Zite
Also known as: Zite, zite.com, Zite AI app builder, Zite Database Agent, Restly
AI app builder for internal tools and portals whose agent plans, builds and edits apps on a built in database in approve or auto modes, with model tier choice, an MCP server, SOC 2 Type II and unlimited users on every plan.
Zite is an AI app builder for business software, from the team behind the Fillout forms product. A builder describes an internal tool, portal or dashboard, and Zite's agent plans and builds it on a built in database, with authentication, workflows and integrations included. Every plan includes unlimited users, which is the company's main pricing pitch.
The agent works in four modes. Plan mode scopes the work first, Chat mode answers without touching the app, Build mode makes changes and asks for approval on certain ones, and Auto mode applies most changes without waiting. Even in Auto, the agent stops to sign into integrations, enter secrets or answer clarifying questions, and it may pause before hard to reverse changes such as deleting a table or widening data access, or before sending email or taking a payment. Every change lands in a Changes panel, auto-approved ones flagged, and version history restores earlier states. A separate Database Agent creates tables, analyzes records, updates fields and connects integrations.
Builders choose the model tier for each task, and Business plans can build with their own AI key. Zite is SOC 2 Type II, offers granular permissions, roles, user groups and enforced two factor authentication on Team, and adds SSO for app users, audit logs and an organization wide AI kill switch on Enterprise.
The agent only works when a person prompts it or an outside agent calls the MCP server; scheduled workflows and webhooks run the generated app's logic, not the agent. Testing is an Improve my app pass that runs automated review, design and accessibility checks, not a stored evaluation suite.
Vendor details
Canonical URL
https://www.zite.com
Category
Coding agent
Subcategory
AI app builder for business software
Funding status
Private company. Restly, Inc. is the company behind Zite and Fillout.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
The agent wires apps to 30+ native integrations, including Gmail, Outlook, Google Sheets, Airtable, HubSpot, Attio, Stripe, QuickBooks, Xero, Jira, Linear, Asana, Notion, Slack, Teams and Twilio, with OpenAI, Anthropic and Gemini available inside app workflows. A Zite MCP server lets outside AI agents query data, build apps and manage the workspace, and a REST API with a TypeScript SDK covers databases, tables, fields, records and webhooks.
In practice
An operations manager prompts Zite for an equipment rental tracker on its built in database, approves the agent's schema changes in Build mode, and shares the app with the whole company at no seat cost.
An agency builds a client portal with custom login and Stripe payments, then switches to Auto mode for a longer redesign pass.
A team lead connects Claude or ChatGPT to the Zite MCP server to query and update records in the company's Zite database.
Sources & related URLs
Related / legacy domains
Research sources
Agentic Index coverage score
9.5 / 14 capabilities · 68%
| Integrations & Tool Calling | Full |
|---|---|
|
The agents take authenticated action. The builder agent works on workflows and data and connects integrations, and Auto mode stops only to sign in to an integration or enter a secret. The Database Agent creates tables, updates fields and connects integrations, and generated apps act in 30+ native integrations such as Gmail, HubSpot, Stripe, QuickBooks, Jira and Slack. The per tool agent actions are described across pages rather than in one reference. Sourcezite.com/llms.txtread 2026-10-02 |
|
| Workflow Orchestration | Partial |
|
Multi step execution happens inside one agent. Plan mode scopes work before building, and Auto mode works through larger multi step builds. Separate chats keep context apart. No multi agent execution or orchestration of the agent inside workflows is documented, and app workflows run the generated app's logic. Sourcezite.com/help/platform/how-to-build/agent-modes/auto-mode.mdread 2026-10-02 |
|
| Knowledge Grounding & RAG | Partial |
|
Context rather than a retrieval layer. The agent inspects the app and its database, and a workspace Context and Integrations page lets builders add business context for the agent. No maintained index or knowledge base over the customer's documents is documented. Sourcezite.com/llms.txtread 2026-10-02 |
|
| Human Oversight & Guardrails | Full |
|
Zite ships approval before changes commit. Build mode, the default, asks the user to approve certain changes. Auto mode is opt in and still stops for sign in, secrets, new apps and clarifying questions, and may pause before deleting tables or fields, widening data access, sending email or taking a payment. Auto approved changes are flagged in the Changes panel, and version history restores earlier states. Enterprise adds an organization wide AI kill switch. Sourcezite.com/help/platform/how-to-build/agent-modes/auto-mode.mdread 2026-10-02 |
|
| Security, Identity & Governance | Full |
|
Zite documents both compliance and access controls. SOC 2 Type II applies on every plan, with 256-bit AES encryption and a security page describing safeguards. Team adds granular permissions, roles, user groups, enforced two factor authentication and per member AI credit limits, and Enterprise adds SSO for app users and audit logs. Sourcezite.com/pricingread 2026-10-02 |
|
| Observability & Auditability | Full |
|
Agent work leaves a runtime record. Every change the agent applies appears in the Changes panel, auto approved ones marked, with an agent summary per turn. Version history supports mass undo, workflow run history keeps detailed logs, and Enterprise audit logs track database changes, logins and permission updates. Retention of the Changes panel is not stated. Sourcezite.com/help/platform/how-to-build/agent-modes/auto-mode.mdread 2026-10-02 |
