# Upsolve AI Aiolum story: https://aiolum.com/story/upsolve-ai/ · added 2026-11-05 · updated 2026-11-05 **In short.** Upsolve AI (Embedded analytics and governed data agents for SaaS products, Not stated in the sources): In production with Fortune 500 users, June 2026. Evidence grade E: no revenue figure was checked. Team: 2 co-founders; Y Combinator lists a team of 5 in November 2026. Launched 17 February 2024; still running. Embedded analytics and governed data agents for SaaS products · Not stated in the sources · AI-native · team of 2 to 10 · still running · group: Documents, data and slides - Outcome: In production with Fortune 500 users, June 2026 (evidence grade of the headline number: E; no revenue figure) - Launch: 17 February 2024; months from launch to the outcome: 28 - Team: 2 co-founders; Y Combinator lists a team of 5 in November 2026 - Outside funding: Y Combinator (Winter 2024); amount not published - Main factor: Domain experience - Meters, 1 to 4: evidence strength 1, copyable pattern 1, capital needed 2 - Tags: AI-native, Small team, Funded, Running, B2B, Business intelligence software Links: - Visit upsolve.ai: https://upsolve.ai/ **The product.** Software that lets a SaaS company put dashboards and an AI data analyst inside its own product, and lets a company's data team build analytics agents that answer questions from the company's data warehouse with permission rules and tests. Users ask questions in plain English; the agent writes the database query, runs it and explains the answer. **Why it worked.** The two founders had built a data product at Palantir before, and they started with a common problem: most B2B SaaS products need customer dashboards, and building them takes engineering time. From the first launch in February 2024 the builder and the question answering used AI. In 2025 and 2026 the product moved from dashboards toward AI agents that a company can trust, with tests and permissions built in. Evidence grades: A = disclosed by a public company, a filing or a payment processor, or verified by one; B = two or more independent reports agree, or a journalist confirmed it; C = a founder claim that no one else verified; D = reported, not verified: one founder claim or a third-party estimate with no second source; E = no revenue figure found, so the headline is a non-money fact and the other facts carry their own grades; calc = computed on this page from graded facts. ## Summary Upsolve AI sells embedded analytics, which means charts and dashboards that a software company places inside its own product for its own customers, plus a data agent, an AI program that turns a question in plain English into a database query (often text-to-SQL, the translation of a question into the SQL query language), runs it and explains the result. Its first launch post on Y Combinator, 17 February 2024, already described "an AI-powered builder" and "AI answering any additional data questions". Ka Ling Wu (CEO) and Serguei Balanovich (CTO) founded it in 2023 after years at Palantir, and it went through Y Combinator in Winter 2024. Y Combinator lists a team of 5. In a launch post of 23 June 2026 the company wrote that it is "already in production with Fortune 500 users"; it names no customer count. No revenue figure was found, so this story has grade E. The page shows what the company sells, its prices, and earlier analytics startups that failed against larger competitors. ## Pricing Upsolve AI prices in November 2026, from the company's pricing page (grade C) | Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit | |---|---|---|---| | Free | 0 | 0 | 2,000 one-time credits, about 200 questions | | Pro | 400 | 500 | 2,000 credits a month; US$0.05 per extra credit | | Team | 1,600 | 2,000 | 10,000 credits a month; embedding, row-level security, multi-tenant | | Enterprise | custom | custom | base fee plus credits, or own model plus surcharge; self-hosting | Plans are priced in credits: one credit is the simplest AI answer, a standard analytical question uses 5 to 10 credits, and the page says a query is capped at 20 credits. The pricing page states a 20% discount on the base fee for a yearly commitment and an overage price of US$0.04 per credit on yearly plans. Embedding in the customer's product starts with the Team plan. Enterprise allows the customer's own model (BYOM) with an Upsolve surcharge. ## How much revenue does Upsolve AI make? The numbers, graded A two-founder, YC-backed analytics company with public prices and no published revenue | Fact | Value | Grade | Note | Sources | |---|---|---|---|---| | Customers, June 2026 | "already in production with Fortune 500 users, 100+ person BI teams, and growth-stage companies" | C | Company's Y Combinator launch post, 23 June 2026. No customer count or names of the Fortune 500 users. | [5] | | Revenue of Upsolve AI | not published | C | No payment-processor profile, interview or filing with a revenue figure was found. See "What is not verified". | [1, 2, 5] | | Founded and team | founded 2023; team of 5 | C | Y Combinator company page, read 5 November 2026. The about page also says founded in 2023. | [1, 10] | | Founders | Ka Ling Wu (CEO) and Serguei Balanovich (CTO) | C | Y Combinator page and about page. Both worked at Palantir; the YC page says Serguei Balanovich spent seven years there and worked on Palantir HyperAuto. | [1, 10] | | AI at the first launch | "an AI-powered builder", "AI answering any additional data questions" | C | Launch post of 17 February 2024. Dashboards could also be built by point-and-click. | [2] | | Earliest dated customer story | Fiber AI, 5 January 2024 | C | Company's impact study on Fiber AI, another YC