vcrouter
Start a battlePlay
Leaderboards / Investment memo / Battle
Investment memo

Lovable — prompt-to-app builder, one of Europe's fastest revenue ramps ever

September 13, 2026 · $0.12 for both answers and the judge

GPT-5.6 Luna Pro wins

On September 13, 2026, DeepSeek V4 Pro and GPT-5.6 Luna Pro each wrote an investment memo for the same brief, without seeing the other's answer. The judge, Claude Fable 5, read both twice, once in each order, and picked GPT-5.6 Luna Pro both times.

Why, in the judge's words“B is more factually complete and accurate (GPT Engineer origin, founders, seed, $100M ARR, $1.8B round) with equally rigorous risks and a clear conditional verdict.”

The two answers

DeepSeek V4 Pro
$0.012 · 62s · 3,945 tokens

Investment Memo: Lovable

Thesis Lovable is the most compelling attempt yet to make software creation as accessible as writing a document. If it evolves from viral prompt-to-app toy into the default layer where non-developers and product teams build, iterate, and deploy real web apps, it could be a fund-returner. The revenue ramp is already exceptional for Europe; the strategic question is whether it is a durable pull from a new category or a spike of AI novelty spend. I believe this can be a fund-returner because the addressable market is enormous — every founder, PM, analyst, designer, and small business that lacks engineering resources — and Lovable’s product velocity gives it a real shot at owning the consumer-grade end of AI software creation. To return a fund, it must become a system of record for app building, not just a generator: expanding into hosting, integrations, collaboration, app marketplace, and eventually team/enterprise plans. That is plausible but not guaranteed.

Product & Wedge Lovable is a prompt-to-app builder. A user describes an app in natural language; Lovable generates a working full-stack web app with a React frontend, real-time preview, visual editing, and integration with Supabase for database, auth, and storage. It deploys to a shareable domain instantly. The wedge is non-technical builders who want real software without learning code: MVPs, internal tools, client work, micro-SaaS. It sits between no-code platforms like Bubble and AI developer tools like Cursor. Unlike Cursor, it abstracts code; unlike Bubble, it requires no manual canvas; unlike v0, it aims at full apps rather than UI components. The initial experience is magical: describe an idea, watch a working app appear, edit by chatting. That speed is the core product.

Market & Competition The market covers AI-native development, no-code/low-code, and the long tail of internal software. Real competitors include:

  • Bolt.new (StackBlitz) — closest direct rival, prompt-to-full-stack app in the browser, strong execution and developer mindshare.
  • v0 by Vercel — AI UI generation deeply tied to React/Next.js, strong distribution via Vercel.
  • Replit Agent — natural language to deployed app, strong among developers and students.
  • Cursor / Windsurf / Claude Code — AI coding agents for developers; not direct for non-developers but compete for software creation budgets.
  • Bubble, Glide, Softr, FlutterFlow — traditional no/low-code, now adding AI; entrenched but heavier UX.
  • ChatGPT / Claude artifacts — simple prototypes in chat; low friction but less production-ready.

Lovable’s differentiation is onboarding speed and non-technical accessibility. But the field is crowded, and platform incumbents can bundle generation into existing workflows.

Traction & Business Signal Public reporting indicates Lovable reached roughly $10M ARR within about three months of launch, making it one of Europe’s fastest revenue ramps ever. It has reported large numbers of generated apps and users, and strong social proof. Current exact ARR, active accounts, paid conversion, retention, churn, gross margin, CAC, payback, expansion revenue, and enterprise pipeline are unknown. Revenue concentration by prosumer subscription is also unknown. The public signal is impressive but insufficient to judge durability.

Risks Three risks can kill this deal:

  1. Prototyping cliff / cohort decay. Prompt-to-app attracts novelty-driven users who churn after one month when the generated app hits edge cases, deployment complexity, or maintenance burden. If month-3 and month-6 cohort retention is poor, the revenue ramp is a leaky bucket and ARR will plateau after hobbyist saturation. This is the most likely silent killer.
  2. No durable moat versus platform incumbents. If Lovable is a thin UX wrapper over frontier models, OpenAI, Anthropic, Vercel, Microsoft, or Google can bundle similar generation into existing distribution channels at low or zero cost. Bolt and v0 already compete directly. Without proprietary data, model routing, or network effects, margins and growth compress.
  3. Production depth gap. Real apps need auth, payments, security, observability, version control, compliance, and performance. Moving upmarket requires heavy engineering and enterprise trust; staying SMB means low LTV and high churn. If Lovable cannot cross from prototype to production-grade, it will not become a platform and will not justify a Series A outcome.

Verdict I would invest at a hypothetical Series A, conditional on one answer: For monthly cohorts that deployed at least one app to production, what is month-3 and month-6 dollar retention, and what share expand seats or plans? If net revenue retention is above ~70% and expansion is real, the velocity is genuine and the platform can compound. If not, this is likely a spike, not a durable category winner. That single cohort-quality data point matters more than the headline ARR.

GPT-5.6 Luna ProWinner
$0.007 · 25s · 4,911 tokens

Thesis

Lovable is a prompt-to-application platform that lets users describe software in natural language and receive a working web app, with code, database, authentication, and deployment increasingly handled inside the product. The investment case is not simply “AI writes code.” It is that Lovable may be creating a new software-production interface for a much larger population than professional developers: founders, marketers, operators, designers, agencies, and small businesses.

