Give an agent an API key and a budget. It calls 40+ services through one endpoint - email, SMS, voice, search, code execution, LLMs, and more - plus OAuth into Slack, Gmail, HubSpot, Salesforce, Jira, and Linear.
Schedule it. Let it delegate to other agents. Publish it to your team. You see every request and every dollar.
New services and operations ship every week. If a provider you depend on isn't in the catalog yet, tell us - we add what our users ask for.
Upivia sits between your agent and every paid API. Every call is metered, budgeted, policy-checked, and logged. Budget the 40+ built-in services alongside your own external APIs and custom endpoints - all through one interface.
Email, SMS/MMS, chat, and voice - metered per send or per minute.
emailsmsmms+2Frontier LLMs, speech synth/transcription, and embeddings - billed by tokens or seconds.
text_generationspeechembeddingLive search, URL fetch, headless browser, RAG knowledge, and CSV parsing.
web_searchwebbrowser+2Sandboxed code execution and outbound webhook fan-out with policy + audit.
codenotificationParse documents, render video, generate subtitles.
documentvideoOAuth-backed calendar, scheduling, CRM, and ticketing operations.
calendarschedulingcrm+1Install once at the org level. Every agent inherits the connection, scoped by policy, with token refresh and audit trail handled for you.
Your agent hits a single endpoint with a Bearer token. We check policies, debit the budget, dispatch the provider, and return a structured result.
$ curl https://www.upivia.com/api/v1/service-requests \
-H "Authorization: Bearer agent_key_live_..." \
-d '{
"operation": "email.send",
"payload": {
"to": "ada@example.com",
"subject": "shipped"
}
}'
{
"status": "executed",
"cost_cents": 1,
"balance_cents": 9999,
"request_id": "req_01HQ..."
}Email + password. No card required.
Pick services. Set a dollar budget per op.
Hand it to your agent. We hash and store securely.
Audit log, balance, and budgets update live.
Invite your team to develop agents together. Assign custom roles with granular permissions—finance approves budgets, engineers build agents, analysts view audit logs. Nobody gets access they don't need.
Role: Engineer ✓ agent.createOwn ✓ service.execute ✓ audit.viewOwn ✗ budgetRequest.review Role: Finance ✓ budgetRequest.review ✓ team.setBudget ✓ audit.viewOrg ✗ agent.manageAny
Execution is step one. Upivia also runs the lifecycle around it: scheduling, delegation, publishing, memory, and files - every action permission-checked, budgeted, and written to the audit log.
Tell an agent “every Monday at 9am” in plain English. Calendar, kanban, and list views; runs logged like any other request.
Agents hand work to other agents - with the delegate's own budget, permissions, and audit trail. No shared keys.
Promote an agent from private to team- or org-wide. Requests go through reviewer approval and a tamper-evident provenance log.
Agents remember across sessions - scoped, size-capped, and inspectable by admins instead of stuffed into prompts.
Agents read and write files through the same metered, policy-checked API. Presigned download, restore, audit included.
Personal access tokens for humans and tools; an MCP server for Claude Desktop and Cursor. Machine-readable docs at /api/v1/agent-docs.
Drop your email for updates as the catalog grows. Or, if you run a service or API you'd like to see wired into Upivia, send us a note and we'll get back with the integration path.