Base44

From PRIMO.ai
Jump to navigation Jump to search

YouTube ... Quora ...Google search ...Google News ...Bing News


Base44 is an AI app builder, owned by Wix since June 2025, that turns a plain-language description into a working web application with a front end, back end, database, user authentication, and hosting already wired together. A user types what the app should do, and the AI chat generates the screens, data tables, permissions, and logic, then publishes the result to a live address that can take a custom domain. The pitch is that the services a typical app needs, from email and file storage to analytics and payments, come built in rather than as separate vendors to sign up for and connect. By late 2026 Base44 had grown into a family of products: the app builder, Superagents for personal and business AI agents, a standalone Base44 Backend for developers, and Base Code for teams working on existing codebases.

Base44 began as a side project of Maor Shlomo, an Israeli entrepreneur who had earlier co-founded the data company Explorium. He built it largely alone, with AI writing much of the code, and the product reached 10,000 users within three weeks of launch and about 250,000 users within six months, while earning a profit of $189,000 in May 2025 after paying for model tokens. On June 18, 2025, Wix bought the bootstrapped company for about $80 million in cash plus earn-out payments tied to performance through 2029, with $25 million of the upfront sum set aside as retention bonuses for the eight employees who had joined shortly before the deal. Base44 kept its own brand and product and continues to operate as a standalone business inside Wix.

Growth accelerated after the acquisition. Wix reported that Base44 had served more than 2 million users by November 2025, and the business crossed $100 million in annual recurring revenue (ARR) around March 2026, reached about $150 million by May, and passed $200 million in late August 2026, according to Shlomo. Along the way it aired its first Super Bowl commercial in February 2026, launched Superagents in March, put a Base44 app inside ChatGPT, and in June released Base 1, a model of its own fine-tuned for building web apps. Wix has made Base44 central to its investor story, saying the shift to Base 1 should lift Base44's gross margin from near zero at the start of 2026 to about 60% in the second half.

The same convenience carries trade-offs. Apps run on Base44's managed backend and JavaScript SDK, so moving an app elsewhere means more than exporting code, and outside researchers found serious flaws in the platform during 2025, including an authentication bypass reported by Wiz that Wix fixed within a day. Reviewers generally describe Base44 as fastest for prototypes, internal tools, and small business software, and less suited to complex systems that a team expects to scale or migrate. Base44 has responded with security scanning, AI penetration testing through Tenzai, enterprise controls, and Base Code for teams that want engineers to review every change.


Base44 treats software as something you describe rather than assemble: by owning the backend, the hosting, the agents, and now the model, it makes a working app the default output of a conversation, at the cost of tying that app to Base44's platform.


Mechanisms

Every Base44 app is built from a conversation in the AI chat, which edits a real codebase behind the scenes. The front end is a React and Vite application that paid users can open and edit in a Code tab, and the back end runs on Base44's own platform, which provides a flexible NoSQL data layer, built-in authentication and access control, realtime subscriptions, and TypeScript backend functions. Integrations with AI models, email, file handling, and outside services run on Base44's servers rather than in the browser, so secrets and tokens never reach the front end, and calls can run either as the signed-in user or with elevated service permissions for admin tasks and automations. Usage is metered in two currencies: message credits for AI chat turns that build or change the app, and integration credits for what the finished app does while it runs.

The chat has distinct modes that control what the AI is allowed to do. Plan mode, offered with the first prompt, holds a credit-free conversation that shapes a plan before anything is built, while Discuss mode lets a builder ask questions or plan the next change without touching code, at a flat 0.3 credits per message on a dedicated model. Build mode acts on prompts immediately, and Edit mode lets the builder click elements in the preview to change text, colors, and layout, either by hand at no credit cost or by asking the AI. Builders on the Builder plan and above can pick which model writes their code and, since October 2026, set an effort level from low to maximum for models that support it.

Model routing is where Base 1 comes in. Announced on June 29, 2026, Base 1 is a fine-tune of an undisclosed open-source base model, trained with reinforcement learning on what Base44 describes as tens of millions of real app-building interactions, including what users asked for, what the system produced, what broke, and what users kept or rejected. Shlomo told The New Stack that the model was not trained from scratch, and Base44's automatic routing sends tasks to Base 1 when it beats alternatives on internal benchmarks, while users can still choose frontier models from Anthropic, OpenAI, and Google. Owning the model gives Base44 more control over latency, cost, reliability, and design style, which matters because Wix projected total AI costs at 30% to 40% of Base44 bookings.

