Base44 Founder Maor Shlomo on AI App Building and Competition

Base44 founder Maor Shlomo joins Harry Stebbings to discuss selling the AI app builder to Wix, competition among model providers, margins and the future of software. Shlomo has financial and professional interests in Base44 and Wix, and says his acquisition compensation includes revenue milestones. His company claims and forecasts are not independently verified.

Shlomo says he began Base44 as a bootstrapped project after a capital-intensive prior company, motivated by a desire to code again. As users grew, he considered remaining independent, raising substantial funds, or joining a larger company. The user and profitability figures he gives are his account.

He initially worried an acquisition might undermine the independence of the product. After conversations with Wix CEO Avishai Abrahami, he saw a shared audience and an opportunity to keep a small product team while using Wix’s marketing, support, legal and operations. His praise of Wix reflects his financial and professional relationship.

Stebbings cites a revenue figure and asks whether Shlomo regrets selling. Shlomo says he cannot comment on revenue because Wix is public and says compensation includes additional milestones. The host’s figure is not confirmed by Shlomo, and annualized run rates should not be confused with recognized revenue.

Shlomo predicts customized AI-built software may eventually absorb parts of CRM, support and task management. Stebbings disputes the idea that most businesses will build their own Salesforce and maintain it. The exchange is a disagreement about future adoption, not evidence that existing software is already obsolete.

Shlomo suggests customers could start from templates or open-source projects and adapt features, rather than prompting an entire enterprise system from scratch. He argues ownership and customization could matter, while acknowledging enterprise adoption would take time. Security, maintenance and economics remain unresolved.

Asked whether small businesses care about code ownership, he says simplicity and fit may matter more. He recalls trying to configure software for his partner’s small business and deciding an LLM-supported application with a database and integrations could be tailored to her workflow. The example is a personal origin story, not a comparative trial.

Shlomo predicts some smaller CRM vendors and other simple database-front-end products may face disruption, while large incumbents have more resources to adapt. His forecast is made by the founder of a competing app-building platform and does not establish any named firm’s future outcome.

Asked about defensibility, he says building a basic AI coding tool is easier than providing reliable applications with databases, authentication, integrations and scheduled tasks. He claims some Base44 projects are very large, but the interview does not audit code size, security or application quality.

He says Base44 focuses on functional applications rather than simple marketing websites, which aligns with Wix’s existing business. He expects simple front ends to become commoditized while more complex, repeatedly used applications may generate more durable demand. These are product and margin predictions.

Stebbings maps potential competition among app builders, incumbents and developer tools. Shlomo expects existing SaaS products to add customization capabilities, offering Monday as an example and saying he does not know Salesforce’s plans. His vision of a customizable interface over a system of record is hypothetical.

He says Base44 is designed for people who do not want to see code, while Cursor and other tools focus on developers. He acknowledges that some software, such as a native firewall, is difficult or impossible for Base44 to build at the time. His comparison is a vendor’s assessment, not a universal capability ranking.

Shlomo describes scheduled tasks and messaging integrations as ways to expand the kinds of applications users can create. He expects developer tools and visual builders to move toward one another, but says different users have different needs and he does not know the ultimate market structure.

He says he worries less about a named app-builder competitor than about model-provider dynamics. If one provider gains a commanding lead, he thinks it could move into application building. He identifies Google as a potential threat because of its compute, distribution and cloud stack; this is a scenario, not a forecast of a winner.

Shlomo describes routing different workloads among model providers based on performance, cost and throughput. He says one model worked well for initial design and another for difficult bugs, and that changing providers can shift spending quickly. These are Base44’s reported experiences, not independently measured rankings.

He contrasts model switching with the harder migration of an entire cloud provider, while observing that model companies also build chat and coding interfaces. He does not know what would happen if a competitor offered a better and cheaper model for months; the revenue consequences he raises are hypothetical.

Shlomo says Wix and Base44 plan around falling model prices and decide not to optimize every temporary limitation. This is a strategic assumption, not a guarantee that model costs will approach zero or that vendors will maintain access and terms.

He declines to disclose the share of each revenue dollar paid to model providers. He argues that routing simple prompts to smaller models could improve margins and speed, while costly autonomous runs may waste resources if they misunderstand users. Stebbings is skeptical; no actual margin is established.

Shlomo says rapid copying makes competition intense and favors investments in harder-to-replicate infrastructure alongside fast product iteration. He describes Base44’s built-in backend, including database and user-management services, as a strategic bet. Its defensibility and migration costs are his commercial assessment.

