# Startup Spotlight: Seals AI automates the wholesale order desk with Titanio > Source: > Published: 2026-09-22 18:45:42+00:00 # Startup Spotlight: Seals AI automates the wholesale order desk with Titanio **Luis Mario Garcia and his Orchata co-founders are applying experience from an operationally difficult delivery business to calls, quotes, orders and ERP entry.** By [Ryan Merket](https://runtimewire.com/author/ryan-merket) ยท Published Primary source: [Y Combinator](https://www.ycombinator.com/companies/seals-ai) ## Why it matters Seals AI is a test of whether vertical AI founders with genuine operating scars can beat better-funded rivals by understanding the order desk, warehouse and ERP as one workflow. [Luis Mario Garcia (@luismariogm)](https://x.com/luismariogm?ref=runtimewire) is building Seals AI around a problem his team encountered while running Orchata, their previous grocery-delivery company. [Y Combinator says](https://www.ycombinator.com/companies/seals-ai?ref=runtimewire) they placed thousands of purchase orders with wholesalers, where quoting, ordering, payment collection, ERP entry and support were handled through phone calls and email. Seals AI's current product, [Titanio](https://www.titanio.ai/?ref=runtimewire), is an AI sales and ordering system for distributors and wholesalers. Titanio plans field-sales routes, prepares account briefs, captures visit notes, handles customer conversations and converts purchase orders, photographs and calls into orders inside existing business software. That scope is wider than the "AI employees" pitch Seals AI used when it entered Y Combinator's Summer 2024 batch. The [original Y Combinator launch post](https://www.ycombinator.com/launches/LqH-seals-ai-ai-employees-for-wholesalers-distributors?ref=runtimewire) concentrated on automating quoting, order taking, payment collection, purchase orders and ERP entry. Titanio now presents those workflows as parts of a broader sales system spanning field representatives, contact centers and digital commerce. The founders arrived at wholesale software after learning a harsher business themselves. Garcia, [Heber Uriegas (@heberuriegas)](https://x.com/heberuriegas?ref=runtimewire), Javier Gonzalez and Fernando Huerta previously worked together at Orchata, a YC-backed grocery-delivery operation in Mexico. Uriegas's [personal site identifies him as Heber Uriegas](https://heberuriegas.com/?ref=runtimewire) and as Seals AI's co-founder and CTO, while the [YC directory lists him as Heber Garza](https://www.ycombinator.com/companies/seals-ai?ref=runtimewire) but links to social profiles using the Uriegas name. This article uses the name on his own site and X account. [YC says Orchata generated $6.3 million in revenue, served 30,000 customers and delivered 300,000 grocery orders](https://www.ycombinator.com/companies/seals-ai?ref=runtimewire). The founders also placed thousands of purchase orders with wholesalers while operating Orchata. Those orders supplied the product thesis for Seals AI. In their YC launch post, the founders said they experienced wholesalers handling quotes, orders, payments, ERP entry and support through phone calls and email. They had been the customer on the other end of the workflow. ### The failure supplied the wedge Garcia's background explains why Seals AI is tackling an operational category where product demonstrations matter less than whether the software survives a busy order desk. According to [his YC biography](https://www.ycombinator.com/companies/seals-ai?ref=runtimewire), Garcia designed Orchata's driver, consumer and warehouse applications while also acquiring customers, operating warehouses, opening cities, recruiting engineers and fundraising. He says he personally delivered 1,000 orders to learn from customers. Orchata produced real revenue without becoming a sustainable business. In an [April 8th, 2025 interview published by Tecnologico de Monterrey](https://conecta.tec.mx/es/noticias/nacional/emprendedores/el-mexicano-que-con-ia-apoya-mayoristas-desde-silicon-valley?ref=runtimewire), Huerta said the founders eventually closed Orchata after concluding that the model would not be profitable. He described years spent trying to make rapid grocery delivery work, including a period when investor interest in the category collapsed. The founders carried Orchata's operating knowledge into Seals AI rather than abandoning the supply chain entirely. That continuity is Seals AI's strongest early advantage. Garcia knows distributor purchasing from the buyer's side, while Huerta brought a longer record in telecommunications and enterprise sales. [YC identifies Garcia as Seals AI's CEO and credits Uriegas with 18 years of coding experience and previous engineering leadership at a YC-backed startup](https://www.ycombinator.com/companies/seals-ai?ref=runtimewire). Uriegas's [personal site](https://heberuriegas.com/?ref=runtimewire) lists him as Seals AI's co-founder and CTO, following roles as Orchata's head of engineering and Prysmex's CTO. [YC says Gonzalez began programming at 12 and independently built an application that reached 500,000 users](https://www.ycombinator.com/companies/seals-ai?ref=runtimewire) before working on four larger software products. Huerta supplies a different type of operating experience. [YC says he helped launch AT&T in Mexico and grew the operation from zero to $120 million in annual recurring revenue as marketing and sales director](https://www.ycombinator.com/companies/seals-ai?ref=runtimewire). YC also credits him with helping make Latin America's first internet connection in 1989. Tecnologico de Monterrey reports that he studied electronic systems engineering there and later spent 15 years at Mexican telecommunications provider Alestra. The result is an unusually seasoned founding group for an early vertical AI product. Huerta had sold and launched technology inside much larger organizations, while Garcia, Uriegas and Gonzalez brought product, engineering and operational experience from Orchata. Seals AI is their attempt to turn those experiences into a repeatable software layer. In the October 19th, 2024 post, Uriegas summarized the founders' operating style: "People often don't expect impressive results from simple ideas." Wholesale order entry looks simple until a system must recognize a customer's language, select the correct products and prices, verify inventory and write clean data into an ERP without delaying the sale. ### Titanio has moved beyond order entry Titanio currently divides its product into three areas: field sales, contact-center automation and AI commerce. For field representatives, Titanio says it can build daily routes, account for drive time and traffic, prepare briefs before visits and turn recorded agreements into CRM notes and follow-up tasks. This part of Titanio targets the work surrounding an in-person sale, including deciding which customer to visit and remembering what happened after the representative leaves. For contact-center work, Titanio handles calls, texts, emails and WhatsApp messages. Titanio says its software can respond autonomously when it has enough information, draft replies for employees to approve and convert customer requests into quotes or orders. A customer can submit a traditional purchase order, send a photograph of a product list or describe an order by phone. Titanio's commerce product gives distributors a branded ordering application and chat interface. Customers can use their account pricing and order history to reorder common products. Titanio also says it can identify accounts that have stopped purchasing and initiate follow-ups. The product is designed to sit on existing systems instead of demanding an immediate ERP replacement. Titanio lists integrations with Epicor, SAP, Microsoft Dynamics, NetSuite, Infor, Salesforce, Zoho, Zendesk and QuickBooks. That approach reduces the size of the initial technology decision for a distributor. It also makes integration quality central to whether Titanio works in production. [YC lists the San Francisco business as Seals AI and directs its website link to Titanio](https://www.ycombinator.com/companies/seals-ai?ref=runtimewire). Garcia's public profiles continue to use Seals AI, while the product website puts Titanio at the front. The clearest reading is that Seals AI is the business identity and Titanio is the customer-facing product, although Seals AI has not labeled the distinction on its public pages. ### The difficult work begins after the demo An AI system can extract line items from a clean purchase order during a demonstration. Distribution introduces the cases that determine whether a deployment saves labor or creates another review queue. The same item may have customer-specific prices, several packaging units or an internal abbreviation understood only by an experienced representative. Stock changes. Substitutions require judgment. A caller may mix product questions, delivery requests and billing problems in one conversation. Titanio's promise requires the software to connect those messages to account history, catalogs, inventory and ERP records while preserving a path for human approval. Titanio's own product copy acknowledges that division of labor. The AI answers what it can and prepares the remaining conversations for an employee to review. That is a more credible deployment model than pretending every interaction can immediately become autonomous. Titanio markets a quote time of 60 seconds compared with 48 hours without its software, along with round-the-clock ordering and fully automated dispatch. Those figures are product marketing, rather than published