Features Everything a production backend needs, configured by your AI.
Data, events, logic, custom code, agents, connectors, multi-tenancy, realtime and docs, in one platform. Every item below is live today.
Data & APIs
Define entities; the API exists.
- Create, get, list, update, delete and count endpoints for every entity
- Search with filters, facets, sorting and pagination
- Lookups, formulas, categories, unique and composite indexes
- Public endpoints for forms and guest lookups, with no login
- Bulk CSV or JSON upload, insert-only or upsert
- File uploads served from a CDN, published publicly or shared by link
Events & automation
Every write can start something.
- Events on create and update of any entity
- Before events check a write and can reject it; after events react to it
- Destinations: templated email, Slack, SMS, webhooks, any connector, writes to another entity, S3, Logic Blocks, agent crews
- Entry conditions and field mapping from the record
- Run in the background or inline; enable and disable without deleting
- Delivery history and redacted traces for every run
- Inbound webhooks from connectors land as records and can start events in turn
Logic Blocks
Business logic as pipelines, not services.
- Read and write entities, branch with AND/OR conditions, loop, call connectors and webhooks, run formulas and transforms
- Run synchronously, or in the background with a job to poll
- Secrets in webhook steps, limited to an outbound domain allowlist
Custom code (Glyphs)
Code only where config stops.
- Python 3.12 and Node 22 functions that run sandboxed
- SDK for your data, HTTP, connectors, email, queues and events, plus a curated standard library
- Run as a Logic Block step, an agent tool, a schedule or its own endpoint
- Drafts, published versions and unlimited restorable history
- Large Glyphs: 2 GB memory and up to 14 minutes on Pro
AI & agent crews
A model in the path only where you put one.
- Typed AI steps: classify, extract, summarize, generate, decide, translate, analyze, validate, moderate
- Answers as text, JSON, a score, a category, a yes/no or a list; images and audio too
- A model and a token limit per step
- Read documents, images and audio; transcribe speech
- Crews: multi-agent graphs where independent agents run in parallel, each with its own model and memory
- Crew caps on tokens and turns, topic guardrails, and sliding-window or summary memory
- Human gates that pause a crew for approval, from the console or your app
- Anthropic, OpenAI, Gemini, DeepSeek, Kimi, Grok, or any provider your org admin defines with an OpenAI-compatible endpoint
Connectors
Adopt once; each workspace brings its own keys.
- Stripe, Paddle, Dodo Payments, PayPal, Email, Slack, Twilio SMS, Salesforce, Claude, Claude Code, Codex, Copilot, Jules, Muse Code, GitHub, Screenshot Service, Serper, Tavily, EasyPost, TaxJar, Google Maps, AWS, Amazon S3, Calendly, Zapier, Webhook
- Several named instances of the same connector, with personal or shared credentials
- Inbound webhooks land in your entities and match the record they belong to
- Any API not listed: the Webhook connector with no auth, bearer, basic, custom header or OAuth2 client credentials
Tenancy, sign-in & access
Built for apps with many customers.
- Single, hybrid or multi-tenant apps, with join-or-create workspaces and invites
- Workspace sign-in with password, Okta or Google, with domain checks and approval modes
- Console SSO with Okta or Google for your own team
- IAM-style access policies with Allow and Deny, enforced on the server
- Each org's and workspace's data in its own partitions; a dedicated stack on Enterprise
- Per-workspace settings, encrypted secrets, email templates and branding
Realtime, schedules & flags
The runtime pieces every SaaS needs.
- Realtime: a wss:// endpoint per app; channels scoped to the org and workspace on the server
- Stream Logic Block AI tokens and progress, every crew turn, and a live activity feed
- Broadcast a Logic Block result to everyone on a channel with notifyall
- Schedules: once, every N minutes, daily, weekly, monthly or cron, in any timezone
- Feature flags: targeting, percentage rollouts, scheduled changes, workspace overrides; JavaScript, React, Vue and Angular clients
Always-current docs
Generated from your config every time.
- Live OpenAPI 3.1 spec, downloadable for Swagger, Postman and other tools
- Architecture Studio: overview, data model, Logic Block and event flows, integration map, infrastructure diagram
- Request walkthroughs that flag themselves when something they pass through changed
- Built-in and custom App Flows with actors, steps and the AWS resources involved
- Asset descriptions that deploy with each asset, and PDF export, on every plan
Ship & operate
From first deploy to the hundredth customer.
- Dev → production publish, diff by content, activate and roll back
- Dashboard with KPIs, call log, error rate, p50 and p95 latency, agent cost and guardrail hits
- Usage per workspace: API calls, records, Logic Block runs, AI tokens and cost
- Delivery traces with redaction, health checks and a public status page
- Playground: try your API in the browser with ready-made scenarios
Build with your AI (MCP)
Ninety tools for any MCP client.
- Hosted MCP server for every org: Claude Code, Claude Desktop, Cursor and others
- Grounding tools and a live API reference so the AI builds against what exists
- Test what it built with test_logicblock and invoke_glyph
- Each call is checked against the caller's access policy and audited; MCP works in dev only