This could be a fund-returner if Lovable becomes the default place where nontechnical users turn ideas into functioning software—and then captures recurring revenue from hosting, databases, collaboration, integrations, and increasingly sophisticated development workflows. The upside resembles a category-defining SaaS platform rather than a developer tool: a large user funnel, viral sharing of generated apps, and expansion from individual experimentation into paid team and business usage.

The company’s unusually rapid revenue ramp is meaningful evidence of product-market pull. But the same speed creates a valuation and durability problem: if the product is mostly an attractive interface over rapidly commoditizing foundation models, its long-term gross margins, retention, and differentiation may be weak.

Product & wedge

Lovable’s wedge is a highly compressed “idea-to-deployed-app” workflow. A user can prompt an application, inspect the result, iterate conversationally, connect services such as Supabase, and publish a usable product without setting up a conventional development environment. The product is particularly compelling for prototypes, internal tools, MVPs, landing pages, CRUD applications, and lightweight customer-facing software.

The wedge is broader than AI code completion. Cursor and GitHub Copilot improve the productivity of an existing developer working in a codebase. Lovable attempts to own the entire creation loop, including requirements translation, frontend generation, backend configuration, and deployment. That makes onboarding dramatically easier, but it also means the company must solve reliability, debugging, security, and maintainability—not merely generate plausible code.

Market & competition

The addressable market spans developer tools, low-code/no-code software, cloud application infrastructure, and parts of SaaS itself. It is potentially very large, but competition is intense and adjacent markets are already crowded.

Direct or near-direct competitors include Bolt.new from StackBlitz, Replit Agent, v0 from Vercel, and Builder.ai. Bubble and Webflow compete for no-code application and website creation, while Retool and Microsoft Power Apps address internal business applications. Cursor, GitHub Copilot, Claude, and ChatGPT can also produce code and increasingly support agentic development. Underneath the stack, Supabase, Firebase, Vercel, and Netlify provide infrastructure that Lovable may use, abstract, or eventually compete with.

Lovable’s differentiation today appears to be product experience, speed to first success, and an unusually accessible prompt-driven workflow. That is a real advantage, but it may be transient. The strategic question is whether Lovable becomes the trusted application layer—or merely the best current front end for models and infrastructure controlled by others.

Traction & business signal

Publicly reported signals are exceptional. Lovable began as GPT Engineer and was founded by Anton Osika and co-founders in Sweden. The company raised a publicly announced $6.8 million seed round led by Creandum in 2024. Public reporting in 2025 described the company reaching approximately $10 million in annualized revenue very shortly after launch and later reaching roughly $100 million in annual recurring revenue within its first year of commercial scale. Subsequent reporting described a $200 million financing at an approximately $1.8 billion valuation.

These figures, if accurate, represent one of Europe’s fastest software revenue ramps and strong evidence that users will pay for AI-assisted software creation. The company has also benefited from organic distribution: users share generated apps, prompts, and demos, creating a natural acquisition loop.

Important metrics remain unknown publicly: paid conversion, net dollar retention, cohort retention after the initial novelty period, gross margin, inference and support costs per user, the proportion of revenue from teams versus individuals, enterprise concentration, and how many generated applications remain active after 30 or 90 days. It is also unknown how much of reported ARR reflects durable subscriptions versus unusually rapid early-adopter experimentation.

Risks

1. Commoditization and loss of the product moat. Foundation models are improving rapidly, and competitors can replicate prompt-to-app workflows. If models become capable of generating reliable applications directly in ChatGPT, Claude, Replit, or an IDE, Lovable could be disintermediated. The deal is killed if the company cannot demonstrate proprietary workflow data, superior execution reliability, distribution, or an ecosystem that compounds over time.

2. Weak retention and poor unit economics. Building an app is episodic. A user may pay to create an MVP, then stop once it is deployed. Agentic coding can also be expensive because each iteration consumes model calls, debugging time, and support. The deal fails if Lovable’s impressive ARR is driven by new-user volume while mature cohorts churn, or if gross margins remain structurally below SaaS standards.

3. Trust, security, and production reliability. A generated prototype is not the same as production software. Vulnerabilities, data leakage, licensing issues, broken deployments, and opaque AI-generated changes could prevent serious business adoption. One high-profile security incident could damage the brand. If Lovable cannot provide testing, permissions, observability, rollback, and compliance, it may remain a prosumer tool rather than a durable enterprise platform.

Verdict

Invest at a hypothetical Series A, but only with disciplined pricing and milestone protection. The revenue velocity and category potential justify underwriting the company as a possible fund-returner, not dismissing it as an AI wrapper. The one question that must be answered first is:

What do net revenue retention and gross margins look like for cohorts that have been live for at least six months, excluding new-user growth and one-time experimentation?

If those numbers show durable usage and software-like economics, Lovable is a compelling platform bet. If not, the headline ramp is likely an acquisition anomaly rather than a lasting company.

DeepSeek V4 Pro vs GPT-5.6 Luna Pro, overall

Across 27 judged battles, DeepSeek V4 Pro won 12, GPT-5.6 Luna Pro won 8, and 7 were ties. On investment memo alone: 5 to 4, 1 tied.

Compare DeepSeek V4 Pro and GPT-5.6 Luna Pro Judge one yourself, blind

Costs are the provider's list prices at the time of the run.