Talk things through in guided conversations, journal your thoughts, upload handwriting, and explore everything in one continuous visual memory you can interact with — helping you understand yourself better and keep track of your reflections.
Free
Launch Team / Built With
Render: Get up to $100K in startup creditsCloud infrastructure for founders who want to move fast and NOT break things
Promoted
ChatGPT by OpenAI
Gemini
A continuously updating “Mind Graph” will hit scale pain around privacy plus safety: sensitive journals, model hallucinations, and graph nodes drifting or merging incorrectly over time.
Best practice is client-side encryption with minimal telemetry, transparent provenance on each node/edge back to the exact entry, and crisis-aware guardrails with tested escalation paths.
How are you storing and versioning the graph (property graph vs embeddings-only), and can users export or audit the exact transformations that created each connection?
@ryan_thill Hey Ryan! Yes, during development I had the same concerns.
To minimize these risks, every node and edge in the Mind Graph keeps references to the original conversation or journal entries they were derived from, using hashed identifiers. This means the system always has a “reverse path” back to the source, similar to how provenance is preserved in tree or graph traversal algorithms.
Merging is indeed the trickiest part. Before merging nodes, the system checks for shared edges, paths, and contextual overlap to avoid losing semantic meaning. Merges are not treated as final. The system includes a self-healing mechanism that continuously re-evaluates past merges as new context appears.
For example, if you talk about your child in one conversation and later mention a child again, the system may initially merge those nodes. If later context indicates these are actually two different children, the merge can be undone and the connections split accordingly. Merges are treated as revisable hypotheses, not permanent truth.
We also maintain a time-based history (“time machine,” as I call it), which allows the graph to be viewed as it existed at a specific point in time.
Regarding storage: we use a real graph database as the primary store. Nodes also carry embeddings, which are used for synonym handling, similarity detection, and linking graph structure back to source content. Embeddings support discovery; the graph remains the source of truth.
Crisis detection and escalation paths are also in place and treated as a first-class concern.
From a user audit perspective, today the UI supports (see screenshots below):
Logging and surfacing every change made to the Mind Graph
Clicking any node and asking the system to explain connections using therapeutic reasoning
Continuing reflection directly from that node with full context
Deeper audit and export capabilities are something we’re actively considering for future releases.
Hope this answers your questions.
Wow, I love it! Does it mean that AI can understand human emotions better than we do ourselves? (personally, I don't always have full clarity on what I'm feeling, so I find the product useful 😁)
@alina_petrova3 ahaha, good point :)
Of course AI can’t "understand" your emotions. However, New Norm is not a generic AI. It designed to help you understand your emtions by noticing and connecting emotional dots that live in your subconscious and naturally surface in the thoughts you share when you speak or think out loud.
New Nörm reflects those signals back to you and represents them graphically through the Mind Graph (actually it is one of meaning of this feature). This helps move what was implicit into conscious awareness, where real clarity can happen :)
Hi team, considering that everything is stored in a continuous memory, I wonder what privacy or security measures should users know about?
@milena_sarukhanyann Hello Milena! Great question. Because New Nörm works with long-term personal memory, privacy and security are core to how the system is built.
In short, all user data belongs to the user. We don’t sell it, and we don’t use personal content to train public models. Memory is private and isolated per user, and you can request a full hard-delete of your data at any time.
From a technical perspective, data is depersonalized internally using hashed identifiers. User accounts and reflective content are stored in separate services, and all data is encrypted both in transit and at rest. Access is tightly controlled and limited to what’s necessary for the system to function.
We’re building New Nörm as a private reflective space, not a data source. It’s designed the same way we’d want our own inner lives to be handled.
What makes New Norm scientifically grounded rather than just reflective writing?
@lusine_abgaryan Thanks for question. The difference is that New Nörm doesn’t rely on free-form reflection alone, or on whatever an LLM happens to generate in the moment.
Our AI agents are guided by scientifically established therapeutic modalities and techniques, developed and reviewed with input from practicing therapists from our own board. These frameworks shape how questions are asked, how reflection is guided, and how responses are structured.
This means the system isn’t just reacting to prompts or trying to please the user with agreeable answers. Reflection is constrained and guided by proven approaches, so the focus stays on awareness, patterns, and understanding rather than quick reassurance or surface-level insights.
In short, New Nörm combines reflective writing with structured, evidence-based guidance, instead of leaving everything to the language model alone.
How does the guided conversation work – is it scripted or adaptive?
@ina_avdalyan1 This is a good one! Guided conversations in New Nörm are built on scientifically proven frameworks developed by our research team and adapted specifically for New Norm. Depending on what you’re discussing, the system can follow a structured, guided flow (scripted) to help you think something through step by step.
At the same time, it’s adaptive. If your direction, emotions, or intent shift during the conversation, New Nörm dynamically adjusts — switching from a scripted path to a more open, responsive dialogue. You’re never “stuck” in a script. The guidance evolves with you, in real time, based on what you share and where the conversation naturally goes.
Congrats, team!
I have one question. How does New Nörm remember me differently than other apps?
@luiza_avetisyan Hello Luiza! Thanks for question.
Most apps remember you as a timeline: separate notes, separate sessions, separate chats. That memory is flat.
New Nörm builds a connected memory. What you share is stored in a graph-based structure where thoughts, emotions, and experiences are linked and updated over time. This means the system doesn’t just recall past entries, it understands how they relate and evolve, and that context actively shapes future reflection.