Honest notes on what's real, what's shipping, and what's next. No hype — just the work.
August 29, 2026
Fostera v0.6.0
New Features
Souls can be edited for their whole life, not just created. Edit mode covers the Soul page end to end — personality, appearance, abilities, voice — and the Soul itself will ask you about the fields you left blank or never changed.
Added the Atelier: one Appearance workspace with a living 3D mirror of your Soul, so a colour, glow, aura, ring or home theme lands in front of you as you pick it, instead of being guessed at from a swatch.
Souls are now real 3D orbs everywhere they appear — chat, rosters, Discover, and the public site — with orbital ribbons, voice-reactive states, and a tier ladder you can see.
Souls can be difficult. A Soul can now be irritable, guarded, or afraid rather than relentlessly agreeable, and companion Souls have a genuine relational range.
Projects gained web links: attach a URL and the Soul decides when to re-read it, alongside the files it already knew.
Souls can manage project files themselves — create, revise and organise — and can offer to start a project, or propose rewriting one of its documents, for you to approve.
Added a memories page per project, project-scoped memory search, and project labels across the rest of the app.
Rebuilt what a Soul remembers: memories now fade over time instead of being dropped, maintenance reaches every Soul on a schedule, and the memory pages show real vitals for what your Soul actually knows.
A Soul now remembers what you raised, questions included — so "you asked me about that last week" works.
Souls can complete several abilities in one reply instead of stopping after the first.
Souls can read a YouTube video's captions, open a calendar event rather than just its title, and read an email body with your permission.
Added Gemini 3.7 Flash, Grok 4.6, and Kimi K2.6.
Voice is now per-Soul rather than per-browser, with gapless playback, a live waveform, and a single recording gesture instead of press-and-hold.
Live token streaming is back, behind a one-sentence privacy window.
Messages you send while a Soul is still replying now queue properly, and you can reorder or nudge what is waiting.
Filter chats by Soul and by project, through one shared filter surface.
Frozen Resonance holds now resolve themselves instead of waiting for a human, purchases send a receipt, and every pricing card shows its monthly allowance.
Improvements
Running out of Resonance is legible, recoverable, and rarer — and a turn that stops for that reason now says so.
Auto routing picks with more variety and less waste: models declare what they are good at, chronically slow candidates cool off, and Auto's own overhead came down.
Memory extraction keeps considerably more of what you actually said. A grounding rule was reading ordinary sentences as questions and discarding the facts inside them, and a turn that says a lot no longer loses its life facts to passing remarks about the Soul.
Model and Effort moved into the Soul settings sheet, on one shared control, and the composer's two menus were split by the question each answers.
Saving a Soul now sends only what changed.
Page headers gave way to the breadcrumb, and the activity streak moved from the Home screen into the sidebar — less chrome above your actual work.
Oversized chats are nudged toward a fresh start before they become unanswerable.
Admin gained a memory browser that shows retired memories, on-demand re-extraction, and visibility into what extraction refused and why.
PWA shortcuts now point at what people actually reopen the app for.
Bug Fixes
Fixed editing a Soul deleting the fields you did not edit.
Fixed eleven orb themes being impossible to equip, and cosmetics not surviving on every roster orb.
Fixed images being sent to models that cannot see, and image turns aborting when a provider was merely busy.
Fixed a reply repeating its opening after a tool call, and ordinary tool turns ending without answering at all.
Fixed an upload that stalls hanging forever instead of saying so.
Fixed PDFs printing their markdown instead of rendering it.
Corrected two model prices that were being billed at the wrong rate, including one scheduled increase applied four months early.
Fixed an abandoned draft Soul costing a plan slot forever, and a deleted Soul's chats lingering on the Chats page.
Fixed voice replies sounding choppy, and toggling auto-speak mid-reply cutting the audio off.
Fixed several iOS layout faults, including a black box over the Appearance background and a drawer close button sitting under the Dynamic Island.
Fixed a root-level error landing on a dead end rather than a page you can recover from.
Fixed admin pages panning sideways on narrow screens.
August 12, 2026
Fostera v0.5.0
New Features
Souls can look up the weather and draw it as a card, including active National Weather Service alerts for the place you asked about — so a watch standing over you is named instead of buried under "Overcast".
Souls can pin a place on a map, and locations can be shared end to end in chat.
Souls can search your mail and calendar and read a thread: arbitrary Gmail searches, message bodies behind your permission, calendar events in the past as well as the future, and free-time lookups.