Inside a finished app, AI shows up in three main forms. AI agents, switched on from the app's dashboard, are chat assistants that can create, read, update, and delete records, search the web, call backend functions, and follow guidelines and up to 10 attached skills, and they can be reached through WhatsApp, Telegram, and LINE as well as in the app. Workflows, which replaced the older automations for apps created from July 6, 2026, are multi-step background processes that the builder describes in plain language, with triggers from Gmail, Google Calendar, Google Drive, Outlook, Slack, data changes, schedules, and sign-ups. Built-in integrations let an app call large language models, generate images, video, and speech, send email, and process files without the builder managing API keys.

Superagents are a separate kind of project: personal or team agents that live outside any one app. A user describes the job, and Base44 assembles the workflows, connects the needed tools, and deploys an agent that keeps persistent memory, runs on schedules and triggers, and acts in the background inside sandboxed environments with permission controls. Owners talk to a Superagent in the browser or through WhatsApp, Telegram, Slack, iMessage, and other channels, and recent additions let it take a dedicated phone number, place outbound calls, and join Zoom or Google Meet meetings as a notetaker. A library of more than 130 prebuilt skills can be added to an agent with one click.

Base Code applies the same chat-driven approach to software that already exists. A team connects a GitHub repository, and Base Code's agent studies the codebase and configures cloud infrastructure until a live preview is running, after which product managers, designers, QA, and operations staff can propose changes in chat on separate branches. Each change is handed to engineers as a pull request, so developers keep control of review and release, and the agent follows repository instructions such as AGENTS.md files and SKILL.md skills. Base Code launched as a standalone product on September 28, 2026, and the Israeli business site Calcalist framed it as Base44's direct challenge to Claude Code and Cursor.

The Physics

Base44's speed comes from making decisions for the builder, and its limits come from the same place. A managed stack removes setup work, but it also fixes where data lives, how code is structured, and how much each change costs. The main constraints are:

  • Credits are the meter: Every build turn spends message credits, and every agent reply, email, or model call inside a running app spends integration credits, with costs varying by model and task, so repeated attempts to fix the same problem can drain a monthly allowance quickly. Discuss and Plan modes exist largely to reduce that waste.
  • The platform is the backend: Generated apps depend on Base44's data layer, authentication, and SDK, so GitHub sync on the Builder plan and above moves the code but leaves the app expecting Base44's backend services, which is the root of most lock-in concerns.
  • Generated code accumulates: Each prompt adds to a codebase the builder may never read, and reviewers report that the AI can get stuck repeating failed fixes as apps grow, which is why Base44 is most often recommended for prototypes, internal tools, and focused business apps.
  • Security defaults have edges: Base44's own documentation notes that authentication tokens are stored in the browser's localStorage because HttpOnly cookies are not available, and that per-app CORS and most HTTP headers, such as Content-Security-Policy, cannot be configured individually.
  • Model choice changes the output: When the security firm Tenzai built 15 apps from the same prompts across five different models on Base44 and attacked them, it found 254 vulnerabilities in total, with the type and severity shifting significantly by model.
  • Mobile is a wrapper: Apps published to the Apple App Store and Google Play run the published web app inside a native WebView, so native-only features such as push notifications and a full offline mode are not supported.
  • Inference is the biggest cost: Base44 entered 2026 with a near-zero non-GAAP gross margin, and Wix tied the expected improvement to moving users onto Base 1, which shows how much of the business model rides on the price of model tokens.

These constraints shape how experienced builders use the platform. They plan before building, keep apps focused, review data permissions and run the security scan before publishing, and export or document anything they may need to move later. Teams that need their code in their own repositories can use Base Code or the standalone backend instead, borrowing Base44's infrastructure without giving up the repository.

