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 · $0.14 for both answers and the judge

Grok 4.6 wins

On September 13, 2026, Gemini 3.1 Pro and Grok 4.6 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 Grok 4.6 both times.

Why, in the judge's words“B is more factually accurate (2025 round, ARR scale), more careful separating claims from verified data, and gives a sharp price-conditional verdict; A misdates facts.”

The two answers

Gemini 3.1 Pro
$0.027 · 24s · 2,230 tokens

To: Investment Committee From: [Investor Name] Date: October 24, 2023 Subject: Investment Memo: Lovable (Series A)

Thesis

Lovable has the potential to be a fund-returner because it is capturing the most valuable real estate in the next paradigm of computing: the translation layer between human intent and functional software. While tools like GitHub Copilot and Cursor make existing developers faster, Lovable is expanding the market entirely. By enabling true "prompt-to-app" generation, Lovable transforms product managers, designers, and domain experts into full-stack developers. If Lovable becomes the default creation engine for the next 100 million software creators, it will command a massive premium in a TAM that effectively encompasses the entire global knowledge economy. Its unprecedented revenue velocity proves acute, immediate market demand.

Product & Wedge

The Product: Evolving from the highly successful open-source project GPT Engineer, Lovable is a browser-based, AI-native software development platform. Users input natural language prompts, and Lovable generates, iterates, and deploys full-stack web applications. It handles the UI, business logic, and backend scaffolding, allowing users to converse with the AI to tweak features in real-time.

The Wedge: Lovable’s wedge is zero-setup rapid prototyping and internal tool creation. By entirely bypassing the command-line interface (CLI), local environment setup, and traditional IDEs, Lovable targets founders, PMs, and designers who want to go from idea to functional prototype in minutes. Once the prototype is built, the wedge deepens as users rely on Lovable to iterate and host the application, shifting it from a prototyping tool to a production dependency.

Market & Competition

The market for AI software generation is arguably the most fiercely contested space in venture capital right now. Lovable is not competing in a vacuum; it faces formidable, well-capitalized opponents.

  • v0 by Vercel: Vercel’s generative UI tool is highly popular, deeply integrated into the React/Next.js ecosystem, and backed by Vercel’s massive distribution and hosting infrastructure.
  • Bolt.new (StackBlitz): A direct competitor offering in-browser, prompt-to-full-stack capabilities with WebContainers, allowing real-time execution of Node.js environments in the browser.
  • Replit (Replit Agent): Replit already owns a massive user base of beginner and intermediate developers and has recently launched powerful autonomous agent capabilities.
  • Cursor & GitHub Copilot: While currently focused on traditional developers, they are moving up the abstraction stack toward natural language app generation.
  • Claude Artifacts (Anthropic): Foundation models themselves are encroaching on this space by allowing users to generate and preview code directly in the chat interface.

Traction & Business Signal

  • Revenue: Lovable has demonstrated one of the fastest revenue ramps in European startup history, reportedly crossing $1M+ ARR in a matter of weeks post-launch.
  • User Growth: Massive organic, viral top-of-funnel growth driven by social media demonstrations (Twitter/X, LinkedIn) of complex apps being built in minutes.
  • Funding: Successfully raised a highly competitive $7.5M pre-seed/seed round led by top-tier European and US investors.
  • Retention & Churn: Unknown.
  • Unit Economics / Gross Margins: Unknown. (Given the heavy reliance on frontier model API calls, margins are likely compressed currently).
  • Active Production Deployments: Unknown.

Risks

These are the three existential threats that could kill this deal:

  1. The "Toy" Ceiling (Churn Risk): It is incredibly easy to generate a flashy 80% complete prototype, but notoriously difficult to use AI to cross the final 20% required for a secure, scalable production app. If Lovable cannot bridge the gap between "cool prototype" and "maintainable production software," users will churn after their initial dopamine hit. High top-of-funnel conversion means nothing if Month-2 retention falls off a cliff.
  2. Foundation Model Overhang: Lovable is essentially a highly optimized wrapper and orchestration layer on top of models like Claude 3.5 Sonnet and GPT-4o. As Anthropic (via Artifacts) and OpenAI continue to build native UI generation and execution environments into their consumer interfaces, Lovable’s core value proposition could be commoditized by the very APIs it relies on.
  3. Distribution and Hosting Disadvantage: Competitors like Vercel (v0) and Replit already own the hosting infrastructure and the developer community. Lovable has to convince users to build on its platform and figure out deployment, whereas Vercel offers a seamless, one-click pipeline from prompt to global edge network. Lovable risks being squeezed out by incumbents with superior distribution.

Verdict

INVEST.

Despite the fierce competition and platform risks, the sheer velocity of Lovable's execution and its historic revenue ramp indicate a team that has found profound product-market fit in a massive, expanding category. In venture, you pay for the option on a generational outlier; Anton Osika and the Lovable team are shipping at a pace that suggests they can outrun the incumbents and build a defensible ecosystem before the foundation models eat the space.