When asked what percentage of users finish a project, he says “finish” is ambiguous and does not provide a rate of sustained usage. He says many applications are published, but distinguishes a working deployment from reaching the day-to-day use a customer intended.

He describes using the sentiment of prompts as one indicator of whether the agent follows instructions, and says feedback changed after model updates. Such sentiment is a proxy rather than a validated measure of correctness, safety or customer success.

Shlomo says category growth matters while margins and healthy retention should still be monitored. He expects model-related costs to change rapidly; he does not provide current unit economics. His view that future routing will improve profitability is a forecast.

Stebbings asks how early-stage investors should weigh fast revenue. Shlomo, who says he is starting to angel invest, looks for growth, resilience against competitors and the possibility of vertical integration. He presents these as his criteria, not a formula that ensures returns.

He uses Cursor’s Composer as an example of a possible strategy: build distribution using external models and later route workloads to a company’s own efficient model. He does not know Cursor’s margins, and the benefit he describes is hypothetical. He says Base44 is considering related options.

He contrasts a product whose value is mostly a prompt around a model with one that owns infrastructure or an operating business. He suggests AI-native services, including law firms and hospitals, as possible investments. These are speculative business ideas, not evidence that AI systems or surgical robots are clinically safe or effective.

Shlomo says angel investing lets him work with founders on new possibilities but is risky. He is cautious about general agent startups when he cannot assess whether model providers will commoditize them. His preferences reflect his own business and investment interests.

On competing with Cursor, he says Base44 began with nontechnical users and built integrations so they would not need to configure a backend themselves. He expects the share of software users can build without inspecting code to grow. These are product claims and predictions, not proof that code review, tests or security checks are unnecessary.

Given hypothetical valuations for Cursor and Cognition, Shlomo chooses Cognition, expecting room for multiple successful firms rather than one winner. His comparison is personal and time-specific, not an investment recommendation or a verified valuation analysis.

Asked about an AI revenue slowdown, Shlomo says he does not personally expect one and believes existing models still have substantial untapped value. He also raises uncertainty over model competition and margins. His expectations do not establish that a slowdown or technical plateau cannot occur.

He contrasts Base44’s lean beginnings with a previous company that raised substantial capital, and predicts greater productivity and lower prices may eventually benefit consumers. He declines to disclose current Base44 headcount. Neither his example nor the acquisition establishes an economy-wide employment outcome.

Stebbings compares reported private AI-builder valuations with Wix. Shlomo says his non-objective view is that Wix is undervalued, but does not know enough about competitors’ internal economics to say they are overpriced. His compensation and role at Wix are relevant interests.

Shlomo says Wix taught him to focus on product and marketing rather than competitors’ publicity. He describes reported improvements in conversion and unit economics after product and marketing changes; these company figures and causal claims are not independently audited.

He says operating outside Silicon Valley has not harmed Base44 in his view, and cites Wix’s financial backing and talent competition in the Valley. His comments are a company-specific assessment, not evidence that location is irrelevant to every startup.

Asked about AI-written code, Shlomo predicts its share will rise and estimates that most Base44 code is produced through prompts. He does not provide a measurement method; his percentages and timeline are estimates and predictions.

For nontechnical builders, he recommends starting with a problem they understand, making a disposable first version, and iterating or reverting quickly. His partner’s CRM is his example. This approach still requires checking whether the resulting application works for its intended use.

Shlomo says money can buy time, help people he cares about and fund future projects, while he does not equate it directly with happiness. These are personal reflections; the sale price discussed by the host is a company transaction amount, not established personal proceeds.

In quick questions he favors Anthropic at the proposed hypothetical prices, praises Jack Dorsey’s builder mindset and reflects on time with friends and family lost while building his previous company. His company comparison is an opinion, not financial advice.

He questions how defensible model providers will be if users can switch to a better or cheaper model, while acknowledging products built above the model may retain users. He does not know the providers’ internal strategies, and his platform competes in the same ecosystem.

Shlomo discusses the value he finds in a partner who understands the demands of entrepreneurship. He presents this as his personal relationship experience, not a universal rule for choosing a partner.

He recalls early outages while operating Base44 alone and the emotional impact of frustrated users. Looking back, he says he would reassure himself that the community could recover and that setbacks could be managed. The episode closes with both speakers thanking each other.