Why This Matters

Base44 is one of the clearest tests of whether AI app building can become a large, profitable business rather than a demo. Its numbers are unusually public because Wix reports them to investors. Its significance shows up in several ways:

  • Speed of adoption: A bootstrapped project with no outside funding sold for $80 million in cash six months after launch and passed $200 million in ARR about 14 months after the acquisition, a pace Wix management called one of the fastest in software history.
  • A turnaround story for Wix: Investors had sold Wix down on fears that AI tools would replace website builders, and Base44 became its counterargument: the stock fell about 27% after first-quarter 2026 results showed heavy spending, then rose more than 60% in August as Base44 passed $200 million in ARR.
  • Vertical integration: Wix described Base 1 as the first proprietary model launched by an app-creation platform, a sign that AI builders may compete on owning their models and infrastructure rather than reselling frontier models.
  • Software by non-engineers: Superagents and Base Code extend the idea from new apps to everyday automation and to changes in production codebases, letting people outside engineering do work that used to wait in a developer backlog.
  • Assistants as distribution: The Base44 app in ChatGPT, built with OpenAI's Apps SDK and Base44's Model Context Protocol (MCP) server, and direct MCP connections from Claude, ChatGPT, and Cursor let people start building from inside other assistants.
  • Security of citizen-built apps: The Wiz and Imperva findings, and Tenzai's research on model-dependent vulnerabilities, made Base44 a reference case for cybersecurity and AI governance teams deciding how far to trust apps built by people who never read the code.

Base44 is usually compared with Lovable, Bolt.new, and Replit, which also generate full applications from prompts, and with in-repository coding agents such as Claude Code, Cursor, and Google Antigravity. Its distinguishing choices are the fully managed backend, Wix's distribution and payment infrastructure, and the breadth of adjacent products, while rivals that sync code to the builder's own repository offer more portability. For the broader landscape, see the AI app builder section of the Development page and the AI Coding Agent Comparison.

AI Offerings

Base44 sells one subscription but ships many distinct AI products and features. The sections below cover each offering as documented in Base44's help center, changelog, and announcements through October 2026. Plan requirements change often, so check the current pricing page before relying on any one of them.

AI App Builder

The core product generates complete web apps and websites from prompts, templates, or reference images. Builders can attach files, including PowerPoint decks since October 2026, answer the AI's clarifying questions before the first build, and draw on a component library and shared design systems. Published apps get hosting, analytics, custom domains on paid plans, and version history with branches, and collaborators can work in the same app with a shared chat queue. A desktop app for Mac and mobile apps for iOS and Android let builders create and edit away from the browser.

Model Picker and Base 1

Base 1 is Base44's in-house model and has been available in the AI chat since June 29, 2026, with automatic routing sending tasks to it when it outperforms alternatives. Builder-plan users and above can select a model manually, compare two models on the same message at double the credit cost, and set an effort level. Recent changelog entries added Claude Fable 5.1, Sonnet 5.5, and Haiku 5.5, GPT 6 Astra, GPT-6.1 Terra, and Gemini 4 Argon to the AI chat, while in-app agents and the InvokeLLM integration gained Claude Opus 5.5 and GPT 6 Sol and Luna. Enterprise administrators can restrict which models members may use.

AI Agents for Apps

App agents turn a chat window inside a Base44 app into an assistant that takes action on the app's own data. Builders enable agents from the dashboard, start from a suggested template or describe one in the AI chat, and configure guidelines, the model, memory, tools, and skills. Each agent can be connected to WhatsApp, where it gets its own number, as well as to Telegram and LINE, and public apps ask visitors to log in first unless the builder turns on anonymous access, so the agent knows whose data it is handling. Conversations are billed in integration credits, starting at roughly 3 credits per message on the default model.

Workflows

Workflows is Base44's native automation engine, available on the Builder plan and above. The builder describes a trigger, steps, conditions, and delays in plain language, and the AI assembles the flow, writes any backend function a step needs, and runs it in the background with no separate infrastructure. Because Workflows lives inside the app, it can see the app's data, users, and agents and suggest automations tailored to them, which Base44 contrasts with bolting on Zapier, n8n, or Make. Workflows replaced the earlier single-step automations for apps created from July 6, 2026, and older apps can switch over in one step.

Built-in AI Integrations

Every app can call a set of managed AI services without its builder signing up with model providers. The InvokeLLM integration sends prompts to language models, and media integrations generate images with models such as Nano Banana 2.1, FLUX 3, Seedream 5.0, and Meta's Muse Image, video with Wan 3.0 and Seedance 2.5 at up to 1080p, and speech. Other integrations send email, upload and process files, and call custom APIs, with credentials held on Base44's servers. Usage draws on the app owner's integration credits, and connectors such as Gmail, Slack, Notion, HubSpot, Salesforce, and Stripe extend apps to outside services.