company with 2 to 10 people. It says Fiber cut two months from building customer analytics. | [9] | | Prices, November 2026 | Free, then US$500 and US$2,000 a month, Enterprise custom | C | Pricing page. US$400 and US$1,600 a month with a yearly commitment. | [7] | | AI models | GPT-4o shown in the product's trace view; Enterprise can bring its own model | C | The home page shows an agent trace with steps labelled "gpt-4o". The pricing page names Azure OpenAI and AWS Bedrock as examples of a customer's own model infrastructure. | [6, 7] | | Review site | 4.8 out of 5 on G2, "High Performer", Fall 2025 | C | As shown on the company's home page; the G2 page itself was not read. | [6] | | Security | SOC 2 Type II; self-hosting on Enterprise | C | Company's llms.txt summary and pricing page. | [8, 7] | | Time from launch to the June 2026 statement | 28 months | calc | Calculated: 17 February 2024 to 23 June 2026. | [2, 5] | How to read this table: a grade C value is the company's own statement, on its site or in its Y Combinator posts. A "calc" value was computed on this page. No value is grade A or B, and no revenue figure was found. ## When did Upsolve AI launch, and how did it grow? Timeline From a dashboard builder for SaaS products to an agent platform for data teams, in 28 months. Each entry carries its grade. - **2023** (grade C [1, 10]): Ka Ling Wu and Serguei Balanovich found Upsolve AI after working at Palantir. - **5 January 2024** (grade C [9]): First dated customer story: Fiber AI embeds Upsolve dashboards in its marketing product. - **17 February 2024** (grade C [2]): Launch post in the Y Combinator Winter 2024 batch: an AI-powered dashboard builder, embedded as React or Vue components, with AI answering users' data questions. "We are live with our customers now!" - **26 September 2024** (grade C [3]): Second launch: a personal workspace where end users build their own dashboards. Examples include an expense platform and a supply-chain platform with Fortune 500 customers. - **31 March 2025** (grade C [4]): Third launch, "GenBI": AI report generation, drill-down in natural language, automatic insights and alerts. - **23 June 2026** (grade C [5]): Fourth launch, Agent Studio: context, permissions, tests and tracing for analytics agents. The company says it is in production with Fortune 500 users. - **November 2026** (grade C [7, 6]): Credit-based pricing from free to US$2,000 a month on the pricing page; the site announces "14 launches in 14 days". ## Why did Upsolve AI work? ### Domain experience The founders say they built a data product at Palantir to more than 50 enterprise customers. They used that background in every launch post, and it is the kind of trust that large buyers ask for. ### One problem every SaaS product has Most B2B software must show its customers a dashboard. Upsolve sold a component that a SaaS team drops into its own product instead of building charts, logins and per-customer data rules itself. ### Follow the AI change Each launch moved further toward AI: an AI builder in 2024, AI reports and insights in 2025, and in 2026 a platform for data agents with tests and permissions that the model cannot override. ## How it was built These details come from the company's launch posts and its own pages. They are grade C. - Dashboards are embedded in the customer's product as React or Vue components, styled to the customer's design system. - Multi-tenant data rules work with the customer's existing login: an SDK in the login flow decides which data each user sees. - The agent layer uses large language models; the home page trace shows GPT-4o, and Enterprise customers can route to their own models. - It connects to Snowflake, BigQuery, Databricks, Redshift, Postgres and other databases, and answers in Slack, Teams, an embedded SDK or MCP clients. - Tests use golden query sets and LLM-as-judge scoring, an AI model that grades another model's answers. Sources: [2, 5, 6, 7, 8] ## Who tried this before Upsolve AI, and why did they fail? Three companies in our database of failed startups sold dashboards or analytics and lost to larger competitors or found too few buyers. The values below are from that database and are not independently verified. The failed startups come from our database of dead startups, retold here in our own words; Its entries are AI-assisted summaries of public sources, so they carry no evidence grade. ### SumAll (2011 to 2019; funding US$15M; cause: Competition) - Promise: One data dashboard for small businesses, without a data scientist. - Why it died: Analytics giants and free tools offered the same dashboard, and its price was hard to justify. - Record: ### Skysheet (2009 to 2012; funding US$200K; cause: Competition) - Promise: Cloud spreadsheets with live collaboration and charts. - Why it died: Google and Microsoft built the same tools into products people already used. - Record: ### Hiptype (2012 to 2015; funding US$500K; cause: No Market Need) - Promise: Reading analytics for e-book publishers. - Why it died: Too narrow a market, and publishers did not change their habits for analytics. - Record: ## What changed The three predecessors sold analytics as a separate product and lost to larger companies that added the same charts to tools people already had. Upsolve faces the same risk from BI vendors and data warehouses that add AI agents of their own. Its answer is to sell what a general model or a single warehouse does not: one agent across several databases, the customer's business definitions and permissions encoded beneath the model, tests before release, and a version that runs inside another company's product. ## Can one person copy Upsolve AI now? **Verdict: Hard for one person.