Base44
AI app builder (a Wix company) with built in database, auth and hosting, plus Superagents: autonomous agents built from a plain language description that run on schedules and events with persistent memory.
Base44 is an all-in-one app builder (a Wix company, acquired June 2025) with built-in database, auth, hosting, and agents for full-stack vibe coding. It reportedly reached $100M ARR, with a free tier and paid plans roughly $40–$100/month.
Vendor details
Canonical URL
https://base44.com
Category
Agent builder
Subcategory
App builder (vibe coding)
Use cases & customers
Primary use cases
Target customers
Deployment options
In practice
You want to build a working app but don't want to wire up a database, auth, and hosting separately. Base44 includes all of them, so you describe the app and it comes together.
Vibe coding falls apart when the backend isn't there. Base44 is full-stack, handling data, authentication, and hosting behind the app you build.
Standing up a prototype usually means stitching services together first. Base44 is an all-in-one builder, so the foundation is built in and you start on the actual app.
Agentic Index coverage score
10.5 / 14 capabilities · 75%
| Integrations & Tool CallingBreadth across classes plus a community contribution path. Prebuilt integrations cover AI, email, file storage, payments and messaging, with Airtable, Slack, Stripe and Zapier named in the documentation index. Beyond the first party set there is a COMMUNITY INTEGRATIONS LIST with a Create Integration flow and per integration approval status, so third parties publish connectors others can adopt (Explorium documents its own onboarding through it). Enterprise admins control which external services are available, manage shared versus app user credentials separately, and can block specific connectors across apps, Superagents and in app agents. Superagents themselves connect outward to third party applications and reach users over WhatsApp and Telegram. The Wix distribution adds Stores, Bookings, Blog, Events, Restaurants and CRM as context sources. docs.base44.com developer documentation, base44.com enterprise pages and Wix App Market listing 2026-07-30 | Full |
|---|---|
| Workflow OrchestrationUPGRADED FROM THE UNSOURCED GRID, which carried Partial, and the reason is a product that did not exist when the June import was written. SUPERAGENTS LAUNCHED MARCH 2026 and are available to all users. The orchestration is two layered. At build time the user describes the task in natural language and BASE44 AUTOMATICALLY BUILDS THE UNDERLYING WORKFLOWS, connects the necessary tools and deploys the agent. At run time the deployed agent OPERATES CONTINUOUSLY IN THE BACKGROUND, responding to triggers, schedules and real time events, sending updates, monitoring systems, executing recurring tasks and triggering workflows WITHOUT MANUAL INTERVENTION, in the vendor's own words. Users run multiple Superagents against different systems simultaneously. That is unattended multi step execution plus automated workflow composition, which is the top of this axis. base44.com Superagents announcement March 2026 and docs.base44.com 2026-07-30 | Full |
| Knowledge Grounding & RAGHELD AT PARTIAL, and this is the weakest axis in an otherwise strong grid. Agents are grounded in the application's OWN DATA ENTITIES, with the documentation describing AI assistants defined with custom tools AND DATA PERMISSIONS, and each entity supporting granular create, read, update and delete permissions. On the Wix surface, agents answer using the merchant's live site data. Superagents additionally reach connected business systems. THAT IS SCOPED DATA ACCESS, NOT A KNOWLEDGE LAYER. No document ingestion, no embedding or vector store, no retrieval augmented generation surface, no knowledge base object and no citation or grounding attribution was found across three passes covering the developer documentation, the enterprise pages and the Superagents launch material. Compare the Full grades in this pass, which rest on named grounding artefacts such as tofu's Playbook or Shopify's federated cross app retrieval. docs.base44.com developer documentation and Wix App Market listing 2026-07-30 | Partial |