The one question I need answered first: "What is your Day-30 and Day-60 retention rate for paid users, and what percentage of generated apps are actively being maintained and used in production versus abandoned after the first session?"

Grok 4.6Winner
$0.009 · 32s · 1,442 tokens

Investment memo: Lovable (prompt-to-app)

Thesis. Lovable sits at the intersection of two compounding forces: collapsing cost of generating working software and exploding demand from non-engineers (and time-starved engineers) who want production-ish apps without a sprint. If the product becomes the default “describe it, ship it” surface for a large class of internal tools, MVPs, and even customer-facing apps, it can own a high-frequency workflow with strong retention and expansion. Europe has produced few software companies that reach global scale at this velocity; a category-defining AI-native builder with network effects around templates, components, and data connectors could return a fund even from a late Series A if it captures even a mid-single-digit share of the no-code/low-code plus “AI coding” TAM. The bet is that prompt-to-working-app is a new primitive, not a feature of IDEs or design tools.

Product & wedge. Users describe an app in natural language; Lovable generates a full-stack application (UI, backend, auth, database) that can be iterated in the same chat, previewed, and deployed. The wedge is speed-to-something-that-runs rather than code quality or pixel-perfect design. That is distinct from “better autocomplete” (Cursor) or “better Figma-to-code” (v0). Early users appear to be founders, PMs, and operators who previously bounced between Bubble, Retool, and a contractor. The product’s defensibility, if any, will come from (1) reliability of generated apps under iteration, (2) integrations and data connectors that make the output actually useful, and (3) a library of remixable apps that compounds. Public demos emphasize end-to-end generation over fine-grained code control.

Market & competition. The addressable market is the union of no-code/low-code (Bubble, Webflow, Retool, FlutterFlow), AI coding assistants (Cursor, GitHub Copilot, Windsurf), and newer prompt-to-app tools (Bolt.new / StackBlitz, v0 by Vercel, Replit Agent, and various open-source GPT-Engineer descendants). Adjacent: Framer, Softr, Glide. Incumbents have distribution (Vercel, Replit, GitHub) and existing users who already pay for hosting or seats. Lovable’s opening is that most of those products still assume the user will touch code or a visual builder; a pure prompt loop that produces a deployable app can steal the “I just need this by Friday” use case. The market is large and growing; the risk is that the category fragments into IDE plugins, design-tool features, and hosting-platform agents rather than a standalone winner.

Traction & business signal. Public reporting and founder statements describe one of Europe’s fastest software revenue ramps: from launch (as GPT Engineer / Lovable) to multi-million ARR in months, with subsequent claims of tens of millions ARR and very high growth rates into 2025. Exact current ARR, net revenue retention, paid conversion, and cohort retention are not fully disclosed in audited form; treat precise figures as unknown beyond “exceptionally fast public ramp by European standards.” Team is Stockholm-based (Anton Osika et al.). Funding: seed and subsequent rounds have been reported, including a sizable 2025 round at a high valuation; exact cap table and last round terms are only partially public. Usage metrics (apps created, DAU, deploy rate) are marketed aggressively but not independently verified. Enterprise vs. SMB mix: unknown. Gross margin and inference cost trajectory: unknown, though the business is inherently GPU-heavy.

Risks (the three that actually kill the deal).

  1. Commoditization by platforms with distribution. Vercel, Replit, GitHub, and Google/Microsoft can ship “good enough” prompt-to-app inside products users already live in. If generation quality converges, Lovable becomes a feature, not a company. Switching costs for generated apps may be low if the output is standard React/Node.
  2. Unit economics and model dependence. Every generation and iteration burns tokens. If frontier models keep improving faster than Lovable’s product layer, customers may go direct to the model or to cheaper wrappers. Persistent negative contribution margin at scale, or a sudden price hike from a model provider, can stall growth or force a down-round. Public data on COGS is unknown.
  3. Quality ceiling and trust. Generated apps that look fine in a demo but fail on auth, data integrity, performance, or security will not survive beyond prototypes. If the product cannot reliably produce maintainable, production-grade software, it stays a toy for hackathons and internal MVPs—large TAM shrinks to a small, price-sensitive segment. Hallucinated architecture and “it worked until I asked for a change” are known failure modes in the category.

Verdict. Invest at a hypothetical Series A only if the round is not already priced for a global winner (i.e., not assuming $100M+ ARR certainty). The public ramp is real and rare in Europe; the product wedge is clear. Pass if the valuation already prices in beating Cursor + v0 + Replit simultaneously without proof of retention and expanding ACV.

One question first: What is net revenue retention and 90-day retention of paying workspaces that have deployed at least one app used by someone other than the creator? That single number separates “fast demo revenue” from a fund-returner.

Gemini 3.1 Pro vs Grok 4.6, overall

Across 30 judged battles, Gemini 3.1 Pro won 10, Grok 4.6 won 15, and 5 were ties. On investment memo alone: 0 to 5, 5 tied.

Compare Gemini 3.1 Pro and Grok 4.6 Judge one yourself, blind

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