** Enterprise data buyers want security audits, self-hosting and engineers on site, and the founders sell on years of Palantir experience. A solo founder could copy the small first version: an embeddable AI dashboard for one kind of SaaS product. 1. Pick one industry of SaaS products whose customers all ask for the same reports. 2. Ship an embeddable component that works with the product's existing login and data rules. 3. Let users ask questions in plain English, and test the answers against a set of known correct queries. 4. Publish dated customer stories and prices from the start. **What is still open.** Analytics agents for one data stack or one industry, where business definitions matter more than the model. ## What could go wrong for Upsolve AI? - **What is not verified.** No revenue figure for Upsolve AI was found. We looked for a payment-processor profile, a founder interview with figures, a funding announcement and a filing. The traction claims are the company's own: "in production with Fortune 500 users, 100+ person BI teams" (launch post, 23 June 2026), while its llms.txt summary says "BI teams of 60+ people". The founders' statement that their Palantir product made "eight figures" in revenue is about Palantir, not Upsolve. The company's location is not stated: YC lists none and its one job post is in London. A reader could look for a funding announcement or an interview with figures. - **Competition from BI vendors and warehouses.** Upsolve's own comparison posts name many rivals: Luzmo and Vizzly as alternatives to the embedded analytics vendor Embeddable (13 February 2025), Explo, and BI products that add agents, such as ThoughtSpot Spotter, Databricks Genie and Hex (llms.txt list of comparisons). Its post of 4 November 2026 says Hex Context Studio launched on 28 January 2026 and cites Forbes as reporting on 27 October 2026 that Hex raised US$100 million. These are a competitor's descriptions, not neutral reviews. - **Dependence on model providers.** The product's answers come from large language models made by others; the home page shows GPT-4o in its trace view. The June 2026 launch post says "GPT and Claude write fine SQL" and places Upsolve's value in context and governance, not in the model. Enterprise customers can bring their own model through services such as Azure OpenAI or AWS Bedrock, which lowers the dependence on one provider. - **The product changed direction.** The pitch moved from dashboards inside SaaS products (2024) to agents for internal data teams (2026). The llms.txt summary calls the two "equal use cases". Readers should check which one a given customer story describes. ## Sources 1. Y Combinator company page: Upsolve AI — https://www.ycombinator.com/companies/upsolve-ai (grade C). Read in full, 5 November 2026. Source for the founders, founding year, team size, batch and the job post location. 2. Y Combinator Launches, 17 February 2024: Upsolve AI - The new way to build and offer analytics to your customers — https://www.ycombinator.com/launches/KQ3-upsolve-ai-the-new-way-to-build-and-offer-analytics-to-your-customers (grade C). Read in full. Company's first launch post. Source for the AI builder, React and Vue components and the first ask. 3. Y Combinator Launches, 26 September 2024: Upsolve AI - Embedded Business Intelligence (BI) — https://www.ycombinator.com/launches/Lvs-upsolve-ai-embedded-business-intelligence-bi (grade C). Read in full. Source for the personal workspace feature and customer examples. 4. Y Combinator Launches, 31 March 2025: Upsolve AI - Embed GenBI in Your App — https://www.ycombinator.com/launches/NBR-upsolve-ai-embed-genbi-in-your-app (grade C). Read in full. Source for the AI report, drill-down and insight features. 5. Y Combinator Launches, 23 June 2026: Upsolve AI - Agent Studio for Data Team — https://www.ycombinator.com/launches/Qyr-upsolve-ai-agent-studio-for-data-team (grade C). Read in full. Source for Agent Studio, the Fortune 500 statement and the remark on GPT and Claude. 6. Upsolve AI home page, read 5 November 2026 — https://upsolve.ai/ (grade C). Read in full. Source for the GPT-4o trace, the G2 rating, the deployment surfaces and the launch-month banner. 7. Upsolve AI pricing page, read 5 November 2026 — https://upsolve.ai/pricing (grade C). Read in full. Source for the plans, credits and the bring-your-own-model option. 8. Upsolve AI llms.txt summary — https://upsolve.ai/llms.txt (grade C). Read in full. Company's own summary for AI clients. Source for the two use cases, databases, SOC 2 Type II and the BI team size statement. 9. Upsolve AI impact study, 5 January 2024: How Fiber AI uses Upsolve AI — https://upsolve.ai/impact-study/fiber (grade C). Read in full. Company's own customer story. 10. Upsolve AI about page — https://upsolve.ai/about (grade C). Read in full. Source for the founding year and the founders' roles. 11. Upsolve AI blog, 4 November 2026: Hex Context Studio vs Upsolve — https://upsolve.ai/blog/hex-context-studio-vs-upsolve (grade C). Read the sections on Hex AI, Context Studio and embedding. A comparison published by Upsolve. 12. Upsolve AI blog, 13 February 2025: We Tested 3 Embeddable Competitors for User Facing Analytics — https://upsolve.ai/blog/embeddable-competitors (grade C). Read the introduction and the list of tools. A comparison published by Upsolve. 13. Database of failed startups, local export — /method/#dead (grade C). Local export. AI-written summaries, not independently verified. Source for the failed predecessors. Contact: aiolum@livyden.com. Run Upsolve AI? Send corrections or a verified figure to aiolum@livyden.com (subject: Correction: Upsolve AI).