|
| Memory & State Persistence | Partial |
|
The agent keeps conversation state within a chat, managed through Multi-chat to control context, plus version history of the app. Workspace business context is authored instructions rather than memory. No memory layer with a stated scope and lifetime is documented. Sourcezite.com/llms.txtread 2026-10-02 |
|
| Deployment & Data Residency | Partial |
|
Vendor hosted only. Apps run on Zite's cloud, with no self hosted, private cloud or on premises option documented. A deployment region page lets customers move form submission data to EU hosting, but it covers Fillout forms and is not stated to apply to Zite apps or databases. Sourcezite.com/llms.txtread 2026-10-02 |
|
| Prebuilt Agents, Templates & Packs | Partial |
|
Few adoptable starting points are documented. The site links a Marketplace and a spreadsheet to app path, and workspaces can be duplicated and imported between organizations, but no template catalog for the agent is documented in the help center. Sourcezite.com/pricingread 2026-10-02 |
|
| Triggers & Channel Coverage | Partial |
|
The agents are user initiated only. Work reaches them through prompts in the builder and the database, and outside AI agents can call the Zite MCP server or the Zite app in ChatGPT. No autonomous wake for the agent is documented, and scheduled workflows and webhooks run app workflows, not the agent. Sourcezite.com/llms.txtread 2026-10-02 |
|
| Model Flexibility & Routing | Full |
|
The customer chooses the model. Builders select a model tier per task (Zite Mini, Zite Pro, Zite Max, Zite Max Fast), Business adds advanced models and fast mode, and Business plans can build with their own AI key. The underlying providers behind the tiers are not named. Sourcezite.com/pricingread 2026-10-02 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
Zite documents interfaces to its own platform. A Zite MCP server lets AI agents query data, build apps and manage the workspace, and a Zite app in ChatGPT inspects and updates databases. A REST API with a TypeScript SDK reads and manages databases, tables, fields, records and webhooks, with separate developer docs. Sourcezite.com/llms.txtread 2026-10-02 |
|
| Testing, Debugging & Optimization | Partial |
|
Zite offers automated review passes, not evaluation. Improve my app runs review, design and accessibility passes, Fix it for me asks the agent to diagnose and resolve an error, and builders can preview as a user and debug workflow runs. No stored, scored test cases or regression suite is documented. Sourcezite.com/llms.txtread 2026-10-02 |
|
| Browser & Computer Use | Unable to verify |
|
No browser, desktop or remote computer control by the agents is documented. The near miss is that generated apps can scrape and crawl pages through the Firecrawl integration, which fetches content rather than operating an interface. Sourcezite.com/llms.txtread 2026-10-02 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Unable to verify against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Recent platform changes
Zite added Auto, a fourth agent mode next to Build, Plan and Chat, that changes an app the way Build does but applies most changes without waiting for an Approve click. It still stops for integration sign ins, secrets, new apps, clarifying questions and pending workflow runs, and auto approved changes are marked in the Changes panel. Build remains the default.
Bears on: Human approval / guardrails
View sourcePricing
$20/mo billed annually ($25 monthly), Pro · free plan
flat organization plan with monthly AI credits and workflow run limits; unlimited users and editors
Included quota
Pro: 100 credits, 3 workspaces, 100,000 records, 5,000 workflow runs a month. Business: 400 credits, 5 workspaces, 250,000 records, 50,000 workflow runs. Team: 800+ credits, 250,000 runs.
What is public
Zite publishes Free, Pro and Business prices, credit allowances, record and workflow run limits and a full feature comparison. Team and Enterprise prices are not published.
Billing mechanics
Organization level plan billed monthly or annually, with a monthly credit allowance that resets on the subscription day and selectable credit multipliers on paid plans.
Cost watchouts
Credits cover AI building, the Database Agent, AI fields and generated documents, and usage varies with request complexity. Credits do not roll over, and the only way to get more before reset is a higher plan. Workflow runs from published apps are capped separately by plan.
Variable cost rationale
Plan fees are fixed and users are unlimited, but agent work draws a credit pool that varies by request complexity and only grows by upgrading.
Additional watchouts
Seats are free, so the budget question is credits: heavy agent building and AI fields draw on the same pool.
Overage / add-ons
No top up purchase; more credits come from upgrading the plan or raising its credit tier, which takes effect immediately for the current cycle.
Sales call required
Mixed (some tiers require a call)
Free / trial
Free forever with 50 monthly credits, a 10 message daily limit, 1 workspace and unlimited users
Lowest paid plan
Pro at $20 a month billed annually ($240 a year)
Commercial notes
Self serve on Free, Pro and Business; Team and Enterprise are scale plans with priority support and advanced security, sold through the Enterprise page.
Key ambiguities
Team and Enterprise prices are not published, and the monthly billed price for Business is inferred from the annual list.
Missing data
Team and Enterprise pricing.
Related vendors
- Cognition — Maker of Devin, an autonomous AI software engineer
- 10Web — Agentic website platform whose specialized AI agents build, host,…
- AgentUI — Managed AI app builder for operations teams that generates and hosts…
- Aider — Open source, model agnostic terminal coding agent that edits your…
- Anthropic Claude Code — Anthropic's agentic coding system across terminal, desktop, IDE, web…
- AppFactor — Agentic platform that runs persistent agents across an enterprise…