Superagents

Launched on March 11, 2026, and available to all Base44 users, Superagents are autonomous agents that a user creates by describing the work in natural language. Base44 builds the workflows, connects tools such as Gmail, Google Calendar, Google Drive, Slack, GitHub, and CRMs, and runs the agent around the clock with memory, files, skills, scheduled tasks, and goals that persist across runs. A Superagent can also operate across its owner's Base44 apps, tracking analytics and users, sending notifications, and managing follow-ups, and developers can reach it through an API and webhooks. Enterprise administrators can disable Superagents, set channel policies for WhatsApp, Telegram, LINE, iMessage, Slack, and email, and govern template links.

Base Code

Base Code is a standalone product for teams with existing software rather than new projects. It imports GitHub repositories, gives each project in a multi-project repository its own preview, starts branches from a chat request or a Google Drive file, and lets the AI read pull request checks, help with merge conflicts, reply on review threads, and resolve fixed comments. Workspace owners can restrict imports to approved GitHub organizations, and secrets are encrypted, shared across branches, and never committed to the repository. Base Code is available to all Base44 users, and the first Base Code turn on an imported app is free.

Base44 Backend Platform, CLI, and Skills

Released on February 16, 2026, the Base44 Backend Platform offers the same backend that powers the app editor as a standalone backend-as-a-service for developers who want their own front end. Projects are created and deployed with the Base44 CLI and accessed through the JavaScript SDK, and the service covers data, authentication, file storage, realtime updates, TypeScript functions, and hosting with custom domains and automatic HTTPS. Base44 publishes agent skills for Claude Code, Cursor, Codex, and other coding agents that teach those agents to set up projects, define data entities, write code, and deploy, and its documentation lists React Native mobile apps among supported uses. Base44's documentation still labels the backend service and CLI as beta.

MCP Server and MCP Connections

Model Context Protocol (MCP) support runs in both directions. The Base44 MCP server lets an MCP-compatible assistant such as Claude, ChatGPT, or Cursor create apps, edit existing ones, and list a user's projects after an OAuth sign-in, with builds running in the background and returning a link to the editor, and a separate documentation server lets assistants search Base44's docs. In the other direction, each account can connect up to 20 outside MCP servers to the Base44 AI chat, which calls a server only when a prompt needs it. Base44 says its own server supports direct connections from Claude, ChatGPT, and Cursor but not third-party MCP gateways.

Base44 App in ChatGPT

On March 23, 2026, Base44 launched an app inside ChatGPT, built with OpenAI's Apps SDK and powered by Base44's MCP server. Users type @Base44 in a ChatGPT conversation, describe the software they need, and get a working app with its database, authentication, logic, and deployment handled automatically. Apps created this way run on Base44's full stack and can be customized further in the Base44 editor, and the app is offered wherever ChatGPT's app directory is supported.

Mobile Apps

Since February 2, 2026, builders can choose Web or Mobile when publishing and prepare an app for the Apple App Store and Google Play from the editor's Mobile app tab. The tab scans the app against each store's guidelines and generates an iOS IPA bundle or an Android App Bundle, with file downloads requiring the Builder plan or above and the builder's own Apple and Google developer accounts. The result is a lightweight native wrapper around the published web app, so most content and design changes appear without a new store submission, but push notifications and full offline mode are not supported.

Security Scanning and Tenzai Penetration Testing

Base44's security tooling has become an offering in its own right. The built-in security scan checks for overly open data, exposed credentials, login gaps, third-party dependencies, and backend functions, flags private fields on readable records, and can apply fixes, which since October 2026 first check whether a finding is actually exploitable. Builder-plan workspaces add static code analysis and can connect their own Wiz tenant, Enterprise workspaces can schedule scans, and since October 27, 2026, owners and admins can launch an agentic penetration test from Tenzai that attacks the live app with a temporary test login. Tenzai pentests are billed to the customer's Tenzai account rather than to Base44 credits, and the full report lands in the Tenzai dashboard.