| Human Oversight & GuardrailsUPGRADED FROM THE UNSOURCED GRID, which carried Partial. Oversight is PRE EXECUTION BOUNDING AT THREE INDEPENDENT LEVELS, which is the leena-ai shaped posture rather than an approval queue. AT THE AGENT LEVEL: users define what each agent can access and do, agents run in SANDBOXED ENVIRONMENTS on Base44 infrastructure with clear permission controls, and each data entity carries granular create, read, update and delete permissions. AT THE CONNECTOR LEVEL: admins control which external services are available and can BLOCK SPECIFIC CONNECTORS across apps, Superagents and in app agents, so a capability can be withdrawn organisation wide without touching individual agents. AT THE WORKSPACE LEVEL: an organisation that has not approved AI agents can DISABLE SUPERAGENTS ACROSS THE ENTIRE WORKSPACE, the same kill switch shape recorded on demandbase in this pass. Add role based publish and visibility controls and per member credit caps. WHAT IS ABSENT is a run time approval gate on individual agent actions, but bounding what an agent CAN do before it runs is a legitimate strong posture under the standing precedent. docs.base44.com enterprise documentation and Superagents announcement 2026-07-30 | Full |
| Security, Identity & GovernanceUPGRADED FROM THE UNSOURCED GRID, which carried Partial, and this is a top tier posture stated unambiguously on a first party trust centre at base44.com/security. CERTIFICATIONS NAMED WITH TYPE: SOC 2 TYPE II AND ISO 27001, plus GDPR. CONTROLS: TLS 1.2+ in transit, AES-256 at rest, secrets through cloud KMS, OWASP based penetration testing, threat modelling, secure design and code review. TWO ITEMS THAT ARE NEW BENCHMARKS FOR THIS INDEX: an ACTIVE BUG BOUNTY PROGRAMME, and a THIRD PARTY VENDOR RISK PROGRAMME COVERING EVERY SUPPLIER, neither of which appears on any other vendor graded in this pass. IDENTITY: SSO through the organisation's own IdP enforced across the workspace and every app inside it, SCIM provisioning with automatic joiner and leaver sync, OAuth 2.0 connectors, IP allowlisting by address, range and subnet, and domain based access control. DATA: enterprise workspaces are OPTED OUT OF MODEL TRAINING BY DEFAULT, which for a platform that ingests the customer's prompts and application code is the commitment that matters most. Two way GitHub export lets a customer run their own security and compliance scanners over the generated code. Runs on Wix's enterprise security infrastructure. base44.com/security trust centre and base44.com/enterprise 2026-07-30 | Full |
| Observability & AuditabilityUPGRADED FROM UNKNOWN, which is what the unsourced grid carried, so this cell was genuinely unrated rather than merely wrong. A PROGRAMMATIC AUDIT SURFACE IS DOCUMENTED FIRST PARTY: enterprise admins create and manage WORKSPACE OWNED API KEYS FOR PROGRAMMATIC ACCESS TO AUDIT LOGS AND WORKSPACE DATA VIA THE MONITORING API, and the enterprise page states plainly that IT MONITORS EVERY ACTION THROUGH THE API. That is a buyer accessible, machine readable audit trail rather than a dashboard, which is the stronger form and lets a customer pipe activity into their own SIEM. GRADED MEDIUM CONFIDENCE BECAUSE OF ONE REAL GAP: the documentation does not establish whether the audit trail resolves to AGENT LEVEL DECISIONS, distinguishing what a Superagent did autonomously from what a user did, which is the distinction that separates hubspot's Audit Cards and manhattan's per decision tracing from ordinary platform logging. No retention policy for audit data was documented either. docs.base44.com enterprise documentation and base44.com/enterprise 2026-07-30 | Full |
| Memory & State PersistenceUPGRADED FROM THE UNSOURCED GRID, which carried Partial. THIS IS ONE OF THE CLEANEST Mem=Full CASES IN THE INDEX because the vendor states the capability directly rather than leaving it to be inferred: Superagents have PERSISTENT MEMORY ACROSS CONVERSATIONS AND TASKS, and REMEMBER PREFERENCES, WORKFLOWS AND PRIORITIES, becoming more effective over time. That is memory as an agent property spanning sessions and spanning task boundaries, which is exactly what this axis asks for and what almost nothing else in the GTM cluster has. Structurally it is the same claim that earned blue-yonder its Full, where the vendor distinguished governed digital workers with persistent memory from stateless assistants that forget context. Underneath it, the platform's built in database provides durable application state independently of agent memory. base44.com Superagents announcement March 2026 2026-07-30 | Full |
