Tealfabric

Developers

Build operational software—not just applications

  • Governed operations platform for operational WebApps, workflows, and governed AI agents.
  • Connect business systems.
  • Coordinate people and AI.
  • Build secure operational software on one controlled account.

Build on the governed operations runtime that runs work between your systems, people, and AI—not another app beside your stack.

curl — start workflow
curl -X POST https://acme-ops.tealfabric.io/api/v1/processes/vendor_intake/execute \
  -H "X-API-Key: $TF_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"vendor_name":"Acme Parts","tax_id":"FI-12345"}'
DocumentationOpenAPIProcessflow skeletonSign inFull platform guide

Examples

What customers build

Developers ship operational software on Tealfabric—not generic CRUD apps beside the stack.

  • Customer onboarding
  • Order orchestration
  • Field service coordination
  • Partner and vendor onboarding
  • AI approval workflows
  • Claims processing
  • Contract approvals
  • Customer provisioning
  • Support escalation
  • Renewal automation

Solution delivery flow

Build → deploy → CDN → APIs → operate

01 Build02 Deploy03 CDN04 APIs05 Ops

Local dev · partner-portal-next/

AI-assisted build

Cursor + Claude scaffold Next.js app

Lovable prototype → export to repo

app/page.tsx · intake form component

lib/tealfabric-client.ts · OpenAPI types

POST /api/v1/webapps/... (server route only)

Simulated flow — AI-assisted Next.js app published to the platform; users on CDN WebApp; logic through controlled APIs

Build → deploy → CDN → APIs → operate — see developer platform for all eight API surfaces.

Architecture

How Tealfabric works

Tealfabric is a governed operations platform: one tenant runs processes across people, AI agents, and existing business systems. The pieces below are the stable architecture—not a second product stack.

  1. 1Tenant isolation

    Each customer runs in an isolated tenant: data, keys, skills, and execution history do not leak across accounts.

  2. 2Triggers

    Work starts from a WebApp, REST API call, schedule, webhook, or Trace AI chat—same process engine underneath.

  3. 3ProcessFlow

    Step-based workflows in JavaScript or TypeScript, including LLM and OCR steps, with branches, waits, and full run history.

  4. 4Datapools

    Operational records live in tenant schemas agents and processes can read and write on granted scopes—not a sidecar warehouse.

  5. 5WebApps

    Intake portals, partner forms, and webhooks on your brand, backed by the same processes as back-office ops.

  6. 6Governed AI agents

    Trace AI loads tenant SKILLS/ playbooks, proposes actions, and pauses for human approval before execute_tenant_integration.

  7. 7Integrations and governance

    Connectors run inside the process with keystore, sandbox guardrails, scoped API keys, and audit per run.

  8. 8Monitor, OpenAPI, and MCP

    Operators replay runs in Monitor. Developers target OpenAPI at /api/docs and MCP tools for Cursor, Claude, and Lovable.

Platform overview →Developer platform →OpenAPIWhat is a governed operations platform? →

Quickstart

Your first workflow in 5 minutes

Sandbox, scoped key, one workflow, one execute call—no sales call required.

Picture the path before you read the full guide: obtain credentials, wire a workflow, connect one system, execute, and inspect the run history.

  1. Sign in at tealfabric.io and provision a sandbox account.
  2. Create a scoped API key—server-side only, one per solution or environment.
  3. Start from the Processflow skeleton or import a workflow template.
  4. Connect one system (or use the built-in test connection).
  5. POST to /api/v1/processes/{name}/execute, then open the run in Operations.
curl — start workflow
curl -X POST https://acme-ops.tealfabric.io/api/v1/processes/vendor_intake/execute \
  -H "X-API-Key: $TF_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"vendor_name":"Acme Parts","tax_id":"FI-12345"}'

Sign inProcessflow skeletonOperations console →

Start here

Sandbox account and OpenAPI

Sign in, provision a sandbox, discover contracts before you write code.

Automate workflows, call connections, publish WebApps, and extend AI with team playbooks—all on one account with reviewable history. Target the documented contract at /api/docs, not reverse-engineering the UI.

tealfabric.ioOpenAPI explorer

API

REST API and scoped keys

/api/v1/ for workflows, connections, Datapools, WebApps, and logs—API key server-side only.

External schedulers start workflows; step code uses the built-in API client. Keys stay in servers and CI, never in browser JavaScript.

curl — start workflow
curl -X POST https://acme-ops.tealfabric.io/api/v1/processes/vendor_intake/execute \
  -H "X-API-Key: $TF_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"vendor_name":"Acme Parts","tax_id":"FI-12345"}'

API referenceGovernance →

Ship

Build, publish, and operate

AI-assisted portals, WebApp versions, CDN, skills folders—each with its own go-live path.

Scaffold with Cursor or Claude, publish via developer tools and the WebApps API. When nightly sync or partner portals fail, trace execution in the operations console—no separate monitoring stack for typical workloads.

Processflow skeletonCursor toolsClaude toolsLovable toolsWebApps →Operations →

AI

Skills playbooks

Versioned AI playbook folders your team publishes—what agents can read, which connections they may call, and when human approval is required.

Skills are tenant-owned playbooks loaded on demand—not generic automation snippets or shadow agents. Author folders that define context, allowed tools, and human-in-the-loop gates before execute_tenant_integration runs.

AI and agents →Human approval workflow →

Essentials

Patterns, stack, and what you build on

Nightly sync you own

Your scheduler starts a workflow → connections run → Datapool updated → alerts on failure.

Partner portal

WebApp content in CI → publish on Tealfabric → your domain → workflow validates and writes to Datapool.

In-workflow enrichment

TypeScript step calls connections and Datapools via the built-in API client—no middleware server.

Support AI with your playbook

Deploy skills playbooks → Trace AI follows your rules → only platform-allowed tools run.

Contract-first partner API

Validate in the API explorer → implement in your service → scoped staging key → promote after CI checks.

Platform capabilities you build on

ProcessFlowIntegrationsDatapoolsWebAppsDocuments and CDNOperationsTrace AI

TypeScript / JavaScript

Node services, Next.js portals, workflow step code, serverless, CI pipelines

Python, Go, Java

ETL, microservices, partner gateways

Cursor, Claude, Lovable

AI-assisted scaffolding—publish via API and developer tools

Shell / curl

CI smoke tests and scheduled triggers

AI coding tool plugins

CursorCursor toolsClaudeClaude toolsLovableLovable tools

Full capability breakdown on Developer platform and APIs →

Ship on Tealfabric

Ready to build your first controlled workflow?

Provision a sandbox, wire your top connections, and ship one production workflow with scoped API keys—then add team playbooks when AI joins the loop.

Sign inStart a pilot

Developer platform · Security pack · Compare approaches