Personal AI
Personal AI is the AI memory platform from Human AI Labs, aimed at telecom carriers and enterprise teams. It pairs Memory Core infrastructure with a no-code Persona Studio and the My AI assistant; access runs through its sales team.
What is Personal AI?
Personal AI is the platform published by Human AI Labs, Inc. and positioned in 2026 as carrier-native memory infrastructure: the layer that turns every identity on an operator's network into an agent that keeps evolving. Its thesis is that memory, meaning what an AI has lived through, plus context, meaning what it has been told, produces an identity that emerges rather than one that is configured. Five memory primitives are orchestrated: encoding, stabilizing, storing, retrieving and updating, across episodic, semantic, working, relational, temporal and procedural memory. Five capabilities are announced: persistent AI identity, recall acceleration, dynamic encoding, centralized governance and portability. The offer is split into three layers: My AI, an assistant aimed at everyone; Persona Studio, a no-code building platform for AI builders and creators; and Memory Core, the infrastructure aimed at developers and AI engineers. Contractually it is a software-as-a-service offering providing AI conversational models. Figures published from the Comcast network report 15 ms to first token and end-to-end voice under 500 ms; a separate compute-cost benchmark against Gemma-27B is quoted at $0.02 per million tokens, a comparison rather than a customer rate, since no price is published anywhere. The home page itself is a conversational demo rather than editorial content.
What it does
- Give every identity a persistent memory that is encoded, stabilized, stored, retrieved and updated
- Run inference on the carrier network, with a stated 15 ms time to first token
- Build and deploy AI personas without writing code, through Persona Studio
- Govern permissions, policies and memory retention from a single control plane
- Carry memory across systems, agents and assistants
- Sustain end-to-end voice exchanges, measured under 500 ms on the Comcast network
When to use Personal AI / When not to
A quick filter to help you decide if Personal AI is the right fit.
When to use Personal AI
- Telecom carriers and network operators, who get memory and inference running on their own network, with tokens metered on the line
- Enterprise product and AI teams looking for persistent per-identity memory instead of a retrieval pipeline replayed at every call
- Regulated verticals the site equips directly: legal (nine documented roles), finance, healthcare and retail
- Non-technical builders, since Persona Studio advertises persona deployment without writing code
- Developers and AI engineers building on Memory Core, which ships public developer documentation
When not to use Personal AI
- Individuals and solo operators: there is no self-service path, no pricing page and no online sign-up, since every call to action leads to a contact form
- Anyone unwilling to sign a B2B framework contract, as access requires a Master Subscription Agreement plus an Order Form
- Mobile-first users: no iOS or Android app is published today, and the home page claims there is no app to install
- Teams that need an interface in a language other than English, none being announced
- Anyone under 18, an age floor set by the privacy notice
How to use Personal AI
A typical end-to-end flow, from setup to results.
- Start with the conversational demo on the home page, which runs in the browser with nothing to install
- Decide which layer you need: My AI for an assistant, Persona Studio to build personas, Memory Core to build on the infrastructure
- Use one of the site's calls to action, from Book a Carrier Briefing to Talk to Sales, Meet your own My AI, Try persona studio or Start building, all of which lead to the contact form
- Discuss scope with the sales team, since there is no self-service sign-up
- Sign the Master Subscription Agreement and the Order Form that sets the subscription term
- Open access for the Authorized Users named in that Order Form
- Sign in through the customer application once the account is live
- In Persona Studio, define the personality, knowledge domains, memory behavior and guardrails, without code
- For a developer project, build on Memory Core using the public developer documentation
Pros & Cons
Pros
- An explicit contractual commitment not to train AI models on customer content in raw form (terms 6.7)
- Ownership of content and outputs left with the customer (terms 5.2)
- A complete, named GDPR setup: identified DPO, EU and UK representatives, Standard Contractual Clauses and a DPA built into the contract
- A public sub-processor list, with prior notification and a right to object
- Performance figures published together with their point of comparison and the network they were measured on
- Public developer documentation and a status page
- Customer content stays accessible and exportable for 30 days after the contract ends
Cons
- No public pricing: no pricing page, no rate card, no entry-level amount, so no cost can be assessed without going through sales
- No free trial and no free plan announced
- No hosting location disclosed, the privacy notice stating that data may be processed anywhere in the world
- No security certification such as SOC 2 or ISO 27001 is claimed on the pages collected
- Personas are not exportable and cannot be used outside the platform (terms 5.6), which creates a strong dependency
- The legal documents date from April 2025, before the carrier-native repositioning, and still describe the consumer product with its Memory Stack and its mobile and desktop apps
- Service Metrics remain exploitable by the publisher for its own purposes even after the contract (terms 5.5), while liability is capped at the fees of the preceding 12 months (terms 8.2)
Pricing & Plans
No price is published. The site carries no pricing page, no rate card and no entry-level amount, and the sitemap contains no pricing URL either; no free plan and no free trial are announced. The cost is set commercially: a negotiated subscription whose fees are fixed in a signed Order Form (terms 1.5 Fees, 1.7 Order Form, 1.14 Subscription Term). Section 3.1 provides that the customer shall pay Personal AI all fees in US dollars, on or before the due date set in the applicable Order Form or, failing that, within 30 days of receiving the invoice. No refund is granted on termination, other than the unused prepaid fees covered by the intellectual property indemnity in section 7.1. The figure of $0.02 per million tokens quoted on the About page is a compute-cost benchmark against Gemma-27B, not a customer rate.
- pricing is handled entirely by the sales team
- My AI
- the assistant layer
- offered without a published price
- Persona Studio
- the no-code persona building platform
- offered without a published price
- Memory Core
- the memory infrastructure layer
- offered without a published price
- Contracting runs through an Order Form
- which sets the subscription term and the authorized users
Data, GDPR & hosting
A consolidated view of how Personal AI handles your data.