Marketing, Payments, and Business Tools

Base44 increasingly bundles the tools a small business needs after launch. Builders can design branded app emails and track their performance, run Google Ads campaigns from a guided wizard, generate social content and post it to LinkedIn, Instagram, and Facebook, and see how AI assistants surface their app in an AI Visibility tab under SEO and GEO on the Pro plan and above. Base44 Payments and Stripe handle checkout, subscriptions, refunds, and payouts, partners can request card payments through Wix Payments, and a presentation builder exports decks to PowerPoint, Google Slides, and PDF. Several of these features draw on Wix's existing commerce and marketing infrastructure.

Enterprise Controls

Business and Enterprise plans add governance for organizations. Workspace single sign-on works with Google Workspace, Microsoft Entra ID, Okta, GitHub, and other OpenID Connect providers, and administrators can enforce SSO across every app, choose US, EU, or UK data residency, set IP allowlists, use a dedicated encryption key, and control which connectors and AI models members can use. The Enterprise tier adds automatic user provisioning through SCIM and audit logs that can be pulled through a Monitoring API, and Base44 says data in Enterprise workspaces is not used to train AI models. For workspaces created on or after October 1, 2026, SSO and data residency require the Business or Enterprise plan.

Offerings at a Glance

The table below summarizes Base44's AI offerings, what each does, and where it is available. Availability reflects Base44's documentation and announcements through October 2026. Credit costs vary by model, action, and plan.

Offering What It Does Availability
AI App Builder Generates full-stack web apps and websites from prompts, with hosting and analytics All plans; the Free plan allows 25 message credits a month and up to five apps
Chat modes Plan, Discuss, Build, and Edit modes control how the AI changes an app All plans; model choice on Builder and above
Base 1 Base44's in-house model, fine-tuned for building web apps In the AI chat since June 29, 2026
AI agents for apps In-app assistants that act on records and reach users on WhatsApp, Telegram, and LINE Billed in integration credits
Workflows Multi-step automations described in plain language Builder plan and above
Built-in AI integrations Language model, image, video, and speech generation inside apps Billed in integration credits
Superagents Autonomous personal and team agents with memory, connectors, skills, and messaging channels All users since March 11, 2026; some features on higher plans
Base Code Chat-driven changes to existing GitHub codebases, delivered as pull requests All users since September 28, 2026
Base44 Backend Platform Standalone backend-as-a-service with CLI, SDK, and agent skills Beta since February 16, 2026
MCP server and connections Builds Base44 apps from other assistants and connects outside Model Context Protocol (MCP) servers to the AI chat OAuth sign-in; up to 20 custom servers per account
Base44 app in ChatGPT Builds apps inside ChatGPT through @Base44 Since March 23, 2026, where ChatGPT's app directory is supported
Mobile publishing Prepares App Store and Google Play bundles from a published app File downloads on Builder and above
Security scan and Tenzai Scans apps for weaknesses and runs AI penetration tests on live apps Static analysis and Tenzai on Builder and above
Enterprise controls SSO, data residency, SCIM, audit logs, and model and connector governance Business and Enterprise plans

Plans and Pricing

Base44 uses a freemium credit model in which each plan buys a monthly pool of message credits and integration credits. The prices below are the annual-billing rates shown on Base44's pricing page in October 2026, which the page says save 20% compared with monthly billing. Credit consumption varies by action and selected model, and the Business tier's price is not published on the pricing page.

Plan Price (Billed Annually) Message Credits per Month Integration Credits per Month Notable Inclusions
Free $0 25 100 Core features, up to five apps, authentication, database, and analytics
Starter $16 per month 100 2,000 Unlimited apps, in-app code edits, custom domain, and a free domain for one year
Builder $40 per month 250 10,000 AI model selection, GitHub integration, Workflows, static analysis, and Tenzai testing
Pro $80 per month 500 20,000 Early access to beta features and the AI Visibility tab
Elite $160 per month 1,200 50,000 Premium support
Business Not published Varies Varies SSO, data residency, IP allowlist, and publishing and connector controls
Enterprise Custom quote Custom Custom Dedicated architect, SCIM, audit logs, and an account team

Security

Base44's security record mirrors the wider vibe-coding industry: fast growth first, hardening after outside researchers found problems. Imperva researcher Yohann Sillam reported several flaws to Base44 in early 2025 and disclosed them publicly in August 2025, including an open redirect that leaked access tokens, a stored cross-site scripting bug that ran attacker code inside the trusted app.base44.com domain, and a design in which the main platform's login token was passed to user-built apps through the URL. Imperva said that combining the scripting bug with weak, client-side enforcement of premium features gave attackers a clear path to full account takeover, and fixes and architectural changes were rolled out over the following months.

