New York's Moonshot AI becomes Shuttlebase after colliding with Beijing's namesake

Why it matters

The rebrand removes a growing sales and search problem while sharpening the founders' bigger bet: merchants will trust AI to change live storefronts, not merely recommend tests.

Aviv Frenkel and Evyatar Segal renamed their New York e-commerce optimization startup Shuttlebase on September 8th, ending its run as Moonshot AI after the shared name created confusion with the Beijing lab behind Kimi.

The founders announced the rebrand three years after starting the business. Frenkel said they had not planned to change the name, but the collision pushed them to reconsider what it should communicate. "We want to build the launchpad that allows eCommerce brands to continuously improve and move faster," he said.

That launchpad is an autonomous conversion-rate optimization system. Shuttlebase says its software studies shopper behavior and site structure, finds places where a store is losing sales, generates changes to design, copy and code, runs experiments against live traffic, and promotes winning versions. The underlying product and founders remain the same. The clean break is in the brand.

Two Moonshots, founded in the same year

Frenkel and Segal founded the New York operation in 2023. Beijing-based Moonshot AI, founded during the spring of 2023, took the same English name while building the Kimi family of foundation models and agents.

The overlap became harder to manage as Kimi moved beyond its home market. RuntimeWire reported in August that Beijing Moonshot was seeking a share of Kimi K3 service revenue from major US cloud providers as it pursued enterprise distribution. The lab had already released Kimi K3's weights in July, giving developers another reason to associate "Moonshot AI" with Yang Zhilin's model operation rather than Frenkel and Segal's e-commerce software.

For a young B2B vendor, that confusion reaches beyond search results. Prospective customers, recruits and investors need to know which product they are evaluating before a sales conversation can begin. Every Kimi release made that explanation harder.

Shuttlebase gives Frenkel and Segal a name tied to their own pitch. It also drops one of the most heavily recycled words in the AI market at a point when the founders are trying to sell measurable revenue gains rather than general-purpose intelligence.

A founder problem turned into software

Frenkel has described Shuttlebase's origin as a problem he encountered while operating an e-commerce business. Each website experiment required coordination among designers, developers and analysts, while the connection between a change and an eventual sale remained uncertain. In Shuttlebase's October 21st, 2025 funding announcement, he called conversion optimization "voodoo."

His background is unusually split between operating and studying startup networks. Frenkel's public profile lists a PhD completed at Bar-Ilan University in 2023. His dissertation examined the role Israeli technology workers in Silicon Valley play in constructing Israel's startup identity.

Segal brought the engineering perspective. In a 2025 interview with AlleyWatch, Frenkel identified Segal as one of Via Transportation's first employees and said his co-founder had encountered similar bottlenecks while running large numbers of digital experiments.

The founders' bet is that those roles can be compressed into software. Each store and experiment is also supposed to improve the models used for later customers, according to Shuttlebase.

That last point is central to the pitch and remains a Shuttlebase assertion. Shuttlebase says its models were trained on millions of conversions, but it does not publish the composition of that data or enough methodological detail to evaluate how results transfer between brands.

Revenue claims carry the usual experimentation caveats

Shuttlebase says it has produced millions of dollars in measurable revenue uplift for customers including Credo Beauty, Kopari, Peter Thomas Roth and P&G's Farmacy Beauty. Shuttlebase reports that DefenAge increased revenue per visitor by 30% and conversion by 14%, while one Alembika experiment generated $114,000.

Those results come from Shuttlebase and its customers. The announcement does not provide the traffic volumes, control-group performance or statistical thresholds behind them. Conversion experiments can also be sensitive to seasonality, promotions, product mix and the period selected for measurement. The figures show the business case Frenkel and Segal are selling; they do not establish a general performance rate for the product.

Trust is the larger commercial test. Shuttlebase wants merchants to let an outside system diagnose problems, change live storefronts and allocate visitor traffic without waiting for a growth team to approve every step. That promise can increase testing volume. It also makes safeguards, attribution and rollback controls part of the buying decision.

The category is moving toward agents

Shuttlebase last disclosed financing on October 21st, 2025, when it raised a $10 million seed round led by Mighty Capital, with Oceans Ventures, Uncorrelated, Garuda Ventures and Almaz Capital participating. Frenkel later told AlleyWatch that the financing brought total capital raised to $13 million. No valuation was reported.

That capital placed Shuttlebase against established experimentation vendors that have been adding their own AI layers. VWO introduced VWO AI on June 1st, pitching a conversational assistant that can analyze user behavior, recommend changes and launch experiments. Optimizely now markets agentic experimentation spanning test ideas, execution and analysis.

Shuttlebase's stated distinction is the amount of work delegated to the system. Frenkel and Segal want the software to own the full cycle from diagnosis through deployment, rather than serve as an assistant inside an existing experimentation program. That is a larger automation claim and a harder product to earn permission to run.

The new name clears away a distraction before Shuttlebase tries to scale that pitch. Frenkel and Segal no longer need to explain why an e-commerce conversion vendor shares a name with one of China's most visible model developers. They can spend that time explaining why a merchant should let their software rewrite the store.