Egoist Machines
AI Passport enables users to carry personal context and preferences across AI assistants with granular control.
Website: https://ego.ist/
Cover Block
Publicly reported
| Field | Detail |
|---|---|
| Name | Egoist Machines |
| Tagline | AI Passport enables users to carry personal context and preferences across AI assistants with granular control. [Y Combinator] |
| Headquarters | San Francisco, USA [Y Combinator] |
| Founded | 2026 [Crunchbase] |
| Stage | Seed [Yahoo Finance, October 2026] |
| Business model | B2C [Y Combinator] |
| Industry | Deeptech |
| Technology | AI / Machine Learning |
| Geography | North America |
| Growth profile | Venture Scale |
| Founding team | Co-Founders (2), Erin Meryl McGurk and Dr. David Khachaturov [Y Combinator, October 2026] |
| Funding label | Seed |
| Total disclosed | ~$3,000,000 [Yahoo Finance, October 2026] |
Links
Publicly reported
- Website: https://www.ycombinator.com/companies/egoist-machines
- LinkedIn: https://www.linkedin.com/company/egoistmachines
Summary and Signal
PUBLIC Egoist Machines is building a consumer AI identity layer that lets users carry approved personal context across assistants, a timely bet as consumers increasingly interact with multiple models and apps that do not share memory by default [Y Combinator] [Cambridge St Catharine’s College, August 2026] [Yahoo Finance, October 2026]. Founded in 2026 in San Francisco, the company was co-founded by Erin Meryl McGurk and Dr. David Khachaturov, with St Catharine’s College stating Khachaturov co-founded the business in May 2026 and Y Combinator listing Egoist Machines in its portfolio the same year [Cambridge St Catharine’s College, August 2026] [Y Combinator].
The core product, AI Passport, is described as a permission-based context and identity layer that can work across assistants including ChatGPT, Claude, Gemini, and Muse; the stated differentiation is user control over what data is shared, with which application, for how long, and with a receipt when that information is used [Pulse 2, October 2026] [Cambridge St Catharine’s College, August 2026] [Henry Li - A* AI | LinkedIn, 2026]. That framing suggests the company is trying to sit above individual model vendors rather than compete as another assistant, with distribution split between direct consumer adoption and integrations into existing consumer applications [Y Combinator] [The Founder’s Pack, 2026].
The team is unusually strong on technical credibility for a two-person company: Khachaturov holds a PhD in Computer Science from the University of Cambridge, and public profiles tie his work to machine learning security and to PaperBench-related research at OpenAI [Y Combinator, October 2026] [Dr David Khachaturov | Department of Computer Science and Technology, 2026] [Dr David Khachaturov | St Catharine's College, Cambridge, 2026]. McGurk’s public profile is less technical but points to audience-building and media experience, including Girl Economics and a podcast, which may matter if consumer trust and distribution become the gating factors for adoption [The Heard, 2026] [Spotify] [Erin Meryl McGurk - Girl Economics | LinkedIn, 2026].
On financing, Egoist Machines announced a $3 million seed round in October 2026 with participation from Y Combinator, Goodwater Capital, a European family office, and scientists from DeepMind and Meta, while the business model is still best read as early-stage consumer infrastructure rather than a disclosed monetization engine [Yahoo Finance, October 2026] [Pulse 2, October 2026] [The SaaS News, October 2026]. The main public traction signal is company-reported: more than 10 B2B2C pilot programs, each involving organizations with more than 1 million users, but no customers, commercial terms, or production deployments have been named publicly [Pulse 2, October 2026].
Over the next 12 to 18 months, the practical questions are whether AI Passport can become a trusted cross-app standard, whether those pilots convert into visible distribution, and whether the company can prove a durable revenue model without losing the user-control premise that defines the product [Y Combinator] [Cambridge St Catharine’s College, August 2026] [Pulse 2, October 2026].
No independent source found -- This section relies materially on company, accelerator, and reprinted company claims, with limited independent public corroboration beyond founder credentials and financing coverage.