The best-known flaw came from Wiz in July 2025. Its researchers found that two undocumented endpoints for registering users and verifying one-time passwords required no authentication, so anyone holding an app's non-secret app_id, which appears in the app's URL, could create a verified account on a private app and bypass every login control, including single sign-on. Wiz confirmed the bypass on enterprise apps used for internal chatbots, knowledge bases, and HR operations containing personal data. Wix fixed the issue within 24 hours of the July 9 disclosure and found no evidence that it had been exploited.

Base44 now describes a much larger security program. The platform says it is SOC 2 Type II and ISO 27001 certified and GDPR compliant, encrypts data at rest and in transit with credentials protected by AWS Key Management Service, runs regular internal and third-party penetration tests and a bug bounty program, and uses PCI DSS-certified payment processing. Recent changes encrypt secrets at rest for Education and Enterprise workspaces, convert API keys pasted into chat into named app secrets, and sign users out on every device after a password change. Base44 still states that each builder is responsible for an app's own security settings.

The remaining risk sits mostly in what builders create. Data access rules decide who can read or change each record, and a permissive rule can expose every user's data even when the platform itself is sound, which is why Base44 suggests rules as apps are built and lets builders test table permissions in preview as another user. Tenzai's research on Base44, in which apps generated from identical prompts by different models carried different vulnerabilities, suggests that model choice is itself a security decision. Treat any app that holds personal, financial, or health data as production software that needs review, scanning, and testing before launch, and watch for prompt injection in apps whose agents read untrusted text.

Getting Started

Most people start Base44 in the browser with a free account and a single idea. The steps below follow Base44's documentation and move from planning through publishing, with notes on where paid plans begin. Keep the first app small and use realistic but non-sensitive sample data until permissions are reviewed.

  1. Create an account: Sign up at base44.com. The Free plan includes 25 message credits a month and up to five apps, which is enough to learn the workflow.
  2. Plan before building: Turn on Plan under the first prompt to shape the app in a credit-free conversation, then click Start building when the plan looks right.
  3. Describe the app clearly: Name the users, the data each screen needs, the main workflow, and any roles, or start from a template and adapt it.
  4. Iterate in the right mode: Use Build for changes, Discuss to ask questions or weigh options at 0.3 credits per message, and Edit to click and adjust visual elements by hand without spending credits.
  5. Review data and permissions: Check the generated data tables, set data access rules so users see only their own records, and test permissions in preview as another user.
  6. Add AI features: Enable agents from the dashboard, describe Workflows in plain language on the Builder plan, and connect services such as Gmail, Slack, or Stripe through connectors.
  7. Scan before publishing: Run the security scan, fix what it finds, and on the Builder plan or above consider static analysis or a Tenzai penetration test for apps that handle sensitive data.
  8. Publish: Choose Web or Mobile, connect a custom domain on a paid plan, and on the Builder plan or above generate App Store and Google Play files from the Mobile app tab.
  9. Choose a model deliberately: On the Builder plan or above, compare models on the same request and set an effort level, remembering that stronger models and higher effort use more credits.
  10. Work as a developer: Install Base44's agent skills or run the Base44 CLI to create a backend project, connect a coding agent to the Base44 Model Context Protocol (MCP) server, or sync an app to GitHub on the Builder plan.
  11. Bring an existing codebase: Use Base Code to connect a GitHub repository, let the agent prepare a live preview, and route every change to engineers as a pull request.
  12. Set up a Superagent: Describe a recurring job, connect the tools it needs, and talk to it in the browser or on WhatsApp, Telegram, or Slack.

A few habits save credits and trouble. Write one clear request at a time instead of stacking several changes into one prompt, use Discuss when a problem is unclear, and roll back to an earlier version rather than asking the AI to fix the same bug again and again. Export your data regularly, keep a record of the prompts and decisions behind the app, and review permissions again after every major change.

Examples

Base44's uses range from weekend side projects to company internal tools. The first example walks through a typical build in general terms, while the others draw on Base44 announcements, documentation, and press coverage rather than invented customer stories. The sample prompts at the end illustrate how requests map to Base44's offerings.

