Sable has raised $45 million for Aidan, an agent designed to run live product demos and onboarding. The company names Notion and Decagon as production customers, but has not published the performance or cost data needed to show that the system can replace customer-facing work at scale.
Sable has raised $45 million to build Aidan, an agent meant to conduct live product demonstrations and onboarding without a human leading the call. The money is clear. The evidence that Aidan can economically absorb customer-facing work is not: the public launch materials contain no accuracy, human-takeover, conversion, retention or cost-per-session data.
Sable said in its press release that Sequoia Capital and 8VC led a $45 million financing round, with BoxGroup, SV Angel, Valor Atreides AI Fund, Sabrina Hahn and Evan Hahn participating. Sequoia partner Shaun Maguire and 8VC founder Joe Lonsdale are joining the board.
The announcements do not break out the stages or amounts. Sequoia says in its investment account that it led Sable's seed round and co-led its Series A, while Sable describes $45 million as both a financing round and its total backing. The distinction matters when comparing the sum with a single conventional venture round.
Demo software has attracted substantial financing before. Demostack raised a $34 million Series B in 2022 and had raised $51.5 million in total at that point, according to a contemporaneous report. Sable's $45 million spans earlier stages, so the figures are not like-for-like, but the comparison limits the idea that investors have only now discovered the value of automating product demonstrations.
The production and demand figures also originate with financially interested parties. Sable names Notion, Decagon and unidentified public companies as production users in its launch post. Sequoia and 8VC say more than 150 companies are on a waitlist. Neither named customer is quoted in the retained sources, and a waitlist does not show how many deployments will reach production, expand or renew.
Sable is less than a year old. It was founded by Harvard friends Nim Ravid, Leon Chen, Linda He and Itamar Rocha; the company and its investors say the team worked on post-training, reinforcement learning and multimodality, with prior experience at companies including SpaceX, Google, Meta and Together AI. A report on the financing identified HubSpot co-founders Brian Halligan and Dharmesh Shah, Valor's Antonio Gracias and Cognition chief executive Scott Wu among the angel investors.
Sable's claimed product distinction is simultaneous interaction. The company says Aidan can perceive a shared screen, speak with a buyer and operate an interface in real time rather than wait for a text prompt. It calls that combination of voice, vision, video and browser control “interactive intelligence.”
The session runs in a LiveBox, a virtual machine containing a functioning product instance. An account from investor 8VC says Aidan can respond to what a user does, cede control of the interface and resume the demonstration. That is more adaptive than a recorded tour, but the description and the claim that the environment contains the “real product” come from an investor backing the company, not an independent product test.
Sable also builds a customer-specific knowledge layer called the Brain. The company says it transforms sales-call recordings, interviews with strong performers, documentation and marketing material into a context graph, then uses subsequent interactions to find knowledge gaps and refine responses.
Sable calls deployment integration-free, yet its own release says mapping an organization's expertise takes weeks. Those statements can both be true if “integration” means no conventional product engineering, but customers still have to supply, structure and govern the material that determines what Aidan tells buyers. The retained disclosures do not identify Aidan's underlying models, compute requirements or the work required to maintain a usable LiveBox as a customer's product changes.
Demostack, Reprise and Walnut were already offering customizable product demonstrations and tours by 2022. Those products cloned or tailored demo environments that sales teams could use live or share with buyers. Aidan's claimed advance is not the invention of automated product explanation; it is an agent that speaks and reacts while operating the product during the session.
The competitive set extends beyond demo platforms. A report comparing the product pointed to BCG's Jamie, an AI sales agent used for internal coaching, while Notion—the customer most prominently cited by Sable—is developing agents of its own. Buyers can also combine human-led calls with documentation, prerecorded tours and chat systems rather than buy one system for the entire journey.
That matters because Sable's labor claim is much broader than its demonstrated interface. The company pitches Aidan across qualification, demos, solutions engineering and customer-success onboarding. Those functions involve repeated explanation, but also discovery, commercial judgment, relationship management and exception handling. The sources show that Aidan is being positioned across all four areas; they do not show that it has replaced any one of them end to end.
8VC says Notion expanded Aidan from setup sessions into custom-agent activation, product demonstrations and partner enablement in dozens of languages, with pricing tied to outcomes. That is the most specific account of customer use and commercial terms in the retained record. It still does not define the outcome, the price, session volume, performance change or who supplied the comparison baseline.
Investor descriptions go further, suggesting expert interaction can be delivered at the marginal cost of the product and in an unlimited number of sessions. No published figures support those economics. Sable says the Brain alone takes weeks to map an organization's expertise, but does not quantify that deployment work, the resources required to run a session or the frequency of human escalation. Without those inputs, there is no basis for comparing Aidan with a sales engineer, a conventional demo platform or a mixed human-and-software process.
Nor is multilingual fluency a proxy for business performance. Maguire told the financing report that he saw Aidan switch among English, Mandarin and Spanish during a demonstration. That establishes what an investor says he observed in one demo, not accuracy across languages or success in unscripted customer sessions.
Sable chief executive Ravid says Aidan can lead interactions end to end without a human in the loop. The company also says customers can build, monitor and continuously improve the agent. Its materials do not explain when a human is alerted, what actions require approval, how errors are corrected or who can change the Brain.
