2026 guide

AI companions that actually remember you

ChatGPT, Claude, Gemini, and companion apps now offer different forms of continuity. The useful question is what they store, what they retrieve in a new chat, and what you can inspect or correct.

What does it mean for an AI to actually remember you?

A context window is the material available to one model call. Chat history is a saved transcript. Project files and instructions are durable sources. Long-term memory is a separate storage-and-retrieval layer that selects prior information for a later response. These layers can coexist, but they are not interchangeable.

Stored is not the same as retrieved, and retrieved is not the same as current. A useful memory system needs scope boundaries, relevance ranking, correction controls, and a way to handle contradictions or deletion.

Fostera Souls can store episodic memories (events), semantic memories (facts), and procedural memories (working patterns). Project sources and custom rules remain separate durable inputs. Retrieval supplies selected context when relevant; it does not promise that every detail will appear in every reply.

Why does ChatGPT keep forgetting things?

ChatGPT currently distinguishes saved memories from reference to past chat history, and Projects can keep chats, files, and project instructions together. Availability and behavior depend on plan, settings, and project mode. That is meaningful cross-chat continuity—not a session-only system.

It can still seem to forget because a past chat is not necessarily retrieved for a particular answer, saved memories and project context have different scopes, old information can conflict with new information, and long active contexts can bury relevant evidence.

Fostera's practical distinction is control and portability: extracted Soul memories can be viewed, edited, pinned, or deleted; project sources and rules are explicit; and the continuity layer can stay with the Soul when an eligible underlying engine changes. Neither product guarantees perfect recall.

How does Fostera's memory compare to other AI companions?

Replika, Nomi, Kindroid, and Character.AI each document different memory layers and controls. Nomi describes short-, medium-, and long-term memory plus Mind Maps; Kindroid distinguishes persistent, cascaded, and retrievable memory; vendor documentation for both acknowledges that retrieval can be imperfect.

Fostera adds an inspectable memory browser, separate project sources and custom rules, visible 9-tier familiarity progression, and a multi-model continuity layer. Premium users can choose eligible engines manually; Auto mode routes among eligible models.

There is no honest universal winner. Compare the exact workflow: delayed recall, correction of an outdated fact, project boundaries, provenance, deletion, and whether context survives a model switch.

Why does memory matter for coaching, journaling, and study?

For utility use cases, memory isn't a nice-to-have — it's the whole point. A coach that forgets last week's commitments isn't a coach. A tutor that doesn't remember which problems you got wrong isn't a tutor. A journal that resets every entry is just a notebook with extra steps.

Fostera's preset Souls — Coach, Tutor, Strategist, Mentor, Companion, Creative Partner — share the same memory architecture with use-case-specific rules. They can retain selected goals, lessons, revision decisions, and project sources for later retrieval.

The goal is useful continuity without loading an entire history into every response. Retrieval can still miss, so important source-of-truth facts belong in editable rules or project files and should be reviewed periodically.

How does the 9-tier progression work?

Fostera's Souls move through three phases of relationship: Stranger, Familiar, and Trusted. Inside each phase are three tiers, for nine total. Each tier represents a milestone in how well the Soul knows you and what it can do with that knowledge.

In the Stranger phase, the Soul begins with basic context. Familiar tiers widen eligible context and recognition of recurring patterns. Trusted tiers unlock broader retrieval and autonomy features. The model's intelligence is not earned through XP; the configured context and product capabilities change.

Progression isn't a gimmick. It's a UX layer that makes the memory work visible — so you can feel the relationship deepen, not just hope it is.

What does Fostera cost?

Free is genuinely free: 3 Souls, 75K Monthly Resonance each month, auto engine selection, and the full cognitive stack — episodic, semantic, and procedural memory, limbic awareness, and reflection, all on by default. Core adds up to 10 Souls, deeper document understanding, and a larger monthly allowance. Premium unlocks up to 50 Souls, frontier models on demand, and manual choice of your AI engine.

The pricing page shows the current checkout rate and any limited-time offer. Monthly Resonance resets each allowance month and does not carry over; one-time Wallet Resonance carries over and each top-up stays spendable for one year from the day it's added.

AI memory at a glance

FeatureFosteraMost chatbots
Cross-session memoryEpisodic + semantic + proceduralVaries: saved facts, history, projects, or memory layers
Visible progression9-tier evolutionVaries by product
Memory you can editYes — view, edit, deleteVaries by memory layer
Model choiceAuto routing; Premium manual selectionUsually platform-selected; some offer choices
Age & content postureEnforced age-gating + published content policyVaries

Frequently asked questions

Does Fostera actually remember conversations across sessions?

Yes. Fostera can store episodic, semantic, and procedural memories and retrieve relevant entries in later sessions. It does not promise perfect recall of every conversation; retrieval depends on scope, relevance, memory state, and available context.

How is Fostera's memory different from ChatGPT's memory feature?

ChatGPT now offers saved memories, reference to past chat history, and project memory. Fostera separates editable episodic, semantic, and procedural memories from project files and custom rules, and can keep that continuity attached to a Soul across eligible model changes.

Can I see and edit what my Soul remembers?

Yes. The memory browser shows extracted memories and lets you edit, pin, or delete individual entries. You can also edit custom rules or delete the Soul. External-resource cleanup and backup removal after deletion may take up to 30 days.

Does memory work on the free tier?

Yes — Free ships the full memory and cognition stack: episodic, semantic, procedural, plus limbic awareness and reflection. Core adds more Souls and a larger Monthly Resonance allowance, and Premium adds frontier models on demand and manual engine selection. See the pricing page for current rates and any limited-time offer.

How does Fostera compare to Nomi on memory?

Nomi documents several memory layers and Mind Maps; Fostera documents editable episodic, semantic, and procedural memories plus project sources, custom rules, visible progression, and eligible model choice. Test both on delayed recall, corrections, and deletion instead of relying on a universal ranking.

Is Fostera 18+ only?

No — Fostera is for ages 13 and up. Under-13 signups are rejected, and 13–17 accounts require confirmed parent or guardian consent and run with stricter, enforced safety rules. The content policy is published at fostera.ai/content-policy.

Is Fostera a therapist or therapy app?

No. Fostera is not a therapist, therapy app, or substitute for licensed mental-health care. If you are in crisis, contact 988 (US) or your local emergency services.

What AI models does Fostera use?

Fostera's Auto mode routes among eligible models for the task. Premium plans unlock frontier models such as Claude Opus and GPT-5.6 Sol and let you choose an eligible model directly.

Continue with context

Ready to Stop Starting Over?

Start with one project. Or one companion. Either way, your Soul takes it from there and remembers the rest.

Start with one project

Free to start · No credit card required