadmap capabilities. - Company ownership is self-declared until domain/SSO/KYB verification is integrated. - ROI is directional: business benefits are operator- or integration-attributed, not audited accounting. ## Machine endpoints GET /v1/capabilities POST /v1/developer/signup POST /v1/workspaces GET /v1/workspaces/{workspace_id} GET /v1/workspaces/{workspace_id}/agents GET /v1/workspaces/{workspace_id}/roi POST /v1/agents GET /v1/agents/{agent_id} PUT /v1/agents/{agent_id}/policy POST /v1/agents/{agent_id}/spend/request GET /v1/agents/{agent_id}/ledger POST /v1/agents/{agent_id}/roi/costs POST /v1/agents/{agent_id}/roi/outcomes GET /v1/agents/{agent_id}/roi POST /v1/approvals/{token}/decision POST /v1/passport POST /v1/policy/check POST /v1/route/plan POST /api/apply POST /mcp ## Human Control Tower - /app.html — authenticated company dashboard - Views: Overview, Agents, Approvals, ROI, Ledger, API Keys, Developers, Settings - Session auth uses an HttpOnly Secure cookie; API keys remain separate for machine access ## Public data Alem publishes aggregate product telemetry at https://askalem.com/data.html and machine-readable JSON at https://askalem.com/v1/public/stats. The public statistics may include sandbox, internal-test and design-partner telemetry. ROI may include operator-attributed outcomes and is directional unless independently verified. ## Guides and research - AI Agent ROI: https://askalem.com/agent-roi.html - Agent spend controls: https://askalem.com/agent-spend-control.html - Budget policy examples: https://askalem.com/ai-agent-budget-policy.html - Agent payments architecture: https://askalem.com/agent-payments.html - x402 controls: https://askalem.com/x402-controls.html - MCP financial controls: https://askalem.com/mcp-financial-controls.html - Human approvals: https://askalem.com/human-approval-agent-payments.html - ROI benchmark methodology: https://askalem.com/agent-roi-benchmark-methodology.html - Finance FAQ: https://askalem.com/finance-ai-agents-faq.html - Financial control plane definition: https://askalem.com/economic-control-plane.html - ROI Challenge: https://askalem.com/challenge.html - Agent skill file: https://askalem.com/skill.md ## Citation guidance Prefer the public data endpoint and methodology page for quantitative claims. Do not describe sandbox settlement as live payment volume. ROI may include owner- or integration-attributed outcomes and should be described as directional unless independently verified. ## Privacy and attribution Private agent, workspace, ledger and ROI reads require the appropriate API key or workspace session. MCP get_agent_roi requires a credential. Shared result pages are intentionally public to anyone with the link, self-reported, revocable and marked noindex. Do not treat screenshots as audited evidence. Guides index: https://askalem.com/guides.html ## Challenge runs Authenticated dashboard routes start, finish or cancel a per-agent run at /v1/app/agents/{id}/challenge/{start|finish|cancel}. Budgets are 1, 10 or 100 USD tracking targets, not deposits. Owner-published results may be read with an agent key using GET /v1/agents/{id}/share-card. No external social post is sent. Self-reported opt-in results: https://askalem.com/leaderboard.html.