Taxonomy Snapshot
| Axis | Value |
|---|---|
| Stage | Seed |
| Business Model | B2C |
| Industry / Vertical | Deeptech |
| Technology Type | AI / Machine Learning |
| Geography | North America |
| Growth Profile | Venture Scale |
| Founding Team | Co-Founders (2) |
| Funding | Seed, total disclosed approximately $3,000,000 |
Company Overview
PUBLIC
Egoist Machines appears to be a newly formed consumer AI infrastructure company based in San Francisco, with a founding year of 2026 and a two-person founding team listed on public startup directories [Crunchbase] [Y Combinator]. The public record is still thin, which is typical at this stage, but the core through-line is consistent across the company profile materials: the business is organized around making a user's personal AI context portable across applications rather than recreating preferences and history inside each tool separately [Y Combinator] [Crunchbase].
The public timeline is short and mostly anchored to 2026. Y Combinator lists Egoist Machines as a San Francisco company in its portfolio and company directory, while Crunchbase records the company under the same name with seed-stage financing disclosed in 2026 [Y Combinator] [Crunchbase]. The clearest public milestone is the October 2026 announcement of a $3 million seed round, which established the company's first visible financing event and coincided with the launch of AI Passport as its initial product surface [Crunchbase] [Y Combinator, October 2026].
One source, partially checked -- This section relies primarily on Crunchbase and Y Combinator company materials, with only partial cross-corroboration on founding and milestone details.
The Product and the Stack
MIXED
The product story is straightforward, even if the implementation details are still lightly disclosed. Egoist Machines describes AI Passport as a permission-based identity and context layer that lets a user carry personal context, preferences, plans, history, and goals across AI assistants and consumer applications, including ChatGPT, Claude, Gemini, and Muse [Yahoo Finance, October 2026] [Pulse 2, October 2026] [Cambridge St Catharine’s College, August 2026]. In the company’s own framing, this is meant to reduce repeated re-entry of the same personal information across AI products while keeping the user, rather than the application, in control of what is shared [Y Combinator] [The Founder's Pack, 2026].
The control model is the core technical claim. Public descriptions say users can decide which applications can access which parts of their context, for how long, and receive a receipt when information is used, with access designed to expire rather than persist indefinitely [Cambridge St Catharine’s College, August 2026] [Henry Li - A* AI | LinkedIn, 2026]. Another public analogy compares AI Passport to a login layer that passes only the specific data requested, such as preferences, budget, or purchase history, instead of transferring a full profile by default [Henry Li - A* AI | LinkedIn, 2026]. What remains unverified from public materials is the deeper architecture: the sources do not establish storage design, model-layer integrations, cryptographic approach, API standard, or whether compatibility with named assistants reflects live integrations, target surfaces, or launch-day interoperability claims [Yahoo Finance, October 2026].
Distribution appears to be split between direct consumer surfaces and integrations into existing consumer applications, which suggests the company is trying to solve both user adoption and partner distribution at the same time [Y Combinator]. That may be strategically sensible for a portable identity layer, but the public evidence here is still early-stage: the company has disclosed more than 10 B2B2C pilot programs involving organizations with more than 1 million users each, according to launch coverage, yet no verified public demo materials in the cited sources identify named partners, implementation depth, or production usage [Pulse 2, October 2026]. The result is a product thesis that is legible and timely, with technical ambition centered on consented portability, but with most of the hard proof points still resting on company-linked descriptions rather than independent validation.
No independent source found -- Core product claims rely primarily on company materials and company-linked launch coverage, with partial corroboration from Cambridge St Catharine’s College.
The Market They Are Entering
Publicly reported The market matters now because consumer AI has expanded faster than the systems that let users carry identity, preferences, and history across products with explicit consent, leaving a coordination gap that Egoist Machines is trying to fill [Y Combinator, October 2026] [Cambridge St Catharine’s College, August 2026].
The available public record does not provide a cited TAM, SAM, or SOM for Egoist Machines, and that omission is material. What can be said from public sources is narrower: the company is positioning AI Passport as a portability layer for personal context across assistants such as ChatGPT, Claude, Gemini, and Muse, with consumers as the initial user and consumer applications as the integration point [Pulse 2, October 2026] [Yahoo Finance, October 2026] [Y Combinator]. That places the startup at the intersection of consumer AI applications, identity and permissions infrastructure, and privacy-oriented data portability, but without a third-party market model it is too early to attach a firm addressable market figure.
Demand drivers are easier to observe than market size. The core user problem described across sources is fragmentation: people increasingly interact with multiple AI products, yet their preferences, goals, and prior context do not automatically follow them from one assistant to another [Cambridge St Catharine’s College, August 2026] [The Founder's Pack, 2026]. Egoist Machines' response is a permission-based layer in which users decide which application can access which part of their context, for how long, and with a record of use, a design that aligns with rising consumer sensitivity to data control as AI products become more personal and more persistent [Cambridge St Catharine’s College, August 2026] [Henry Li - A* AI | LinkedIn, 2026].
There is also an ecosystem-level tailwind if developers want lower-friction personalization without storing every piece of user context themselves. The company says it is pursuing both direct consumer distribution and integrations into existing consumer applications, and one launch report said it had more than 10 B2B2C pilot programs involving organizations with more than 1 million users, though those counterparties were not named and the claim remains company-sourced [Y Combinator] [Pulse 2, October 2026]. If that pattern holds, the near-term market may look less like a standalone consumer app category and more like middleware for AI-enabled consumer software.
The nearest adjacent markets are identity, consent management, consumer privacy tooling, and application-layer memory for AI products. The substitute, from a buyer perspective, is not only another startup offering context portability, but also the default path in which each major assistant builds its own native memory and preference stack, reducing the need for a neutral layer [Yahoo Finance, October 2026]. That creates a mixed market setup: interoperability becomes more valuable as users add more AI endpoints, but platform owners may prefer vertically integrated memory systems if they can keep users inside one ecosystem.
Regulatory and macro forces point in both directions. A permissioned, user-directed architecture fits the broader policy drift toward user control over personal data, which supports the company's framing even if the cited sources do not name a specific statute [Cambridge St Catharine’s College, August 2026]. At the same time, any company handling sensitive personal context will face a higher trust bar as AI policy debates move from model performance to data rights, retention, and cross-platform sharing, which means adoption could depend as much on governance and product design as on raw AI demand.
| Market lens | What the sources support | Evidence |
|---|---|---|
| Core category | Consumer AI context portability and permissions infrastructure | AI Passport is described as a permission-based identity and context layer for moving personal AI data across assistants [Pulse 2, October 2026] [Yahoo Finance, October 2026] |
| Initial user | Consumers | YC describes direct-to-consumer distribution alongside integrations into consumer applications [Y Combinator] |
| Integration buyer | Consumer apps with existing distribution | YC says the company also plans to integrate into consumer applications that already have distribution [Y Combinator] |
| Early demand signal | B2B2C pilot activity, company-claimed | More than 10 pilot programs with organizations involving more than 1 million users each, according to launch coverage citing the company [Pulse 2, October 2026] |
| Main substitute | Native memory inside major AI assistants | Public sources frame compatibility with major assistants, implying those same platforms could also internalize the function [Yahoo Finance, October 2026] |
The table points to a market that is legible in workflow terms before it is legible in sizing terms. The investable question, on public evidence, is less about headline TAM and more about whether a neutral identity-and-memory layer can become necessary infrastructure before assistant vendors make interoperability optional.
One source, partially checked -- Section relies on a mix of company-originated descriptions carried by Y Combinator, Cambridge St Catharine's College, Yahoo Finance, and Pulse 2, with no independent third-party market sizing report in the cited sources.
The Competitive Field
MIXED Egoist Machines is trying to position itself one layer below the visible AI assistant battle, arguing that the durable control point is not the model or chat surface, but the user-owned context that could move between assistants and apps [Y Combinator, October 2026] [Cambridge St Catharine’s College, August 2026].
The competitive map is easiest to read in three bands. The incumbents are the large assistant platforms named in the company’s own product framing, ChatGPT, Claude, Gemini, and Muse, which already own user attention and can each build native memory, identity, and preference layers inside their own products [Yahoo Finance, October 2026] [Pulse 2, October 2026]. The challengers are startups attempting to define a portable identity or memory layer across those ecosystems, but no specific rival is named in the sourced materials, which is itself notable because it leaves the company competing more against platform behavior than against a clearly disclosed startup set [Crunchbase] [LinkedIn]. Adjacent substitutes are the status quo workflows users already tolerate, re-entering preferences manually, relying on account logins, or accepting whichever memory product a single assistant offers, which means Egoist Machines has to change user behavior at the same time it asks apps to integrate [Cambridge St Catharine’s College, August 2026] [Henry Li - A* AI | LinkedIn].
Where Egoist Machines appears to have an edge today is in the specificity of its product thesis and the founding mix behind it. The company is not presenting AI Passport as a broad consumer app, but as a permission-based identity and context layer with expiring access and usage receipts, a framing that gives it a clearer privacy and control posture than a generic personalization feature would [Cambridge St Catharine’s College, August 2026] [Pulse 2, October 2026]. That is helped by founder-market fit on paper: David Khachaturov’s Cambridge PhD and published work around machine learning security support the trust and security narrative, while Y Combinator backing and the reported seed round provide some early distribution credibility with developers and consumer app partners [Dr David Khachaturov | Department of Computer Science and Technology, 2026] [Y Combinator, October 2026] [Yahoo Finance, October 2026]. The harder question is durability. Product framing and technical credibility are real advantages at seed, but they are perishable if assistant platforms decide that portable memory should be native and free.
The company is most exposed where it does not own the user relationship or platform rail. OpenAI, Anthropic, Google, and other assistant operators can bundle memory, identity, and preference management directly into their products, then constrain third-party access through policy, API design, or default settings; the sources confirm Egoist Machines wants to work across those assistants, but they do not show a disclosed commercial partnership with any of them [Yahoo Finance, October 2026] [Pulse 2, October 2026]. That leaves Egoist Machines dependent on two external decisions at once: consumers must care enough about cross-assistant portability to adopt a separate layer, and consumer applications must decide that adding an outside context passport is worth the integration cost [Y Combinator] [Cambridge St Catharine’s College, August 2026]. The reported presence of more than 10 B2B2C pilots suggests some partner interest, but because those organizations are unnamed and the claim comes from launch coverage, it does not yet establish a moat [Pulse 2, October 2026].
The most plausible 18-month scenario is not a clean startup-versus-startup race, but a standards battle between independent portability and platform-contained memory. Egoist Machines is the most plausible winner if consumers begin using multiple assistants in parallel and if distribution partners decide neutrality matters more than dependence on any one model provider, because that would make a cross-platform passport economically useful rather than merely conceptually elegant [Y Combinator, October 2026] [Cambridge St Catharine’s College, August 2026]. OpenAI is the most plausible winner if users prove comfortable keeping their context inside one primary assistant and if native memory products improve faster than third-party interoperability layers can spread. In that scenario, the loser is less likely to be a named startup rival, none is disclosed in the record here, and more likely to be the independent portability category itself.
One source, partially checked -- Core positioning is corroborated by Y Combinator and Cambridge St Catharine’s College, but competitive inferences rely on company framing and no named direct competitors are confirmed in the sourced record.
Opportunity
Opportunity
PUBLIC The prize here is not a better consumer app but a control layer for personal AI context, and if that layer becomes standard across major assistants and consumer software, the economic value could sit closer to identity infrastructure than to a single-feature app [Y Combinator, Unknown] [Cambridge St Catharine’s College, August 2026].
The clearest upside case is that Egoist Machines becomes the default permissions rail for user-owned AI memory. The company is early, with a disclosed $3 million seed round in October 2026 and a reported two-person team, but the product thesis is unusually direct: let a user carry preferences, history, plans, and goals across assistants such as ChatGPT, Claude, Gemini, and Muse, while deciding what each application can access and for how long [Yahoo Finance, October 2026] [Cambridge St Catharine’s College, August 2026] [Y Combinator, Unknown]. That outcome is still speculative, but it is not arbitrary. The evidence shows a recognizable wedge, consumer utility around repeated context sharing, and early distribution intent through both first-party products and integrations into existing consumer applications [Y Combinator, Unknown].
The path to scale likely depends less on winning one app and more on becoming the neutral layer many apps can accept. That makes the reported traction signal, more than 10 B2B2C pilot programs with organizations said to have more than 1 million users each, directionally important even if still company-sourced and unnamed [Pulse 2, October 2026]. It also helps that the startup is backed by Y Combinator and Goodwater Capital, with participation from scientists affiliated with DeepMind and Meta, since this is a category where founder credibility and ecosystem access may matter as much as early revenue [Yahoo Finance, October 2026] [The SaaS News, October 2026].
