long-horizon agent contracts<\/a> across lending, healthcare, and three other verticals. The combined team will ship a platform called Horizon. This follows Sierra’s $950 million Series E in May 2025 at a $15 billion valuation, $150 million in ARR, and a FedRAMP High certification that opened the federal market.<\/p>\nWhat this means for your business<\/h2>\n
The architecture question Sierra is answering, whether an AI agent should complete a task in seconds or orchestrate work across days, is one that most enterprise IT shops haven’t formally answered yet. If your current agent deployments are scoped to single-session interactions like a support deflection or a form fill, you’re building on a different set of assumptions than what Sierra and Takeoff are selling. That gap matters most to CIOs who are mid-contract with a point solution vendor and haven’t yet locked in an architecture for the next layer.<\/p>\n
Takeoff’s traction is the detail worth sitting with. Eight figures in ARR from three people in seven months across five regulated industries suggests that outcome-based pricing, where the vendor charges for results produced rather than API calls consumed, is finding real buyers. That pricing model shifts risk toward the vendor in theory, but in practice it pushes the customer to instrument outcomes clearly before signing. Organizations that can’t define what “resolved” or “completed” means for a multi-week lending workflow will negotiate badly on these contracts, regardless of the platform’s technical merits.<\/p>\n
Sierra’s trajectory, from customer support tool in 2024 to Agent OS 2.0, FedRAMP High, a SoftBank Japan partnership, and now a long-horizon runtime acquisition, reads less like a product roadmap and more like a land-grab across every enterprise surface where agents can plausibly touch revenue. The vendors most exposed aren’t the model providers; they’re the workflow automation incumbents whose value proposition assumes a human closes the loop. If Horizon delivers on multi-week autonomous execution at Sierra’s current deployment speed (Nordstrom in five weeks, Cigna in eight), the renewal conversation for any RPA or BPM platform sitting in your stack gets harder to win on status quo alone.<\/p>\n
Concept deep-dive: Long-horizon agent runtime<\/h2>\n
Most AI agents today are stateless, meaning they handle one request, return a result, and forget the context. A long-horizon agent runtime is the infrastructure layer that gives an agent persistent memory, a task queue, and the ability to pause, wait for external events, and resume, much like a human project manager who picks up a loan file on Tuesday after waiting for a credit check to return on Monday. The business case is any workflow where completion takes longer than a single conversation.<\/p>\n
Based on reporting from Sierra Acquires Takeoff For Long-Horizon AI Agents<\/a>, originally published 2026-07-24 10:46:00.<\/em><\/p>\n","protected":false},"excerpt":{"rendered":"Share with your CIO Sierra is betting that the next enterprise AI moat isn’t a better model, it’s a runtime that keeps agents working across days or weeks rather than single turns. The company acquired Takeoff, a three-person startup that reached nearly eight figures in ARR in seven months by selling outcome-priced, long-horizon agent contracts […]<\/p>\n","protected":false},"author":1,"featured_media":6694,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[148],"tags":[185],"tmauthors":[],"class_list":["post-6693","post","type-post","status-publish","format-standard","has-post-thumbnail","category-ai-marketing","tag-cio"],"_links":{"self":[{"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/posts\/6693","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/comments?post=6693"}],"version-history":[{"count":0,"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/posts\/6693\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/media\/6694"}],"wp:attachment":[{"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/media?parent=6693"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/categories?post=6693"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/tags?post=6693"},{"taxonomy":"tmauthors","embeddable":true,"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/tmauthors?post=6693"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}