Example: An Internal Request Tracker

Consider an operations manager who tracks equipment requests in a shared spreadsheet and a group chat. She describes the app to Base44, listing requesters, items, status, and due dates, answers the AI's clarifying questions, and receives a working app with a request form, a status board, user logins, and a data table for each kind of record. She then sets data rules so that requesters see only their own requests while managers see everything.

Next she adds a Workflow that emails a requester whenever a status changes, and an app agent that answers questions about a request in the app or on WhatsApp. Before sharing the app, she runs the security scan and tests the permissions while viewing the app as a requester. The build can be finished in an afternoon, though the credits it uses depend on the model and on how many revisions she asks for.

Superagents for Everyday Work

Base44's own examples for Superagents center on recurring tasks that span several tools. Its documentation suggests starting with jobs such as monitoring email, generating reports, or tracking updates across tools, and its marketing shows an agent that watches flight prices between London and New York, finds family events for a trip, and sends updates to WhatsApp. Because a Superagent can reach its owner's Base44 apps, it can also track app analytics and users, send notifications, and manage follow-ups across an app's database. Newer features extend the same agent to phone calls and to meetings, where it can join Zoom or Google Meet as a notetaker.

Base Code for Product Teams

Base Code targets the common situation in which a small change waits weeks in an engineering backlog. In Base44's launch materials, a product manager, designer, or QA analyst connects to a repository the company already runs in production, asks the agent for a change, checks it in a live cloud preview, and submits a pull request for engineers to review. Base44's chief product officer, Yoav Orlev, described the goal as letting non-engineers make changes to existing codebases directly while engineers keep leading the review process. Enterprise controls limit imports to approved GitHub organizations so that the workflow fits existing governance.

Developers and Coding Agents

Developers use Base44 differently from non-technical builders. With the Backend Platform, a coding agent such as Claude Code or Cursor installs Base44's skills, creates a project with the CLI, defines data entities, writes functions, and deploys, while the developer keeps a custom front end in any framework. Base44's developer site features examples such as a Chrome extension, a desktop whiteboard tool with AI chat and real-time collaboration, and a company that used the backend to get a complex front end running in two days. Wix also connected its headless commerce platform to Claude Code, Codex, and Base44 in 2026, so AI-generated front ends can use Wix's business infrastructure.

Building From Inside ChatGPT

The ChatGPT app shows how Base44 reaches people who never visit its site. A user who is already discussing a problem in ChatGPT can type @Base44, describe a CRM, reporting dashboard, or internal tool, and receive a working app built on Base44's stack. The app can then be opened in the Base44 editor, where the full set of design, agent, and workflow features is available.

Sample Prompts

The table below pairs illustrative prompts with the Base44 offering each one uses and a practical note. The wording matters less than naming the users, the data, the action, and any limits. Plan requirements reflect Base44's documentation in October 2026.

Prompt Offering Used Practical Note
"Build a booking app for my dog-grooming business with a calendar, deposits, and reminder emails." AI App Builder, Base44 Payments, Workflows Workflows need the Builder plan; try payments in test mode first
"Before building, help me plan a volunteer scheduling app and ask me what you need to know." Plan mode Planning does not use credits
"Add an agent that answers customers' order questions on WhatsApp using their order history." AI agents for apps Billed in integration credits; public apps ask users to log in
"Every Monday, email each manager a summary of open requests and flag anything overdue." Workflows Scheduled trigger; Builder plan and above
"Check my inbox each morning and send me a WhatsApp digest of anything from our top customers." Superagents Connect Gmail and WhatsApp, then review what the agent may access
"Change the checkout button text and open a pull request." Base Code Engineers review and merge
"Prepare this app for the App Store." Mobile publishing Requires an Apple developer account and the Builder plan for files
"Scan this app and fix any data that strangers can read." Security scan Review each fix before applying it