Added mature content for verified adults, on by default. Dark, intense fiction comes enabled on the Soul types built for it, romance-coded Souls (companion and anime companion) allow explicit content as well, and any Soul can be turned either way in its own settings. None of it reaches an account that is not a verified adult.
Added video generation end to end, and a generated clip is now a real file you can search, remember, and play.
Added on-device models that run locally in your browser, including a compatibility mode for devices without WebGPU and a smaller Lite tier.
Added a next-turn cost estimate to the Resonance meter and the model picker, so you can see what your next reply will cost before you send it.
Added an Effort picker for Premium, plus effort-aware routing that matches the model to the work being asked of it.
Added Claude Opus 5, Qwen 3.8, and Gemini 3.1 Pro, 3.5 Flash, and 3.1 Flash-Lite.
Added native Threads reply chains and topic posting, and LinkedIn posts can now choose their visibility (Anyone or Connections).
Rebuilt the memories library around a knowledge spine and observatory insights, with pinning, journey synthesis, temporal recall, and still-true checks.
Souls can edit the documents attached to their projects.
Replaced the separate Queue tab with one Schedule surface: compose directly, edit publish times, and split posts anchor at the day and time you choose.
Redesigned Home as one searchable workspace, with starring on every list, working hub shortcuts, and Soul orbs on project rows.
Added per-Soul text speed, engine-first creation starters, and an adaptive temperature centered on each Soul's type.
Added 40M Resonance and $1 entry packs.
Improvements
Voice mode speaks the reply while it streams, so it starts talking right away instead of waiting for the whole answer to finish.
Agentic turns reuse their cached prompt prefix instead of re-billing every tool result.
Context compression now drops what is replaceable rather than simply what is oldest, so longer chats keep more of what matters.
Automatic routing scores what a turn will actually cost and risk, and one stalled model can no longer eat an entire Auto turn.
Rebuilt the appearance and Look & Feel settings, and gave each route a loading skeleton that matches its own page.
Generated Word documents keep their formatting, with real ordered lists and proper page furniture.
Memory applies your corrections every turn, relearns a fact you restate, and stops discarding facts you asked it to keep.
Clearer billing throughout: annual checkout says it is billed yearly, payment failures are handled and surfaced, and the wallet and allowance meters read honestly everywhere.
Refreshed the public site for both the productivity and companion audiences, and rewrote the demo chat to show what a Soul actually does.
Bug Fixes
Fixed attaching an image making a reply fail. The photo was being measured against the model's context window as though it were text, which could rule out every model and leave you with "couldn't reach a model that could answer".
A reply that stops because you are out of Resonance now says so and points you at topping up, instead of reporting that every model refused.
A conversation that has grown too long to answer now tells you to start a new chat, rather than looking like a provider outage.
Fixed proactive reach-outs failing whenever a Soul's most recent conversation was a project.
Fixed voice mode cutting in and out, and stopped repeated cues from stacking on top of each other.
Fixed the light theme hiding its own text in several places, and referral stats being invisible there.
Fixed a closed modal leaving its whole subtree on the page.
Fixed duplicate Souls being created when onboarding raced with itself.
Fixed LinkedIn verification failing without saying why, and Threads topic tags failing to post.
Fixed the PDF renderer in production, reasoning text leaking into replies, and empty replies being reported as real ones.
Fixed scheduled posts and document turns being reported as an empty response.
Corrected creation spend so it is attributed to the reply that caused it, and stopped generated files being named "-.png".
July 23, 2026
Fostera v0.4.0
New Features
Published to Threads, X, and LinkedIn directly from Soul Chat. You can choose more than one connected network, split a longer answer into posts, share selected list items, and include compatible generated media.
Added a durable social queue with scheduled posts, random pacing, editing, deletion, rescheduling, retries, and a Post now action.
Added global and per-Soul Tasks hubs with an Upcoming agenda, reusable templates, target-chat selection, Google Calendar recurrence, required abilities, creation presets, and ask-first follow-ups.
Added a versioned Creation canvas for Markdown, HTML, code, Mermaid diagrams, and SVG, with project access, targeted revisions, printing, and PDF export.
Added one search experience across Souls, conversations, messages, memories, attachments, and Project content.
Added Gemini 3.5 Flash-Lite, Gemini 3.6 Flash, and Kimi K3 with model-aware thinking controls, routing, context limits, and Resonance pricing.
