Questions, answered.

Opposite — fewer steps. A chatbot hands you a wall of prose and forgets it the next time you open the app. We turn the same back-and-forth into structured outputs — an itinerary, a budget, a decision log — and keep them current as the conversation moves on. The AI also reads the whole thread, not just the last message, so it knows what the group already agreed.

Anything with people, choices, and moving parts: a trip, a move, a kitchen renovation, a long weekend with friends, a wedding, a household budget, a recurring Saturday with the other families. The templates cover the common shapes; the underlying artifacts (itinerary, budget, places list, decisions, timeline, materials, packing, phase plan) compose into anything else.

Yes — that's the headline feature. Decisions are recorded as their own artifact with who voted, why, and what they ruled out. Six weeks later when someone asks "didn't we already decide the quartz?" the answer is there with the reasoning, not buried in scrollback. Memory facts (allergies, addresses, hard budget caps) persist across Spaces so you don't re-state them.

Friends and small groups (two to ten people) who plan things together — trips, moves, renovations, ongoing household decisions, recurring social plans. Less for solo personal task-management; less for big-org project software. The product assumes the conversation is the planning surface.

They mostly don't have to. One link drops them straight into a Space that already has a draft plan in it — not an empty room asking them to set up an account. They can read, vote, and reply over chat or email. The AI keeps the plan current even from the members who never open it. The only ones who need the full app are the people doing the AI conversation themselves.

Both, eventually. Today the chat lives in the app. We're building lightweight presence in the channels people already use (group chat, email reply-to) so the plan updates from where the conversation is actually happening. For now: send the link, people land in the Space, the conversation happens there.

Override it. The AI proposes; the group decides. Every artifact edit is reversible and every decision can be re-opened. Thumbs-down on an AI message is a real signal — it shapes future replies and goes into our evaluation set so the same mistake doesn't repeat. The AI never silently changes a decision the group made.

Yes. Owners can remove any member, and non-owners can leave on their own. An owner who wants out can hand the Space off to another member first — ownership transfer is built in — or archive the Space.

Yes. Connect with someone and you can start a direct message — 1:1 or a small group — separate from any Space, with reactions, edits, and photo or file attachments. You also get a private thread with Pango that's always open. And an activity feed on Home keeps you caught up on what your connections and your Spaces are doing without ten separate threads.

No. It drafts, organizes, surfaces tradeoffs, and records what the group chose — but it doesn't transact. Booking is still a thing humans do, on the actual vendor sites. We hold the plan; you hold the credit card.

It routes between a small/fast model and a strong model per turn. "Did everyone get the date?" is a quick check; "reconcile the new mover quote against the budget and the deposit timing" is real work. You don't pick the model — the routing layer does, with the goal of cheap-enough-for-the-question and strong-enough-when-it-matters.

Less than a default chatbot would on the same prompt, because every artifact is the source of truth, and the AI reads from them instead of inventing facts. When it doesn't know, it says so — and when an extraction is ambiguous it surfaces a confirmation chip instead of silently writing. That said: it's still an LLM. Don't treat AI-drafted dates or prices as final without a glance.

Each AI action — a reply, a photo caption, a public-summary regen — debits a small number of credits. Everyone is on one plan with a monthly base allowance that refreshes every month; when you need more, you add recurring credit blocks that raise your monthly ceiling. Hitting the cap drops the AI to a lighter fallback model rather than locking you out, and warnings start at 75% so it's never a surprise. Full math on the pricing page.

Yes — there's one plan and it starts free, with the whole product and a monthly AI credit allowance that's enough to actually plan a weekend or a small trip. No feature is locked behind a paywall, and no card is needed to start; sign in and go.

Add a credit block — a recurring +1,000 credits every month for one simple price. Add as many as you want; each renews monthly and you can remove any of them anytime. Remove them all and you're back to free. The exact credit amount and price live on the pricing page so we keep them in one place rather than risk drift here.

The AI drops to the cheaper fallback model — slower, simpler answers, but still working. You don't get locked out and the plan doesn't freeze. Warnings start at 75% so it's never a surprise. Want the heavy work back sooner? Add a credit block and it resumes right away.

Yes. Each credit block is its own monthly subscription — remove one, keep the others, or remove them all with a single button. No email to support, no retention dance. You keep access through the end of the billing period, then drop back to the free base. Your Spaces and history aren't deleted.

Anything you put in is yours. The AI sees the conversation in the Space it's reading, plus the artifacts in that Space, plus your personal memory facts you've opted to keep. We don't sell data, don't train external models on your conversations, and don't build a profile of you for advertisers — we don't have advertisers.

No. Each Space is a scope. The AI in your renovation Space can't read your trip Space, even though both are yours. Cross-Space recall happens only via memory facts you've explicitly opted to keep (e.g. "my home address is X", "my hard cap on travel is $Y") and you can see and edit those at any time.

We default to not extracting facts about other people unless they've opted in. You can flag information about other members; we don't write it to their personal memory without their say-so. Spaces are shared context, not a back-channel dossier on the people in them.

Yes. Archive hides; delete removes. Deleted Space data is purged from primary storage immediately and from backups within the standard retention window (30 days). Public share links die instantly when revoked — the cached summary is dropped too, so a re-published Space doesn't surface stale text.

Primary storage is Postgres on managed cloud infrastructure (US-East today). The AI requests go to Anthropic. We're a small team, the security model is straightforward: signed JWT auth, role-based access per Space, no public read of private data without an owner-issued share token.

Web, iOS, and Android — all feature-equivalent today. The web app is a real installable PWA; the iOS and Android apps are native. Picking one platform doesn't lock the rest of the group out of theirs.

Reads work offline once a Space has loaded. Writes queue and sync when you're back online. AI replies obviously need the network.

Not yet, deliberately. We're shipping the product first; an API surface comes once the data shapes are stable. If you have a specific use case, tell us — early integrations shape the API.

There's a Send feedback option inside the app (Settings → Feedback). Goes straight to the team, replies are real humans. For security issues, email security@stateful.com.