Known Limitations

  • Credit-driven costs: Heavy iteration, stronger models, higher effort levels, and busy in-app agents all consume credits, which makes monthly costs harder to predict than a flat subscription.
  • Platform dependence: Apps rely on Base44's hosted backend and SDK, so leaving the platform means rebuilding data, authentication, and functions elsewhere even after syncing code to GitHub.
  • Complexity ceiling: Large apps with intricate business logic are harder to steer by prompt, and reviewers report that the AI sometimes loops on the same failed fix.
  • Builder-owned security: Base44 secures the platform, but each app's data rules, exposed fields, and agent permissions remain the builder's responsibility.
  • Fixed security headers: Authentication tokens live in localStorage, and per-app CORS and Content-Security-Policy settings cannot be customized.
  • Mobile limits: Store apps are web views without push notifications or full offline mode.
  • Plan gating: Model choice, Workflows, GitHub sync, Tenzai testing, and mobile file downloads require the Builder plan or above, and SSO and data residency require Business or Enterprise for workspaces created since October 1, 2026.
  • Training data: Base 1 is trained on real build sessions, and Base44 says whether workspace data can be used for training depends on the plan, with Enterprise workspaces excluded automatically, so other customers should check the terms for their plan.
  • Rapid change: Base44's changelog records dozens of changes a month, so documentation, tutorials, and pricing can go out of date quickly.

History

Recent Developments

  • Late 2024 to early 2025: Maor Shlomo built Base44 as a side project after returning from military reserve duty, and it reached 10,000 users within three weeks of launch.
  • May to June 2025: Base44 reported a $189,000 profit for May after model costs and reached about 250,000 users.
  • June 18, 2025: Wix acquired Base44 for about $80 million in cash plus earn-out payments through 2029.
  • July 2025: Wiz reported an authentication bypass on July 9, Wix fixed it within 24 hours, and the research was published on July 29.
  • August 2025: Imperva disclosed earlier-reported flaws, including stored cross-site scripting and leakage of platform login tokens to user-built apps.
  • November 19, 2025: Wix said Base44 had served more than 2 million users and was on track for at least $50 million in ARR by year-end.
  • February 2026: Mobile publishing to the App Store and Google Play went live on February 2, the "It's App to You" ad aired during Super Bowl LX on February 8 alongside a $50,000 app contest, and the Base44 Backend Platform launched on February 16.
  • March 2026: Base44 passed $100 million in ARR, launched Superagents on March 11, and launched the Base44 app in ChatGPT on March 23.
  • April 16, 2026: EU and UK data residency became available for new apps on eligible plans.
  • May 2026: Wix reported Base44 at about $150 million in ARR, while Wix shares fell about 27% as heavy marketing and computing costs returned the company to losses.
  • June 29, 2026: Base44 released Base 1, its in-house model for building web apps.
  • July 6, 2026: Workflows replaced single-step automations for new apps.
  • August 2026: Wix's second-quarter report projected about 60% gross margin for Base44 in the second half, Superagents gained the ability to join Zoom and Google Meet meetings, and Base44 passed $200 million in ARR late in the month.
  • September 2026: Base44 added a presentation builder, new Claude and GPT models, and dedicated phone numbers for Superagents, and launched Base Code on September 28.
  • October 2026: SSO and data residency moved to the Business and Enterprise plans for new workspaces, a Mac desktop app arrived, Tenzai penetration testing launched on October 27, and GPT-6.1 Terra joined the model picker on October 28.

Further Resources

Additional Viewing

Load video
YouTube

Solo founder, $80M exit, 6 months: The Base44 bootstrapped startup success story is Lenny Rachitsky's roughly 90-minute Lenny's Podcast interview with Maor Shlomo, recorded shortly after the Wix acquisition. Shlomo explains how he built Base44 largely alone with AI writing much of the code, why he chose to sell even though the company was profitable, and how building in public on LinkedIn drove growth.

Load video
YouTube

Base44 Superagents - Your Personal AI Agent is Base44's own introduction to Superagents, the agent product launched in March 2026. It shows an agent connecting to a user's tools, automating workflows, managing an inbox, and analyzing data from plain-language instructions.

Load video
YouTube

Base44 Tutorial for Beginners 2026 - How to Use Base44 from Metics Media is a step-by-step introduction for first-time users. It is a good starting point for the app-building workflow described in Getting Started.

Load video
YouTube

Base44 Tutorial: How to Build App with AI + Publish to App Store (No Code) from Mikey No Code walks through building an app with Base44's AI and publishing it to Apple's App Store without writing code. It is a practical companion to the Mobile Apps section above.

Load video
YouTube

I Built the SAME App in Claude Code vs Google Antigravity vs Base44 from Mikey No Code builds one app in three different AI tools to compare the experience. It illustrates the trade-off between Base44's managed, prompt-first approach and the in-repository coding agents discussed in Why This Matters.

References

Retrieved from "https://primo.ai/w/index.php?title=Base44&oldid=52603"