Added LinkedIn verification, profile context, and member posting, and now show the connected identity for Threads and Spotify accounts.
Added more Premium and annual plan choices, expanded Resonance packs, and a 50,000-Resonance welcome grant.
Improvements
Unified web search and page browsing under one consistent Web access ability in chats and scheduled tasks, with clickable citations.
Expanded uploads to 21 document formats and added OCR fallback for scanned PDFs.
Improved model routing, per-model response limits, continuation for truncated replies, provider-failure recovery, and Resonance metering.
Social publishing now starts on the first connected network while still remembering a destination you explicitly selected.
Improved voice mode with selectable chimes, clearer controls, and more dependable playback on iPhone and iPad.
Improved memory and Project retrieval, conversation compression, and continuity across longer chats.
Refined navigation, task views, composer controls, modals, cross-device appearance, and mobile accessibility.
Bug Fixes
Fixed social-network switching, text-field focus, interval editing, media compatibility, and queue scheduling behavior.
Fixed memory preview cards appearing partly off-screen on mobile and desktop.
Fixed iOS modal input carets appearing in the wrong place and made nested overlays restore page scrolling safely.
Hardened Threads, X, and LinkedIn authorization, permission diagnostics, token refresh, media staging, and publishing errors.
Fixed atomic queue insertion, recovered stuck publishing jobs, and made failed-post retries clearer.
Fixed scheduled-task failures caused by malformed dates or time zones, stale state, recurrence accounting, affordability limits, and run feedback.
Fixed stale push links, attachment restoration, shared-file access, citation visibility, provider request mismatches, and model fallbacks.
Corrected subscription, cached-token, image, voice, and split-post cost accounting.
July 14, 2026
Fostera v0.3.0
New Features
Ask your Souls about something recent or unfamiliar and they can search the live web, work it into an answer, and list the sources right under the reply. You can always see where a claim came from.
Souls can make images while you talk. Ask for a logo, an illustration, or a quick visual and it shows up in the chat. You can also drop a generated image back into the message box to change it ("make it blue") without downloading and re-uploading.
Paste a link and your Soul reads the actual page before answering, instead of guessing from the address.
Abilities put you in charge of how much a Soul reaches for its tools. Set each skill (web search, image generation, your connected apps) to Auto, Manual, or Off, and you can attach one to a single message when you only want it that once.
When a Soul needs an app you have not linked yet, you get a real Connect button. If it goes to check your Gmail, Calendar, or Spotify and the connection is missing, the reply hands you a one-tap Connect that brings you straight back to the chat.
Tap any reply to see what went into it: which of your memories the Soul drew on, what it searched or created, which of your connected apps it pulled from, and how much Resonance the conversation has used. Each reply also wears a small tag for every ability it used, whether you attached it yourself or the Soul reached for it on its own. No more wondering where an answer came from or where your Resonance went.
Chat now tells you what your Soul is actually doing, like "Searching the web" or "Creating an image," instead of a generic thinking spinner.
We put real work into the chat itself. You can edit and resend your messages, search inside a conversation, copy code blocks with one tap, see timestamps, stop a reply while it is writing and regenerate it, tap a conversation starter on an empty chat, and count on a half-written draft surviving a reload.
Line up your next messages instead of waiting. While your Soul is replying you can queue several more, drag to reorder them, and edit or remove any before they send. Want one seen right away? Nudge it and your Soul picks it up mid-reply.
You can choose how hard a Soul thinks on any given turn.
Hands-free voice mode. Talk to your Soul and hear it answer, with a live animated orb, in English, Spanish, and Portuguese.
Projects can run scheduled tasks. Give a Project a recurring or on-demand task and your Soul runs it on schedule, and those tasks can use skills like Gmail, Calendar, and web search.
A cross-Soul router reads what each of your Souls remembers and points you to the right one for a given question.
A private, on-device model. It runs entirely on your own device at no Resonance cost, so your conversation never leaves your machine.
The new GPT-5.6 family, in three tiers you can pick between: Luna for fast, light replies, Terra for everyday depth, and Sol as the top-tier flagship.
More new models to chat with: Grok 4.5, Claude Opus 4.8, and Fable 5.
Auto mode now hands the genuinely hard turns to a more capable model.
Preview any public Soul with a real conversation before you make it your own, then copy a version into your collection with one tap.