Evidence from another computer-using agent cannot measure Aidan. It can clarify what buyers should demand. OpenAI's Operator system card identifies mistakes and on-screen prompt injection as risks, recommends human oversight, and says its computer-use model performs best in browser-sandboxed settings. OpenAI added confirmations, monitoring and supervision requirements to its own product. Sable's LiveBox supplies isolation, but the public record does not describe comparable permission boundaries, confirmation steps or defenses against misleading on-screen instructions.
The knowledge layer creates a separate governance problem. A system trained on successful calls may preserve expertise, but “successful” can depend on the account, market or product version. Sable says Aidan learns from interaction patterns; it has not publicly described how a customer inspects, approves, rolls back or segments those changes.
Sable now has the capital to move companies from its waitlist into production. The unresolved decision for buyers is whether live, agent-led sessions improve commercial outcomes enough to justify their cost and control requirements.
That case needs comparable evidence: completion and answer-accuracy rates; latency and human-takeover frequency; conversion, activation or onboarding changes against a defined baseline; renewal and expansion behavior; and total cost per successful session. Buyers also need retention policies, permission boundaries and a clear account of who can approve changes to the Brain.
Until named customers provide that evidence, the round establishes investor conviction and company-reported deployments—not that one AI system can reliably replace the people who sell, explain and onboard enterprise software.
OpenAI has put conversations and Projects back in its redesigned ChatGPT desktop app and enabled cloud Work threads to move across devices, correcting the launch's biggest usability failures without merging local Work or Codex histories.
Meta reportedly plans to put departing AWS compute executive Dave Brown to work on its data-center buildout, adding hyperscale operating experience while leaving any customer-facing cloud business conditional and undefined.
China has paired a five-year AI training offer for developing countries with cooperation centers, a weather-warning rollout and a new 29-country organization. The package gives Beijing a platform for influence, but no budget, selection rules or delivery timetable has been published.
RoboTTT extends a robot policy’s visuomotor context to 8,000 timesteps and raises its average rubric score from 42% to 79%, but the evidence comes from three author-run assembly tasks and the longest task was completed in only two of ten trials.
Zhipu reportedly reached $1 billion in annual recurring revenue in July, roughly four times a March estimate, but the unconfirmed run rate is not annual sales and still sits far ahead of recognized cloud revenue while margins remain thin.
Nebius has arranged its first senior secured facility against an operating GPU deployment and one customer’s cash flows. The deal adds project-level debt to its expansion toolkit, but an unnamed customer and a larger rival financing limit what it proves about the rest of Nebius’s backlog.
Cursor says automatically running a repository-root git.exe on Windows does not meet its criteria for patching, while the researcher calls it an untrusted-search-path defect and separate research shows the same weakness across several competing AI coding tools.
Huawei publicly displayed a 16-cabinet Atlas 950 configuration rated at 1 EFLOPS in FP8, providing tangible evidence of its system-scale AI strategy while leaving price, power use and sustained workload performance undisclosed ahead of the full system's planned fourth-quarter release.
CIA Director John Ratcliffe said US intelligence is consistent with an estimate that Russian recruits last 20 to 30 minutes on Ukraine’s battlefield, but the public trail leads to an unsourced claim about assault troops and does not establish a representative average.
UK testing places leading open-weight models four to seven months behind selected closed-model cyber results, yet longer attack chains, U.S. benchmarks and mixed cost comparisons show why that interval is a warning signal rather than a universal capability clock.
TSMC reached the top of its second-quarter revenue guidance and raised its 2026 outlook, but a one-off investment gain boosted profit growth while the company committed more capital to 2-nanometer production, advanced packaging and an undated Arizona expansion.
Moonshot AI's Kimi K3 added to a global technology selloff with near-frontier performance, but unreleased weights, mixed cost comparisons and a recommended 64-accelerator deployment leave its effect on chip demand unresolved.
Moonshot AI has made Kimi K3 available through its apps and API, pairing a 2.8-trillion-parameter architecture with early frontier-level results, but the model's open-weight claim cannot be tested until its weights and technical report arrive.
Two binding EU decisions require Google to give rival AI services comparable access to 11 Android features and offer eligible search competitors a restricted, anonymized dataset, but phased deadlines, certification, pricing and privacy safeguards leave the competitive effect unproven.
Databricks has signed a term sheet for a Coatue-led financing at a $188 billion valuation, while unidentified sources put the round at $3 billion. The proposed capital would deepen its push into AI governance, data agents and operational databases, but the transaction remains open and the company supplied no new operating figures.
Netflix says generative AI workflows were used on roughly 300 titles in 2026, but its only quantified example tied to that disclosure covers 17 minutes and does not establish how much finished material, spending or labor changed across the slate.
Gemini 3.5 Pro missed its expected June rollout. An anonymous-source account says a late-June data change fell short of Google's coding goals, but Google has confirmed only partner testing—not the reported cause, a new date, or public results and pricing.
Google is bringing Gemini Omni editing and a reusable face-and-voice avatar to Vids, but the sharper distinction is account-level identity across Vids and Gemini rather than a new category of AI video; Vids already offered Veo generation and customizable avatars, while specialist rivals already sell digital presenters.
Intel plans to adopt Gemini Enterprise across engineering, supply-chain and corporate work while extending chip-simulation capacity into Google Cloud, but the companies supplied no rollout schedule, contract value or performance baseline.
The Navy has approved an immediate department-wide framework for turning data into operational effects, but the public rollout leaves budgets, owners, deadlines, performance measures and assurance rules to a promised implementation roadmap.