| Deployment & Data ResidencyHELD AT PARTIAL, but on real evidence rather than the unsourced guess. RESIDENCY IS GENUINELY OFFERED: enterprise customers CHOOSE WHERE WORKSPACE APP DATA IS STORED ACROSS US, EU OR UK, documented first party and framed explicitly as meeting data sovereignty and compliance requirements. That is region selection, the first of the four deployment shapes this pass has formalised, and it is more than most of the GTM cluster offers. DEPLOYMENT CONTROL IS ABSENT: no on premises, no air gapped option, no private VPC, no single tenant and no self hosting. THE PORTABILITY CEILING IS THE SHARPER LIMITATION and is worth stating for buyers: the two way GitHub integration EXPORTS FRONTEND CODE ONLY, while the backend, the database and the deployed agents remain on Base44, so an exit is a rewrite rather than a migration. Hosting runs on Wix infrastructure. Half the axis served, which is Partial. docs.base44.com enterprise documentation and third party platform analysis 2026-07-30 | Partial |
| Prebuilt Agents, Templates & PacksDOWNGRADED FROM THE UNSOURCED GRID, which carried Full on no evidence, and this is the one axis where the June import was too generous. THE PRODUCT'S PREMISE IS GENERATION, NOT TEMPLATES: a user describes what they want and the platform composes the application or agent from scratch, which is a different thing from instantiating a prebuilt one. What does exist and earns the Partial: contextual suggestion of starting points (on the Wix surface, detecting installed apps and proposing a sales dashboard, staff scheduler, content calendar or client history view), the Community Integrations catalogue as reusable building blocks, and documented skills for AI coding agents setting up projects. WHAT WAS NOT FOUND across three passes: a named prebuilt agent library, an agent template marketplace, importable plays, or a gallery a user can browse and clone. FLAGGED FOR RE VERIFICATION, since a template gallery is the kind of surface that is easy to miss and would move this to Full. docs.base44.com, base44.com and Wix App Market listing 2026-07-30 | Partial |
| Triggers & Channel CoverageUPGRADED FROM THE UNSOURCED GRID, which carried Partial. THREE DISTINCT TRIGGER CLASSES are documented for Superagents rather than one: event triggers, SCHEDULES, and real time events, with agents running continuously in the background between them. That is a genuine trigger framework, not a single signal type, and schedule based invocation in particular is absent from most of the GTM cluster. Channel coverage is broad and unusual for this category: WHATSAPP AND TELEGRAM are named first party as ongoing interaction surfaces for a deployed Superagent, alongside email and SMS (both billed through integration credits), plus the deployed web application itself and in app agents. GRADED MEDIUM CONFIDENCE because the trigger taxonomy comes from the launch announcement rather than from a configuration reference, so the granularity a builder actually gets is not documented. base44.com Superagents announcement and docs.base44.com 2026-07-30 | Full |
| Model Flexibility & RoutingGENUINE MULTI MODEL ARCHITECTURE, NO CUSTOMER CONTROL, which is why this sits at Partial rather than either extreme. Third party platform analysis reports that the builder AUTO SELECTS BETWEEN CLAUDE SONNET 4.6 AND GEMINI 2.5 PRO depending on the task, so this is not the single undisclosed provider dependency found on the four vendors graded immediately before it in this pass; two named frontier models are in play and routing exists. BUT THE ROUTING IS THE VENDOR'S, NOT THE BUYER'S: no model selection, no routing configuration, no bring your own model and no bring your own key surface was found first party. At the application layer a builder can call AI integrations and is billed for LLM calls through integration credits, which implies some choice inside a built app, but no first party documentation of which models are selectable was located. NOTE THE ROUTING ITSELF IS THIRD PARTY SOURCED, so under the standing rule it confirms the capability EXISTS and does not establish its limits. docs.base44.com and third party platform analysis 2026-07-30 | Partial |