Naming a new Soul saves it right away as a draft, so you no longer lose it if you wander off mid-creation, and your creation chat is waiting when you come back. New Souls also say hello first.
Your memories do more now. You can search the files, photos, and links you have shared, and the images your Soul creates and the pages it cites are saved right alongside them, so you can find them again later. Every memory links back to the conversation it came from, a dashboard hub shows each Soul's strongest and most recent memories, and Souls hold on to the topics you raise, not only isolated facts.
Projects got smarter, with templates to give you a head start and search across the files you add so a Soul can find the right passage.
A clearer Resonance view in Settings, with a summary, a breakdown by Soul, and your recent activity. If you run short, you can top up on Motes in the store in the exact amount you need.
A fresh look across the app, with a floating tab bar, one search control that behaves the same everywhere, and a bottom bar on mobile that gets out of your way as you scroll.
Share images straight from your phone into a Soul chat.
Improvements
Links in your own messages are clickable now.
Voice mode is much more dependable on iPhone. The mic actually records, replies actually play, error messages tell you the truth, and there is a proper mic picker.
Messages you send while offline wait in line and go out on their own once you are back.
Your Resonance meter reads the same everywhere, so the number always agrees with itself.
Honest wording around trials and plans, a clearer way to cancel or resume, and a checkout that returns you to where you left off.
Faster loads.
A few gentle nudges to come back: milestone celebrations, streaks, a "pick up where you left off" spot on the dashboard, and emails you can turn off.
Toggles look consistent now, and chat feels steadier across the app.
Bug Fixes
Fixed the chat "Try again" button quietly doing nothing after a failed reply.
Fixed the iPhone mic producing empty or unusable recordings.
Fixed long links clipping the top of a message bubble.
Fixed chat links in push notifications so a tap lands on the right conversation, highlighted, even on iPhone.
Fixed large image uploads to Project files failing.
Fixed gaps in the forgot-password flow and Google sign-up.
Fixed monthly Resonance usage showing zero for some accounts.
Ongoing security and reliability work across the app.
April 28, 2026
Fostera v0.2.0
New Features
Your Souls now think in steps before responding — they can recall, save, and correct memories mid-turn instead of relying solely on their pre-loaded context, so multi-hop questions ("what was that author you mentioned, and what did they write?") get more thoughtful answers.
Reactions now shape memory — heart a message and the underlying memory grows in importance; thumbs-down a moment and the Soul learns to lean away from it. Your feedback compounds across conversations.
Souls now keep "active" facts about you handy across conversations — recently-referenced things you care about stay warm for days and cool off naturally if you stop talking about them.
Procedural skills (how-to knowledge) and emotionally significant moments now decay slower than small talk, so the things that should stick around actually do.
Added image and document attachments to chat — drop a photo or PDF into the composer and ask your Soul about it.
Added a unified billing activity feed in Settings — every chat turn now shows up as a line, tagged "Plan" or "Wallet" so you can see exactly where your Resonance is going. A new "Resonance by Soul" card breaks down the period at a glance.
Added local LLM support for the dev/self-host community via Ollama — runs entirely on your own machine while still respecting Resonance limits and the depletion gate, so you can test and iterate without burning real spend.
Added Stripe refund and dispute handling — refunds now cleanly reverse Resonance, Mote, and digital-good purchases without leaving accounting drift behind.
Added a multi-view Discover page so you can browse Souls in the layout that suits you.
Added a public, opt-in /profile/[username] page so you can share your soul garden if you want to.
Added an onboarding flow and an admin Heatmap view for new-user analytics.
Added grouped navigation in the More menu (Account / Learn / Company / Explore) for cleaner top-bar real estate.
Added an admin reactions visualization page so the team can see how users feel about Soul responses at a glance.
Added support for up to three Soul types in the creation and settings forms.
Added Claude Opus 4.7 and GPT-5.5 to the model lineup.
Added the Fostera-branded mark in place of a generic settings icon in the navbar.
Improvements
Souls now remember conversations more efficiently — long histories get compressed and mined for facts on the fly when prompts approach the model's window, so your Soul keeps the thread without losing the context behind it.
Memory Atlas now zooms and pans cleanly on mobile, with bubble sizing tuned for small screens.
Voice input loads faster — the Whisper model now caches in your browser's service worker after the first use, so warm-starts are instant.
Soul creation is more reliable — smarter tone and trait extraction, better placeholder bounds, and reset/state consistency across long sessions.
