Two widely known AI entrepreneurs, Ashutosh Garg and Varun Kacholia, are back with a new company, Viven, and a $35 million seed round to go along with it. Eightfold’s co-founders are creating AI-powered digital twins to allow workers to stop and query a colleague’s knowledge when that person is offline, taking a break, or simply heads-down. The funding comes from Khosla Ventures, Foundation Capital, FPV Ventures and other backers hoping to see a path toward turning individual expertise into a perpetually available corporate asset.
Viven’s proposition is simple, but it is ambitious. The startup trains a bespoke large language model, honed to match the tone and style of a given employee — one that can read from that person’s emails, chat threads, documents, and project history, but never store or transfer them between devices. Colleagues can then pose work-related questions and receive instant, context-aware replies that even reference source materials.
Why digital twins are taking off in modern workplaces
And in a globally distributed team, the cost of waiting for the right person is very real. Studies by both McKinsey and IDC have long estimated that knowledge workers lose from a fifth to a third of their time every day looking for information — or looking for colleagues who know where the information is. That drag is magnified during handoffs across time zones and onboarding, when new hires unblock themselves by pinging experts over and over again.
By making personal context a service that can be queried, Viven seeks to transform slow, interrupt-driven flow into on-demand answers. Picture a sales engineer dragging the precise email snippet one product manager used to send a client, or if a finance analyst can float “why we accrued $500K last quarter” without hauling in the owner, too. It’s not about replacing people, but smoothing over logjams caused by availability.
How Viven thinks about privacy and context
Prominent in Viven’s design is what the company refers to as pairwise context and privacy. Instead of a universal database, each twin reviews the disparities based on who is asking and what they have access to in terms of content, people, and project borders. In practical terms, the twin can respond “What did we decide about that Q3 pricing test?” for a project teammate, but refuse or redact details for someone not in the loop.
To further promote norms, Viven only shares the entire query history with the owner of each twin. That visibility provides a kind of social check on fishing expeditions or inappropriate requests. The system also separates out personal context from professional files, and weeds these out, as topics of overlap are likely to be concerned with private matters. Under the covers, this looks like a combination of retrieval-augmented generation scoped to a user’s universe, access control lists enforced at inference time, and built-in auditability from the get-go.
Early customers and the investor thesis behind Viven
Viven says it’s already in use at several enterprises like Genpact and Eightfold itself. The two founders are staying at Eightfold and working on Viven, no doubt a signal that the companies believe their products are not competitive with each other within the broader enterprise AI stack. With regard to the latter, investors refer to the partnership’s history in managing large-scale machine learning, and the distinction of their person-centric model as reasons for confidence.
Vinod Khosla is an outspoken proponent of AI copilots for narrow use cases and is now funding Viven through his VC firm, Khosla Ventures. Foundation Capital and FPV Ventures add to its enterprise credibility. The bet is that fine-grained, relationship-aware sharing becomes a lasting moat when trying to adopt safely within organizations where off-the-shelf assistants come up short on permissions and provenance.
Crowded field or an open lane for individual twins
There is no lack of enterprise AI tools trying to help answer questions at work. Microsoft Copilot, Google’s Gemini tools, Anthropic’s Claude for Work, and the enterprise offerings of OpenAI all deliver company-aware assistants, ranging in levels of customization. Companies like Slack, Notion and Atlassian have introduced retrieval features that dig through chats and documents. Where Viven claims it’s unique is in its organizational unit: the individual digital twin, as opposed to a company-wide search agent.
That framing matters. A twin knows the sublimated history of a teammate’s choices and drafts — not just the final files. It also makes it possible to customize guardrails: what you can ask “my manager’s twin” might not be the same as what you can ask “a peer’s twin,” even if both documents overlap significantly. Whether that granularity scales without friction will be a critical test of product-market fit.
Risks, governance, and what happens next for Viven
No enterprise AI endeavor escapes the glare on security and compliance. Potential customers will need confidence around where data resides, control of consent on ingestion into personal mailboxes, and compliance with things like SOC 2, ISO 27001, and GDPR. Accuracy and attribution are also important; teams will want answers that are rooted in auditable sources with clear fallbacks when the twin is uncertain.
If Viven can demonstrate that pairwise permissions, transparent query logs, and high-quality retrieval are worth the wait for a colleague to co-work with an AI twin, the label of AI coworker twins may move from gimmick to game-changer.
For now, the fresh capital provides runway for the company to flesh out its privacy tech, bring that deeper integration beyond email and docs, and demonstrate measurable time savings with its early adopters.