| APIs, SDKs & MCP ExtensibilityONE OF THE STRONGEST Ext CASES IN THE INDEX, and unusually AGENT NATIVE. TWO FIRST PARTY MCP SERVERS SHIP, which no other vendor in this pass matches: app.base44.com/mcp for creating and managing projects, listing projects and querying project data behind OAuth, and docs.base44.com/mcp for searching the documentation from any AI tool with no authentication required. Alongside them: a full developer platform at docs.base44.com, a CLI, a typed SDK with published references, backend functions running custom server side logic in a secure runtime, two way GitHub integration, and workspace owned API keys exposing audit logs and workspace data through a Monitoring API. Generated apps are STANDARD REACT BUILT WITH VITE rather than a proprietary runtime. An llms.txt documentation index is published at docs.base44.com/llms.txt. The onboarding path is itself agentic, with the documentation instructing users to let an AI coding agent install the CLI, create the project, define entities, write code and deploy. THE LIMITATION, recorded because it is material: GitHub export covers the FRONTEND ONLY. docs.base44.com developer platform documentation 2026-07-30 | Full |
| Testing, Debugging & OptimizationUPGRADED FROM UNKNOWN, which the unsourced grid carried, so this cell was unrated rather than wrong. A REAL ITERATION LOOP EXISTS FOR THE ARTEFACT BEING BUILT: generate, preview instantly on built in hosting, inspect, and refine either by prompting again or through a visual editor that allows direct click to change edits to layout, copy and styling, with real time collaboration and version control tracking changes. Two way GitHub export lets a team run their OWN external security tooling and specialised compliance scanners over the generated code, which is customer controlled testing of a kind few vendors permit. WHAT IS ABSENT IS EVALUATION OF THE AGENT: no harness, no sandbox for rehearsing a Superagent against test inputs before it runs live on schedules and real events, no automated scoring, no regression testing when a prompt or integration changes. Third party review reports that debugging consumes build credits without necessarily resolving the problem, which is a cost characteristic of the loop rather than a capability claim and was not graded. docs.base44.com, base44.com/security and third party platform reviews 2026-07-30 | Partial |
| Browser & Computer UseNone, and consistent with the test this pass settled on the shopify-sidekick build: Comp is non zero only when an agent operates software the vendor does not control because no programmatic interface exists. Base44 is the opposite case. Superagents reach external systems through integrations, OAuth 2.0 connectors and APIs, and reach people through WhatsApp, Telegram, email and SMS, all of which are programmatic channels. Backend functions execute server side logic in a secure runtime rather than driving an interface. No browser control, no headless session, no screen operation and no robotic process automation was documented across three passes covering the developer platform, the enterprise documentation and the Superagents launch material. The sandboxed environments agents run in are runtime isolation, credited under oversight, not computer use. docs.base44.com developer documentation and Superagents announcement 2026-07-30 | Unable to verify |
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
Base44 now allows users to resolve GitHub merge conflicts directly within the platform using AI. The system also automatically generates concise English-language branch names based on the user's initial message.
Bears on: Integrations
View sourceUsers can now restore deleted applications from the Trash view for up to 30 days, with workspace admins able to restore any app across the workspace. The restoration process automatically rebuilds the backend, though custom domains and integrations must be reconfigured.
Bears on: Security / enterprise
View sourceBase44 introduced persistent memory capabilities for its Superagent, allowing it to automatically save, categorize, and refresh important facts and preferences. The update includes working context tracking per conversation, a reviewable memory log, and full user control to edit or delete stored memories.