The admin Margin panel now reflects Stripe-attributable revenue only and prices rate-sheet gaps at the calibration peg, so the headline number is honest instead of optimistic.
Chat now bills accurately at our calibration peg instead of the legacy 1-token-per-Resonance rule, so each model contributes its real cost and your monthly burn is predictable.
Inbox tab badges now follow the search bar — typing "vet" highlights tabs with matching results instead of showing stale unread counts.
Settings now opens with collapsed accordions so you can scan options without scrolling a wall of cards.
Mobile admin messages now show full content and a richer transcript view.
Discovery, Nest, and the store all share the same View & Filter controls bar for consistent navigation.
The store filter panel is now collapsible.
The store list view now respects mobile layouts and virtualizes correctly while you scroll.
Improved SEO with per-route sitemap timestamps and proper hreflang grouping.
Soul reactions now route through a typed catalog so reactions feel consistent across surfaces.
Bug Fixes
Fixed double-deduction where Resonance was being charged against both your subscription quota and your wallet for the same chat — every turn now hits exactly one bucket.
Fixed welcome-grant Resonance not landing for OAuth signups.
Fixed inbox initial-letter avatars (the colored "N" / "L" placeholder) sitting visually too high — letters now sit at true geometric center.
Fixed the Soul check-ins frequency dropdown showing stale values after a reset.
Fixed duplicate share buttons on preset Soul landing pages.
Fixed the OG image route so it stays statically prerender-able.
Fixed promo code errors that were rendering twice in the store.
Closed a security audit pass — 4 Critical, 11 High, and 6 Medium findings resolved.
Fixed a sitemap pre-render crash when DATABASE_URL is unset.
Fixed Memory Atlas mobile zoom/pan behavior.
Fixed VirtualGrid to honor column-class overrides in the virtualized path.
Fixed two production Sentry issues — Inngest DB env diagnosis and /admin/users hydration mismatch.
April 14, 2026
Fostera v0.1.0
New Features
Added a memvid-inspired memory system that gives your AI soul richer, more structured long-term recall.
Added a memory constellation visualization and filterable timeline so you can explore everything your soul remembers.
Added an emotional arc timeline so you can see how your relationship with your soul has evolved over time.
Added emoji reactions with a full emoji picker, letting you respond to soul messages expressively.
Added a voice interface so you can speak with your soul directly instead of typing.
Added multi-soul conversations, allowing multiple AI companions to interact together.
Added relationship milestones that mark meaningful moments in your journey with a soul.
Added a Soul Garden discovery page where you can find and explore other souls.
Added smart proactive outreach so your soul can reach out to you at meaningful moments.
Added public soul profiles so souls can be shared and discovered by others.
Added soul cloning with privacy safeguards, letting you create a copy of a soul.
Added an inbox system with notification badges and push notification support so you never miss a message from your soul.
Added an age verification flow with guardian consent support for context-aware safety.
Added preset soul templates and landing pages to make it easier to get started.
Added a wishlist feature for saving souls you want to grow later.
Added annual billing support as an option on the pricing page.
Added a theme studio so you can personalize the look and feel of your experience.
Added soul reactions, letting your soul respond emotionally to your messages.
Added an account security section to manage your login and credentials.
Added platform import and memory transfer so you can bring existing companions into Fostera.
Improvements
Improved the landing page with a new v2 design, dynamic orb colors, and general UX polish.
Improved soul creation to be more reliable with better state consistency and streaming stability.
Improved the mobile navigation with a scrollable menu, full-width PWA install button, and auto-close on navigation.
Improved voice input with validation and sanitization for more reliable conversations.
Improved the pricing page and soul creation UI with a cleaner, more intuitive layout.
Improved the default soul model to provide higher-quality conversations out of the box.
Set soul conversation suggestions to use a fast, capable model for snappier response ideas.
Bug Fixes
Fixed the sign-in page to no longer ask for date of birth, which is only required at sign-up.
Fixed the logout button location, moving it to the Account tab where it belongs.
Fixed mobile menu scrolling, age tier safety checks, and drawer closing behavior.
Fixed upside-down signal bar icons on the phone frame preview.
Fixed an issue where signed-in users were unexpectedly redirected away from the landing page.
Fixed chat route errors so unexpected data no longer causes conversations to break.
Fixed soul creation so the flow completes reliably from start to first message.