Summation launches a $60 AI analyst that promises to check its work

Why it matters

Summation is testing whether enterprise-grade verification can become a self-serve product. The $60 plan lowers adoption friction, while shared context could give Summation a foothold that general chatbots struggle to hold inside companies.

Opendoor co-founder Ian Wong (@ianwong_) launched a self-serve version of Summation on September 10th, pitching an AI analyst that can work across corporate data without making executives verify every answer by hand.

https://x.com/ianwong_/status/2098106891737977191

In a 38-post thread on X, Wong positioned Summation against general-purpose products such as ChatGPT and Claude. Summation connects to data warehouses, enterprise resource planning software, customer relationship management systems and spreadsheets, then preserves definitions and operating context across an organization.

That shared context is central to Wong's pitch. A finance leader can teach Summation how revenue is recognized, for example, and colleagues can reuse that definition instead of rebuilding it inside separate chatbot sessions. Summation also turns recurring analysis into scheduled workflows with controls over what the software can do.

The newly published pricing makes Summation available beyond the large-enterprise deployments that defined its first year. The Pro plan costs $60 a month for 12,000 credits, while Max costs $200 for 40,000 credits, point-in-time data snapshots and priority support. Summation sells larger deployments at custom prices, including private-cloud and on-premise options, access controls, audit logs and forward-deployed support.

The $35M round happened last year

Wong's launch thread paired the product release with Summation's $35 million funding figure. The financing predates Thursday's release: Summation disclosed the $35 million when it emerged from stealth on October 1st, 2025.

Benchmark led Summation's seed round, while Kleiner Perkins led the Series A. Benchmark partner Chetan Puttagunta and Kleiner Perkins partner Josh Coyne joined Summation's board. Wong also thanked Basis Set, A*, Opendoor co-founder Eric Wu and several individual backers in Thursday's thread.

Summation was founded in 2024 by Wong and Ramachandran "RC" Ramarathinam, who worked together at Opendoor. Wong served as Opendoor's chief technology officer after previously becoming Square's first data scientist. Ramarathinam led the technology behind Opendoor's core home transactions, according to earlier reporting.

Wong's $10 billion-plus revenue shorthand for Opendoor is supported by its peak scale. Opendoor reported $15.6 billion of revenue in 2022, when it sold more than 39,000 homes. The same filing reported a $1.35 billion net loss, an important distinction for a business whose revenue largely reflected the resale price of houses.

Wong says the experience of pricing and operating that business shaped Summation. Executive meetings could begin with prepared dashboards and spreadsheets, then stall when leaders asked why a metric changed or what action should follow. Analysts sometimes needed weeks to produce an answer, by which point the decision had moved on.

Verification is the product

Summation's claim to "nearly zero hallucinations" rests on multiple agents checking numbers, calculations and assertions against underlying data. Users can trace figures to their sources, while Summation says its software independently recomputes outputs rather than asking another language model whether an answer looks plausible.

Summation's evidence remains largely based on tests designed and published by Summation. In an August 26th evaluation, Summation ran a financial-model verification task 20 times with a frontier model. Eighteen outputs were correct, one was visibly broken and another produced a balanced model that violated its own cash constraint. Summation said an independent mathematical check caught every error deliberately planted in a separate set of models.

That methodology addresses a real weakness in enterprise AI: polished output can conceal a broken calculation. It does not establish an independent hallucination rate for Summation across customers, data sources and business workflows. Wong's near-zero claim should be read as Summation's product target rather than an audited performance measure.

Wong named Fanatics and Lineage as customers. Summation's website carries testimonials from Fanatics Commerce CEO Andrew Low Ah Kee and Lineage data science executive Elliott Wolf. Summation says Fanatics cut some analyses from weeks to near-instant results, while Lineage reduced an anomaly-investigation workflow from weeks to minutes. Those performance figures come from Summation and its customers.

The self-serve release broadens Wong's original enterprise thesis. Summation entered the market selling an intelligence layer for executive planning and reporting. At $60 a month, Wong is betting the same verification machinery can become a daily tool for individual finance, revenue and operations teams, then spread through the organization as their definitions, workflows and institutional memory accumulate.