Lovable

What people rate it for. Reviewers consistently credit speed and accessibility: a written idea becomes a working web prototype quickly, which compresses the scaffolding work founders and product teams would otherwise pay for. No Code MBA's August 12, 2026 review treats the workflow as practical for prototypes and regular iteration. The real-code and Git path draws separate praise, because a technical person joining later inherits a repository instead of a closed builder. Reviewers also rate the breadth of built-in backend: database, auth, storage, hosting, integrations and AI without assembling a stack first. Aggregate scores back this up unevenly. G2 sits at 4.6/5 from more than 270 reviews and Trustpilot at 3.9/5 as of May 2026, both secondary snapshots worth rechecking.

Credit economics, and the billing terms behind them, are the loudest complaint. The recurring criticism targets cost predictability, not output quality. Reviewers describe debugging loops that consume many credits for one outcome, and a secondary roundup quotes a developer whose single bug fix used 30 credits. One Trustpilot reviewer, quoted in a June 2026 roundup, called the experience "less like engineering and more like gambling," and described uncertainty about what the next prompt would break. The billing terms press on the same worry. Trustpilot carries a detailed allegation that a same-day subscription refund was refused, with the reviewer writing that "your refusal to issue a refund for my subscription constitutes a breach of European Union consumer protection law." That is a reviewer's allegation, not a finding against Lovable, but the August 28, 2026 terms do state that fees are non-refundable except where required by law. Budget for the iteration as well as the build.

Reliability falls off as projects get complex. Reported patterns include repeated bugs, fixing one thing and breaking another, incomplete logic on complex requirements, and projects that become difficult to scale. So Lovable is strongest for prototypes and app scaffolding, and you shouldn't treat generated output as production-ready without testing, security review, version control and cost monitoring.

An alleged 2026 access-control incident is secondary-sourced only. A May 9, 2026 account on webaistack.com says that between February 3 and April 20, 2026, authenticated users could reach public project chat histories and source code belonging to other users, and the issue went undetected for about 11 weeks. The same account says Lovable fixed the vulnerability within two hours of disclosure and made historical public projects private by default. No official Lovable incident advisory was found, so this stands as an allegation awaiting primary confirmation rather than a documented breach.

The data-training policy changes on September 9, 2026. Lovable's documentation says that from that date, Free and Pro customer data may be used to train, develop and improve its models unless the user opts out in account settings, while Business and Enterprise data is excluded by default. That runs against a November 2025 data processing agreement saying customer personal data is not used for AI or ML training, so the governing version and scope need reconciling. If you're on Free or Pro and the data matters, check the opt-out toggle.

The growth numbers are the company's own. TechCrunch reported on August 12, 2026 that Lovable told it annualized run-rate revenue reached $500 million in June 2026, and coverage says it passed $100 million annualized within eight months of public launch. Both are company claims, not audited figures. The $13.3 billion valuation attached to the August 2026 round is press-reported, and the 539-employee headcount comes from a third-party record with no exposed methodology or date.