GDPR overview
GDPR implementation is named rather than merely claimed. The privacy notice in force is dated 9 April 2025. A Data Protection Officer is identified, Kenny Huynh (kenny.huynh@personal.ai). An Article 27 EU representative is appointed, Ametros Ltd in Cork, Ireland (gdpr@ametrosgroup.com), alongside a UK representative, Ametros Group Ltd in Hereford, England. The legal bases cited for the EU and the UK are performance of the contract, legitimate interest, consent and legal obligation. Rights covered are confirmation, access and portability, rectification, erasure, restriction and objection, and withdrawal of consent; they are exercised through a dedicated data request page, with the option of lodging a complaint with a supervisory authority. International transfers rely on the EU Standard Contractual Clauses, with a UK Addendum referenced in the DPA, which is itself attached to the contract as Exhibit 2, its Exhibit A carrying the clauses. Privacy contact: privacy@personal.ai.
Who owns the data?
Section 5.2 of the terms of service is explicit: the customer owns all right, title and interest in the Customer Content, the Output and the related intellectual property rights. Personal AI keeps the rights to its own Personal AI IP (5.3). Personas sit in between. Section 5.6 states they are not exportable from the platform and may not be used independently of it, while Personal AI undertakes not to use or share one customer's Personas with another customer without prior written agreement. Section 5.5 lets Personal AI collect Service Metrics and use them for its own purposes, during and after the contract. The privacy notice makes Personal.ai controller for website users' data and processor for its business customers' Customer Data.
Reuse rights
Customers can reuse their own material without asking: section 5.2 assigns the Customer Content, the Output and the related intellectual property rights to the customer, with no further permission required. The exception is Personas, which under section 5.6 cannot be exported from the platform or used independently of it. On the vendor side, section 6.7 carries the central commitment: Personal AI represents, warrants and covenants that it shall not train any AI model using Customer Content in raw form, except for the Persona(s) expressly agreed with the customer or otherwise for that customer's exclusive benefit. The same section rules out access to a Persona's memory unless the customer requests it, a court order requires it, or a check on compliance with the contract makes it necessary, and it allows a Persona to be stopped or deleted at the customer's request. The privacy notice adds that the Memory Stack and related data are not sold or shared for monetary consideration. Two caveats remain: aggregated or anonymized data may be reused without limit and shared with third parties, and profiling and automated decision-making appear among the stated uses of account data. Data obtained through Google APIs, such as Gmail, falls under the Google restrictions: no advertising use, and no human reading outside the listed cases.
Data retention & training
Hosting summary
The site is precise about its own lack of precision. The privacy notice states that all personal information processed by Personal AI may be transferred, processed and stored anywhere in the world, including but not limited to the United States or other countries. No hosting country or region is declared, and no cloud provider is named on any of the pages collected. Transfers outside the EEA, Switzerland and the United Kingdom are covered by the EU Standard Contractual Clauses, with a UK Addendum referenced in the data processing agreement. The sub-processors that handle customer data are published on a dedicated updates page, with prior notification and a right to object. One clarification is worth making: the site's IP address resolves to Amazon (AS16509, anycast), but that describes where the website is served from, not where customer data is stored. In short, the contractual framework for transfers is documented while the physical location of the data is left open.
Things to keep in mind
Risks and trade-offs to weigh before adopting Personal AI.
- The legal documents, privacy notice and terms of service both dated 9 April 2025, still describe the consumer product with its Memory Stack and its mobile and desktop apps, none of which is corroborated by the current product pages
- The integrations listed on this page come from that April 2025 privacy notice, not from an up-to-date product page: check with the vendor before relying on any of them
- No security certification is claimed, the site presenting security documentation supplied on request as the only evidence
- Service Metrics remain exploitable by the publisher after the contract has ended
- Profiling and automated decision-making are named among the uses of account data
- Hosting is not localized, the privacy notice reserving processing anywhere in the world
- The company incorporation date is not published: this page carries an archive-based fallback of 18 March 2016, earlier than the domain registration of 16 December 2017
Setup & Integrations
Technical difficulty
The difficulty lies less in the technology than in the procurement: there is no self-service start, so every deployment begins with a sales contact and a signed Order Form. After that, the effort depends on the layer. My AI is announced as requiring no setup and no configuration. Persona Studio is configured without code, covering personality, knowledge domains, memory behavior and guardrails. Memory Core targets developers and AI engineers and comes with its own documentation. A carrier deployment is the heaviest case, since it involves network integration alongside the hardware partners named on the site, HPE and Lanner Electronics.
Deployment
Integrations
Behind Personal AI
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Personal AI.
Frequently asked questions
How much does Personal AI cost?
Is there a free trial or a free plan?
Is my data used to train their models?
Who owns the content and the outputs?
Is the product GDPR compliant?
Is there a mobile app?
Should you pick Personal AI?
Personal AI is solid on the legal and governance side: a contractual commitment not to train on raw customer content, a named data protection officer, EU and UK representatives, a published sub-processor list and a DPA built into the contract. The counterweight is complete pricing opacity, since no rate appears anywhere and no cost can be assessed without a sales conversation, with neither a free plan nor a trial to soften the entry. The site is also mid-repositioning: the home page has been reduced to a conversational demo, while the terms of service and the privacy notice, both effective 9 April 2025, still describe the earlier consumer product. Taken together, the signals point to a single audience, telecom carriers and enterprises able to negotiate a framework contract and integrate memory infrastructure. An individual user will find no way in.
- Choosing a selection results in a full page refresh.
- Opens in a new window.