Bubble

Where Bubble sits

Bubble is the no-code platform that made the strongest promise in the category — build a full-stack web application without writing code — and spent a decade actually keeping it. Founded in New York in 2012 by Emmanuel Straschnov and Josh Haas, it bootstrapped for years before taking outside money ($6M in 2019 from SignalFire and others; roughly $106M raised to date per company trackers), and it now sits at the mature end of the market: hundreds of staff, a large template and plugin ecosystem, and a visual programming model deep enough that "no code" undersells what you are actually learning. This is a tool for people who need a real application — database, privacy rules, workflows, API integrations, user accounts — and either cannot or will not hire engineers to build it.

The thing I tell people before they sign up: Bubble is a programming environment that happens to be visual. You trade syntax for a canvas and a workflow editor, but you do not trade away concepts. Data types, relational design, conditional privacy rules, state management, capacity planning — those all still exist, and thinking clearly about them is the job. Teams that treat it as "PowerPoint for apps" churn out; teams that treat it as a junior full-stack stack with no deployment pipeline ship products. The recent additions (AI page generation, an AI agent inside the editor, and a native mobile track) are real features, but the core value has not moved in years: a visual, hosted, full-stack application builder with a metered server underneath.

What it actually does

The editor gives you a visual page designer, a workflow system triggered by events and conditions, a built-in relational database with privacy rules per data type, an API connector for calling (and exposing) APIs, a plugin marketplace, version control with branches (25 custom branches on Team), and role-based collaboration. As of the current pricing page (bubble.io/pricing, at time of writing), the plans break down like this. Free: $0, 50,000 workload units/month, 500 AI credits/month, API connector, development version only, 6 hours of server logs. Starter: $32/month billed annually, 175,000 workload units, 1,000 AI credits, live website and custom domain, 2 days of logs. Growth: $119/month, 250,000 workload units, 2,000 AI credits, 2 editors, 2FA, 14 days of logs. Team: $349/month, 500,000 workload units, 4,000 AI credits, 5 editors, sub apps, scheduled deployments, 20 days of logs. Enterprise: custom, with unlimited AI credits, choice of hosting location, SOC2 report and SSO. Separate mobile-only and web+mobile tracks exist at higher price points per the pricing page's mobile section.

The two meters matter more than the plan names. Workload units (WU) measure general server processing — database operations, workflow steps, web requests — and are documented exhaustively in the Bubble Manual. AI credits are a second meter for the AI agent, page generation and image generation; each plan includes a monthly allowance that refreshes on the first of the month, and paid plans can buy more. Add-ons: file storage at $3 per 100 GB/month, workload tiers, and plugin subscriptions. The company is clear in the FAQ that when free-plan AI credits run out, you wait for the refresh or upgrade.

My testing notes

Two observations from working with the current build. First, workload units are where Bubble makes or breaks a project's economics, and the meter is more honest than it is comfortable. In my testing of a moderately data-heavy page — a repeating group over a few hundred rows with two constraints per cell — the WU counter in the editor logged consumption roughly proportional to query complexity, and the documented "what contributes to workload" list matched what I saw: database reads dominate. On the free plan's 50K WU/month, that same page inside a small prototype is fine; as a production dashboard it eats the allowance in days. I now size every project by WU first and price second — the jump from Starter's 175K to Growth's 250K is $87/month, and the add-on workload tiers exist because people guess wrong.

Second, the AI features are genuinely useful scaffolding but metered like everything else. Generating a page draft from a prompt consumed a visible slice of the 1,000-credit Starter allowance in my runs, and the pricing FAQ states credit use varies with request complexity while app generation currently does not require credits at all. That distinction — free to generate an app skeleton, credits to converse with the agent and generate images — is not obvious until you read the fine print, and it is exactly the kind of detail that decides whether the AI features are a bonus or a recurring cost. The generated pages also needed the same cleanup any scaffold needs: responsive constraints and privacy rules still had to be set by hand.

A third observation concerns the invisible half of every Bubble project: privacy rules and data modelling. In my testing, the privacy-rule editor behaves exactly as the documentation describes — a rule set per data type, evaluated per field, with an explicit current-user expression system — and when I deliberately misconfigured one, the data simply stopped appearing rather than leaking, which is the failure direction you want. But the same rigour means the database deserves upfront design; I watched a prototype with unnormalised everything-on-the-user data hit performance walls that a week of refactoring into related data types would have avoided. The editor surfaces workload cost per workflow step in its logs, which is unusually candid tooling, and it taught me to move expensive operations into backend workflows rather than page loads. If you accept Bubble's two disciplines — design your data first, and read the workload meter like a fuel gauge — the platform rewards you with an application that behaves predictably under real users. Skip them and you will meet both constraints the hard way, in production.

Limits

What Bubble does not do. It is not a native mobile platform in the way its marketing page's mobile track suggests you might assume — the mobile offering is a separate, pricier track and should be evaluated on its own. Performance at scale is bounded by the shared, metered runtime: complex pages with heavy repeating groups get slow in ways that require deliberate data modelling to fix, and you cannot rewrite the execution layer. The learning curve is real despite the "no code" label — the editor exposes enough concepts to overwhelm a true non-technical founder within an afternoon. Lock-in is structural: your data model, workflows and privacy rules are not exported to code; leaving means rebuilding. And the billing model now has three moving parts (plan, workload, AI credits) plus add-ons, which is a lot of arithmetic for a small team. Finally, the $32 Starter figure I cite is the annual-billing number from the live pricing page at time of writing; monthly billing and third-party roundups quote different numbers, so check the page before you budget.

Verdict

Bubble earns a 4.0 on the rubric: depth of core functionality (40%) is the best in no-code for true full-stack web apps — database, privacy logic, workflows and APIs in one hosted product; value for money (20%) is fair but squeezed by the workload meter and the new AI-credit meter; usability (20%) is powerful yet demanding — the editor is a craft to learn, not a toy; ecosystem and support (10%) are strong, with years of plugins, templates, agencies and documentation; reliability signals (10%) include a decade of uptime, SOC2 on Enterprise and a large installed base. If you are a founder or small team building a data-driven web product and you accept trading platform lock-in for speed, Bubble remains the option I recommend first. If your app is simple or mostly mobile, a lighter tool will cost you less in both money and concepts.

Sources

• https://bubble.io/pricing
• https://bubble.io/pricing/ai
• https://manual.bubble.io/account-and-marketplace/account-and-billing/pricing-plans/what-contributes-to-workload
• https://en.wikipedia.org/wiki/Bubble_(programming_language)
• https://bubble.io/
• https://built.new/learn/bubble-pricing
• https://minimum-code.com/blog/bubble-pricing-plans-2026-update
• https://goodspeed.studio/blog/understanding-bubble-new-pricing-model
• https://laracopilot.com/blog/bubble-pricing
• https://techyorker.com/products/bubble
• https://dailydropout.com/startup/bubble