Bears on: Memory / state
View sourcePricing
From $20/mo
credits
Included quota
Starter (entry paid, $16/mo annual): 100 message credits + 2,000 integration credits. Builder ($40) steps to 250 message + 10,000 integration credits and adds custom domains, GitHub (frontend) export, backend functions, and a free domain for year one.
What is public
Base44 (a Wix company) publishes five tiers (annual / monthly): Free ($0), Starter $16/$20, Builder $40/$50, Pro $80/$100, Elite $160/$200; annual saves ~20%. Plus a custom Enterprise option (dedicated architect, account support). Tiers are largely a credit ladder; production features (custom domains, GitHub frontend export, backend functions) unlock at Builder.
Billing mechanics
Flat monthly tier gated by two separate credit pools: MESSAGE credits (consumed when you prompt the AI to build/edit — ~0.3/credit in discussion mode, more for builds) and INTEGRATION credits (consumed at runtime when the live app calls an integration — LLM call, file upload, email, SMS, image gen — at 1 credit each regardless of type). Neither pool rolls over; you can't buy credits separately — you upgrade or wait for the monthly reset.
Cost watchouts
Two independent credit pools (either can block you), non-rollover credits, no separate credit top-ups, error/retry loops burning credits, backend/database/auth don't export (frontend-only GitHub export)
Variable cost rationale
Both build-time and runtime meter to credit pools, so heavier iteration or higher live-app usage forces tier upgrades — moderate exposure, but no surprise per-credit overage bill
Additional watchouts
Backend isn't exportable (lock-in), managed-backend outage risk, and browser-rendered apps have SEO limits; the published tier is a starting point — heavy live usage can force upgrades
Overage / add-ons
No per-credit overage or top-ups — hitting either pool's monthly cap stops new builds (message) or live integration actions (integration) until the next billing cycle or an upgrade. Existing apps keep running.
Sales call required
Mixed (some tiers require a call)
Free / trial
Free tier
Lowest paid plan
Free; ~$40–$100/mo
Commercial notes
Annual saves ~20%; reported up to 50% student/teacher discount on Starter; backend/database/auth are managed (no separate hosting/Supabase bill); acquired by Wix (~$80M, 2025)
Key ambiguities
Exact message-credit cost per build prompt varies by complexity; the Free plan's integration-credit allowance is listed inconsistently (100 vs 500) across sources/pages
Cancellation / refund
Monthly or annual (~20% off); cancel anytime via billing dashboard with access through the paid term; annual plans typically aren't prorated for refunds
Support SLA / resale
Priority support at Elite; dedicated architect/account support on Enterprise; standard support below
Missing data
Enterprise pricing is custom (contact-only); the Free plan's integration-credit number varies by source; per-prompt message-credit costs aren't fixed. Note: live paid tiers start at Starter $16/mo (annual), below the seed's $40 'from' anchor (which matches Builder).
Related vendors
- Lindy — No-code AI agent builder focused on automating repetitive business…
- Activepieces — Open-source automation platform with a self-hosted Community Edition…
- Adopt AI — No code agent builder that turns existing applications and manual…
- Agent.ai — A marketplace and network of hundreds of prebuilt AI agents plus a…
- AgentX — No code platform to build, evaluate, and deploy multi agent…
- AI Library — No code enterprise AI platform to build and run AI agents across the…
Alternatives to Base44
The closest documented capability profiles to Base44 among agent builders tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Aigensei10.0 / 14Fuller documented coverage on Knowledge Grounding & RAG
- Teneo10.0 / 14Fuller documented coverage on Model Flexibility & Routing
- FLOWX.AI11.5 / 14Fuller documented coverage on Knowledge Grounding & RAG and Deployment & Data Residency
- Gumloop11.5 / 14Fuller documented coverage on Deployment & Data Residency and Prebuilt Agents, Templates & Packs
- Joget9.5 / 14Fuller documented coverage on Deployment & Data Residency
- Superblocks11.5 / 14Fuller documented coverage on Knowledge Grounding & RAG and Deployment & Data Residency
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded