Role Model AI
Role Model AI builds photorealistic talking avatars from a single portrait photo. Its Concierge product lets that avatar hold live video conversations, join Zoom calls, stream through OBS, sit on your website and remember your ongoing story between sessions.
What is Role Model AI?
Role Model AI is an avatar platform whose product, Concierge 2.4, turns a single portrait photograph into a photorealistic talking avatar you can hold a live video conversation with in your browser. Creation runs through a guided flow: upload the photo, then set the avatar's name, personality, opening script, knowledge base, accent colour and voice. Voices come from presets, from a written description, or from an audio sample cloned on the paid tiers.
The same avatar is then reused across five surfaces. Talk runs the face-to-face session in the browser, with a live transcript, microphone controls, camera and screen-share vision tools and a real-time credit meter. Stream with Avatar pushes it into OBS through a Browser Source URL, adds a phone-based mute dock, and runs a public-safe mode that keeps private memory, goals, inbox and connected apps off air. Meetings sends the avatar into Zoom, Google Meet and Microsoft Teams. Living Bio publishes a public page where visitors meet your AI twin while you are offline. Living Website embeds the avatar on your own site as a chat bubble, a fullscreen launcher or an iframe.
Behind the avatar sits a personal layer the vendor calls Life Concierge: durable memory, a private timeline, goals, a daily brief, a weekly life review and natural-language reminders gathered in one in-app inbox. A tools hub connects Notion, LinkedIn and Home Assistant, with Google Calendar and Gmail announced but not yet shipped. Paid users can also code by voice: a Concierge MCP server and a Cursor companion extension let the avatar queue jobs for Cursor or Claude Desktop.
Everything is metered in credits at three credits per minute of talk time. The first avatar is free and further ones cost 300 credits each. A marketplace lists community avatars, and an announced ownership layer called AuRA promises consent-first control over voice, likeness and knowledge — though it remains a waiting list. The company describes itself only as a company powered by Manifest Anything and as part of the SELF Labs family, and publishes no legal entity or postal address anywhere on the site.
What it does
- Turn one portrait photo into a photorealistic talking avatar with its own voice, personality and knowledge base
- Hold a live face-to-face video conversation in the browser, with real-time transcript and screen or camera sharing
- Send the avatar into Zoom, Google Meet and Microsoft Teams meetings
- Put the avatar on a live stream through an OBS Browser Source, muted from a phone
- Publish a public bio page where visitors talk to your AI twin, or embed the avatar on your own website
- Keep a persistent personal layer — memory, goals, a daily brief, a weekly review and reminders
- Delegate coding jobs to Cursor or Claude by voice through the Concierge MCP server
When to use Role Model AI / When not to
A quick filter to help you decide if Role Model AI is the right fit.
When to use Role Model AI
- Creators and influencers who want a public Living Bio page where followers can talk to their AI twin while they are offline
- Live streamers and video producers who need an on-air avatar driven from OBS with a phone-side mute control
- Founders, operators and managers who want a voice-driven assistant that keeps memory, goals, daily briefs and reminders in one place
- Developers on the paid tiers who want to queue coding jobs for Cursor or Claude by talking to an avatar over Concierge MCP
- Small teams and solo businesses on the Expert plan who want a face-to-face avatar embedded on their own website
When not to use Role Model AI
- Regulated buyers who need a GDPR position, a data processing agreement, a named hosting jurisdiction or a sub-processor list — none of which the vendor publishes
- Anyone under 18, since the Terms require the age of majority to open an account or use paid features
- People looking for licensed medical, legal, financial or therapeutic advice, which the vendor explicitly says its avatars do not replace
- Free-tier users hoping to publish, stream or embed: 15 welcome credits buy roughly five minutes of talk, and Living Bio, streaming, meetings and embedding all require a paid plan
- Buyers who need an unbranded white-label widget or a mobile application, as embedded avatars always carry Role Model AI branding and the iOS app link on the site is dead
How to use Role Model AI
A typical end-to-end flow, from setup to results.
- Sign up with an email address on the Concierge application; the account opens with 15 welcome credits and one free avatar
- Upload a clear portrait photograph to build the avatar
- Configure the name, personality, opening script, knowledge base and accent colour
- Pick a voice: a preset, one designed from a written description, or a clone of your own voice on Pro and Expert
- Open Talk to start the face-to-face session in the browser; on Safari, tap Start conversation so the microphone is granted
- For streaming, open the Stream with Avatar panel, paste the Browser Source URL into OBS and keep the mute dock open on a phone
- For meetings, open Send Avatar to Meeting and pick an avatar marked READY before joining Zoom, Google Meet or Teams
- For a public page, go through the Living Bio creator flow, claim a username, add links, choose whether talk auto-starts, then publish
- For a website widget on Expert, open the Embed Avatar panel, enable a personal avatar and copy the bubble, fullscreen or iframe snippet
- For voice-driven coding on Pro or above, generate an MCP token, paste the configuration into Cursor or Claude, and install the companion extension from the VSIX file
Pros & Cons
Pros
- One avatar serves five distinct surfaces — browser talk, streaming, meetings, a public bio page and a website embed
- A public demo runs on the site without an account, and the free tier lets you build a real avatar before paying
- Pricing is published in full, with a feature-by-feature comparison table across the three tiers
- Credits are fungible between talk time and extra avatars, and packs or custom top-ups from five dollars work on every tier
- Annual billing is genuinely cheaper: two hundred dollars a year for Pro against two hundred and forty monthly, and nine hundred and ninety-nine for Expert against one thousand one hundred and eighty-eight
- The public-safe streaming mode deliberately keeps private memory, goals, inbox and connected apps off air
- Integration is built on the open Model Context Protocol, so it works with Cursor and Claude Desktop rather than a proprietary client
Cons
- No legal entity, no postal address and no country of establishment appear anywhere on the site
- No GDPR statement, no data processing agreement, no sub-processor list and no declared hosting location
- The vendor never addresses whether customer material is used to train models, and offers no opt-out
- Version claims are inconsistent: the brand says Concierge 2.4 while the legal pages and help centre still say 2.2 and the web manifest says 2.3
- The iOS App Store link published on the site returns a 404 and the application has been delisted
- Embedding is Expert-only at ninety-nine dollars a month, streaming and meetings need Pro, and the free tier cannot publish a Living Bio
- Visitor conversations on a Living Bio or an embedded widget are billed to the creator's credit balance, and Role Model AI branding cannot be removed from the widget
Pricing & Plans
A permanent free plan is available at no cost, providing 15 welcome credits — approximately five minutes of avatar talk time — and one custom avatar. The lowest paid entry point is the Pro plan at USD 20.00 per month, or USD 200.00 when billed annually. The Expert plan is priced at USD 99.00 per month, or USD 999.00 annually. Usage is metered at three credits per minute of avatar talk time. Credit packs range from USD 5.00 for 500 credits to USD 149.99 for 25,000 credits, and custom top-ups are accepted from a USD 5.00 minimum on every tier.
- 15 welcome credits (about five minutes of talk time)
- one free custom avatar
- live video avatar chat
- live transcript
- recent chat history
- memory
- daily brief
- goals and reminders in the application
- 2
- 000 credits per month (about eleven hours)
- extra avatars at 300 credits
- voice cloning
- Stream with Avatar over OBS with the mute dock
- Zoom
- Google Meet and Microsoft Teams
- in-session image and video generation
- 10
- 000 credits per month (about fifty-five hours)
- everything in Pro
- website embedding
- advanced AI models
- higher image and video generation limits
- priority avatar processing
- analytics and usage insights
- Credit packs (all tiers) — USD 5 for 500 credits
- USD 6.99 for 600
- USD 14.99 for 1
- 500
- USD 34.99 for 4
- 000
- USD 79.99 for 10
- 000 and USD 149.99 for 25
Data, GDPR & hosting
A consolidated view of how Role Model AI handles your data.
GDPR overview
There is no GDPR implementation to report. The words GDPR and CCPA appear nowhere on the site, which was checked exhaustively against the JavaScript bundle that carries every page of this single-page application. The Privacy Policy does say, without naming any regulation, that depending on your location users may have rights to access, correct, delete or export their personal data and to object to or restrict certain processing, and it routes those requests to the support mailbox. Beyond that, nothing: no Article 27 EU representative, no data protection officer, no data processing agreement, no published list of sub-processors, no hosting jurisdiction, and no statement on whether customer material is used to train models. Treat this as a documentation gap rather than as an implied position.
Who owns the data?
Under the Terms you keep ownership of everything you upload: portrait photos, voice samples, knowledge material and creative work. You grant Role Model AI a limited licence to host, process and display that content, and only in order to deliver the service you asked for. The vendor states that it does not sell personal information, and that data is shared with hosting, analytics, payment and email providers under contractual confidentiality — but it never names them, apart from Stripe for payments. AuRA is presented as a future layer for controlling how avatar assets are shared and licensed; it has not shipped. No data processing agreement is offered, and the legal pages carry a single effective date of 24 June 2026.
Reuse rights
The Terms grant the end user no reuse rights over anyone else's material. You may work with what you create through your own avatar, since ownership of uploaded content stays with you, but the acceptable-use rules forbid uploading portraits or voices belonging to other people without their permission, and forbid using an avatar to deceive, defraud or misrepresent identity. Premade and marketplace avatars cannot be embedded on your own site or used on stream, and only personal avatars you built yourself qualify for embedding. Embedded widgets always display Role Model AI branding, which cannot be removed. Any wider licensing of avatar assets to third parties is deferred to AuRA and the marketplace, whose terms are described as still to come.
Data retention & training
Hosting summary
The vendor declares no hosting jurisdiction at all. Neither the Privacy Policy, the Terms nor the Cookie Policy names a country, a region or a data centre. The only security statement is generic: industry-standard safeguards, encryption in transit, access controls and secure infrastructure, with the usual caveat that no internet transmission is entirely secure. Service providers are described by category only — hosting, analytics, payment processing and email — and Stripe is the single third party named anywhere on the site, in the cookie section. What can be observed from outside the documentation is not the same as a commitment: the marketing site is served from Netlify, the application runs on a Supabase backend, and the domain resolves to an Amazon Web Services anycast address geolocated in the United States. None of that is stated by the vendor and none of it constitutes a contractual guarantee about where personal data lives. Organisations with residency requirements will need to ask directly before signing.
Things to keep in mind
Risks and trade-offs to weigh before adopting Role Model AI.
- Your face and your voice become reusable assets: portraits and voice clones are uploaded and stored, and the vendor has published no retention period and no training policy for them
- Analytics and advertising trackers load on the homepage before any consent action, including a Meta advertising pixel and Google Analytics
- Credits are consumed by other people: every visitor who talks to your public bio page or embedded widget spends your balance, and enabling auto-start makes that happen the moment a page opens
- A photorealistic avatar of a real person is a persuasive artefact; the vendor forbids impersonation and deception in its terms, but the technical capability is in your hands and misuse falls on you
- Outsourcing memory, goals and daily planning to an assistant is comfortable and can quietly erode the habit of holding your own priorities and reviewing your own week
- Avatar conversations are framed as coaching-style guidance and are explicitly not a substitute for licensed medical, legal, financial or therapeutic advice, yet the format invites exactly that confidence
- Announced features — AuRA, the wider marketplace, Google Calendar, Gmail — should not be counted on when choosing: they are waiting lists rather than shipped capabilities
Setup & Integrations
Technical difficulty
Basic use is easy: sign up with an email, upload a portrait, set voice and personality, then talk in the browser — nothing to install. Streaming is a middle step, pasting a Browser Source URL into OBS and keeping the mute dock open on a phone. Embedding means copying a snippet, with an optional host script for page actions. The genuinely technical part is voice-driven coding: generate an MCP token, paste the configuration into Cursor or Claude, then install the companion extension from its VSIX file through the extensions menu rather than by opening the file.
Deployment
Integrations
Supported languages
Behind Role Model AI
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
How do credits work on Role Model AI?
Is there a genuinely free plan?
What does the paid subscription add?
Can the avatar join a Zoom, Meet or Teams call?
How do I put the avatar on my own website?
Is Role Model AI GDPR compliant?
Does Role Model AI train its models on my conversations?
Who is behind Role Model AI?
Is there a mobile application?
How do I contact support or get API access?
Should you pick Role Model AI?
Role Model AI is unusually complete on the product side and unusually thin on the company side, and both halves matter to the decision. What it builds works and is easy to verify: a single portrait becomes a photorealistic avatar you can actually talk to in a browser, and that same avatar then shows up on a live stream, in a Zoom call, on a public bio page and inside a website widget. The credit model is simple and readable — three credits a minute, first avatar free, top-ups from five dollars — and the pricing page publishes a full comparison table rather than hiding behind a sales call. A public demo runs without an account, so nothing has to be taken on trust before spending.
The reservations are almost entirely about governance. The site names no legal entity, publishes no postal address, never mentions data protection regulation, offers no processing agreement, names no sub-processors and declares no hosting jurisdiction. It also says nothing at all about whether the material you upload — portraits, cloned voices, knowledge bases, session transcripts — feeds model training, and provides no way to exclude it. Small inconsistencies reinforce the impression of a fast-moving team: the brand advertises Concierge 2.4 while the legal pages still describe 2.2, and an App Store link on the site now points at a delisted application.
The practical read is that this suits creators, streamers, coaches and solo operators who want a striking presence quickly and are spending their own money. Organisations with procurement, residency or compliance requirements should get written answers on entity, hosting and training before committing anything sensitive to it.
- Choosing a selection results in a full page refresh.
- Opens in a new window.