measurements across named deployments. Seals AI's larger market claims deserve the same treatment. YC's profile says the United States has 700,000 wholesalers and frames the opportunity as converting $100 billion in payroll expense into software spending. Seals AI presents those numbers as its market thesis. Garcia's recent public posts show Seals AI concentrating on field deployment and enterprise sales. On [his LinkedIn profile](https://www.linkedin.com/in/luismariogarcia?ref=runtimewire), he has written about visiting distributor markets around San Diego and Tijuana, meeting large retail operators and learning customer operations in person. The approach resembles his Orchata habit of delivering orders himself: Seals AI's founders are again going to the operational floor rather than selling the category from a distance. ### Wholesale AI has become a funded category Seals AI is entering a market where several competitors have already raised larger rounds and published customer figures. Canals has expanded from distributor order entry into accounts receivable, purchasing, receiving and customer inquiries. In April 2026, [Canals said](https://www.canals.ai/blog/canals-launches-end-to-end-operating-ai-suite-for-distributors?ref=runtimewire) that over 100 distributors used its platform. On May 28th, 2026, Canals [announced $35 million in funding](https://www.globenewswire.com/news-release/2026/05/28/3302917/0/en/Canals-Raises-35M-to-Eliminate-Friction-from-the-Industrial-Supply-Chain.html?ref=runtimewire) led by Base10 Partners and said it had processed eight million sales orders and $5 billion in payables. Didero is attacking adjacent procurement work. On February 12th, 2026, [Didero announced a $30 million Series A](https://www.didero.ai/blog/series-a-announcement?ref=runtimewire) co-led by Chemistry and Headline, with Microsoft's M12 participating. Its agents manage supplier communication, order tracking and exceptions inside existing ERP systems and email workflows. Digit is pursuing a deeper replacement strategy. Digit describes its product as an AI-native operating system combining inventory, manufacturing, material planning and warehouse management. On March 13th, 2026, investor HPA [announced Digit's $3 million seed round](https://hpa.vc/digit-follow-on/?ref=runtimewire), bringing its reported funding to $6.3 million. Titanio's breadth places it between focused order-entry tools and full operating systems. Seals AI wants to influence the complete commercial loop: choosing accounts to visit, capturing demand, generating quotes, entering orders, answering service questions and bringing customers back for another purchase. That breadth can help Seals AI expand inside a distributor after winning one workflow. It also creates a demanding product roadmap. Each added function introduces another data source, business rule and opportunity for an AI-generated mistake to reach a customer or ERP record. ### Experience is the pitch; deployment will be the proof Y Combinator is Seals AI's publicly verified institutional backer. Angel investor [Ivan Montoya](https://www.linkedin.com/posts/ivanmontoya1_i-am-very-excited-to-back-luis-mario-garcia-activity-7242557358822858753-nlPr?ref=runtimewire) has also said he invested after previously backing Orchata. Montoya cited the founders' years of working together and the discipline they developed inside an operationally difficult business. Garcia has separately identified Nowports founder [Alfonso de los Rios as an investor](https://es.linkedin.com/posts/luismariogarcia_best-mentors-are-founders-best-investors-activity-7219061713185243136-R-fW?ref=runtimewire). That investor case rests heavily on founder-market fit. Seals AI's public materials provide extensive detail about the founders and Titanio's capabilities, while the measurable case for Titanio remains company-supplied. Canals has published customer and transaction figures. Seals AI is still asking the market to place greater weight on the founders' history and their understanding of distributor work. The wager is reasonable. Vertical AI products gain an edge from knowing which workflow to automate, where employees need control and how software must fit into an existing operation. Garcia, Uriegas, Gonzalez and Huerta acquired that knowledge through a previous business that generated millions of dollars, processed hundreds of thousands of deliveries and still failed to become profitable. Seals AI is turning that failure into a focused second attempt. Titanio will be judged by whether distributors trust it with live prices, customer relationships and orders. The founders already understand the cost of getting operations wrong. That experience gives them a useful starting point in a category where polished AI output is easy and dependable execution is the product.