| Scenario | What happens | Catalyst | Why it's plausible |
|---|---|---|---|
| Consumer memory rail | AI Passport becomes a common sign-in and permissions layer for consumer AI apps, similar in workflow to identity login but for selective context sharing | A high-usage consumer application adopts Passport as an integration path for portable preferences and memory | The product is already framed as a permission-based identity and context layer, and the company says it is integrating into consumer applications with existing distribution [Pulse 2, October 2026] [Y Combinator, Unknown] |
| B2B2C distribution partner | Large consumer platforms use Passport to reduce repeated onboarding and improve personalization without taking full custody of user context | One of the reported pilot programs converts into a named production deployment | The company has reported more than 10 B2B2C pilots involving organizations with more than 1 million users, which at minimum suggests outbound demand for testing the model [Pulse 2, October 2026] |
| Trust and compliance layer | Passport is adopted as a user-control standard because it offers expiring access, selective disclosure, and usage receipts | A broader shift toward permissioning and auditability in AI products makes portable user consent a product requirement | Public materials consistently emphasize granular control, time-bounded access, and receipts when information is used, which maps to a trust and governance narrative rather than pure convenience [Cambridge St Catharine’s College, August 2026] [Henry Li - A* AI |
These scenarios point to one common growth engine: every additional integration makes Passport more useful to the next app and more worth maintaining for the next user. If a consumer sets context once and can reuse it across multiple assistants, switching costs can move from the application layer to the permission layer. The early signs are modest and mostly company-reported, but they line up with that flywheel: first-party distribution on one side, integration into existing consumer applications on the other, and pilots that appear aimed at scale distribution rather than one-off individual adoption [Y Combinator, Unknown] [Pulse 2, October 2026].
The more subtle compounding effect is reputational. In AI products, users are being asked to hand over increasingly intimate context, yet today that context is fragmented across apps. A product that gives the user selective disclosure, expiring access, and a receipt trail could become the trusted default if large platforms decide that asking for less data is a better growth tactic than asking for everything [Cambridge St Catharine’s College, August 2026] [Henry Li - A* AI | LinkedIn]. That does not guarantee a moat, but it suggests the company is trying to own the rules of exchange, not just the interface.
Sizing the upside is difficult because the public record here does not include market-size studies, revenue, or named customer contracts. The closest grounded way to frame the win is by analogy to infrastructure categories that became default rails once multiple applications depended on them. If Egoist Machines were to become the standard permissions and context layer across a broad set of consumer AI products, the resulting asset could resemble identity or developer-infrastructure platforms in strategic value, though no direct public comparable is established in the source set. On the evidence available, the fair public framing is that this could become a strategic infrastructure company with platform economics if the integration model proves real at scale (scenario, not a forecast) [Y Combinator, Unknown] [Yahoo Finance, October 2026].
No independent source found -- This section relies materially on company-originated claims and republished launch coverage, with limited independent corroboration beyond Y Combinator, Cambridge St Catharine’s College, and funding announcements.
Sources
Publicly reported
[Y Combinator, October 2026] Launch YC: Egoist Machines: user-owned hyper-personalisation infrastructure for Web 4.0. | https://www.ycombinator.com/launches/St3-egoist-machines-user-owned-hyper-personalisation-infrastructure-for-web-4-0
[Cambridge St Catharine’s College, August 2026] Catz startups turning research to real-world benefit. | https://www.caths.cam.ac.uk/startups
[Yahoo Finance, October 2026] Egoist Machines AI Passport Goes Live Today - Only Way To Make Your Personal Context Portable Across AI Assistants Like ChatGPT, Claude, Muse and Gemini - But Still Owned By You | https://finance.yahoo.com/technology/ai/articles/egoist-machines-ai-passport-goes-132300417.html
[Pulse 2, October 2026] Egoist Machines Raises $3 Million Seed Round And Launches AI Passport. | https://pulse2.com/egoist-machines-raises-3-million-seed-round-and-launches-ai-passport/amp/
[Crunchbase] Egoist Machines - Crunchbase Company Profile & Funding | https://www.crunchbase.com/organization/egoist-machines
[The SaaS News, October 2026] Egoist Machines Raises $3M Seed | https://www.thesaasnews.com/news/egoist-machines-raises-3m-seed/
[LinkedIn, September 2023] Erin McGurk on LinkedIn: I am hugely excited to share that I have been shortlisted for the FCDO… | https://www.linkedin.com/posts/erin-mcgurk_i-am-hugely-excited-to-share-that-i-have-activity-7105631613522198528-YHDy
Articles about Egoist Machines
- AI Passport Could Let Users Carry Their History Across ChatGPT and Claude — The Y Combinator-backed startup is pitching a user-owned profile that travels between AI assistants, with a seed round from Goodwater and DeepMind scientists.