Pricing
One tier. $2.99 per seat per month. That's the whole price list.
The trial
14 days. $0. No credit card.
You get full workflow access — every feature. Plans, items, Board, Slack, Path A endpoints, the lot.
AI works in two places during trial:
- Per-item AI sidebar — the one that opens beside each item with the
akey. Funded by us up to $1 of usage (about 40 calls). That's our pitch — try the moat feature on us. - AI Assist tab (global chat) — needs your own Anthropic key from day 1. It's a free-form chat surface; the burn rate is too unpredictable for us to fund. Add a key in Settings → ACCOUNT → AI provider, or use the per-item sidebar instead.
When the $1 hits or day 14 lands (whichever first), AI moves entirely to your Anthropic key.
After the trial
Free — keep using Vibstr. One active project at a time. Older projects auto-pause; the data stays. Want them all active? Upgrade.
Pro — $2.99/user/month — unlimited active projects, unlimited team members, every workflow feature unlocked. Bring your own Anthropic key for AI.
That's it. No Solo vs Team vs Enterprise. No "first 500 users lock in $X forever". No bulk discount at 10 seats. No annual prepay with a confusing discount. 5 people = $14.95/month. 50 people = $149.50/month. Linear math.
What $2.99 buys
The workflow:
- Unlimited active projects
- Unlimited team members per project
- Slack integration (Discuss in Slack threads + reply mirror)
- Plans + canonical plan format parser
- Board + Kanban + swimlanes + stages
- Path A endpoints for AI-agent automation
- 7-importer migration support (VivifyScrum + GitHub Projects + Trello + Jira + Linear + Notion + Asana)
- Public shipped log
- Everything else
What it doesn't buy:
- AI compute. We don't supply AI. We don't markup AI. You pay Anthropic directly using your own API key.
- Priority support. Same docs, same community, same response time as everyone.
- SSO / SAML / SOC2. Not on the roadmap. If you need those, we're not for you yet.
BYO Anthropic key — the only AI path after trial
Five trust signals on how your key is protected (also in Settings → ACCOUNT → AI provider above the input):
- End-to-end encrypted at rest via Supabase Vault. The encryption key lives in our database's protected vault, not alongside your data. Even our platform admin cannot read your API key via SQL — only the runtime AI Edge Functions can decrypt it at the moment of an Anthropic call.
- One-way after save — once you save your key, this app only ever shows you the last 4 characters. To change it, you replace it; to remove it, you clear it.
- Only your account can read it via the app — row-level security prevents other users from accessing your key, even on the same workspace.
- Used only to call Anthropic — never sent to a third party, never logged in our audit trail, never appears in error messages or analytics.
- You pay Anthropic, not us — Vibstr never bills you for AI usage. You see the bill directly in your Anthropic dashboard.
What we actually use your AI key for — and what each call costs
Vibstr uses Claude Haiku 4.5 for every AI call. Haiku is the cheap-and-fast model. We deliberately don't use Claude Sonnet 4 (the more expensive model) for any current feature — Haiku handles structured extraction + free-form chat well enough, and the cost difference is 3-5×. We pick the cheap option on your behalf.
Six AI surfaces use your key:
| Surface | What it does | Typical call cost (Haiku 4.5) |
|---|---|---|
Per-item AI sidebar (a key in drawer) | Context-aware Q&A scoped to one item — knows the description, comments, attached plan, sibling items, and recent activity. | ~$0.006 per turn |
| AI Assist tab (global chat) | Free-form chat with access to your project's items, plans, and builds. The highest-volume surface; gated to BYO from day 1. | ~$0.008 per turn |
| Plan derive | Turns a markdown plan into structured items (zero AI cost when the body is canonical-format — the deterministic parser handles it; AI only fires for freeform plans). | ~$0.011 per 5-task plan |
| AI semantic matcher (super-function) | After a build is reported, matches Claude Code's output to open items that aren't named explicitly via closes #NN. Deterministic #NN parser runs first — AI only catches what the parser missed. | ~$0.009 per build |
| Build changelog auto-summary | Generates the markdown changelog shown on the Builds tab from your build's Claude Code output. | ~$0.010 per build |
| Bulk import AI extraction | Parses freeform pasted text (notes, meeting transcripts, email lists) into structured items. Optional — bulk import has a deterministic mode that uses zero AI. | ~$0.012 per 10-item import |
The math: at Haiku 4.5 pricing ($1/million input tokens + $5/million output tokens), the average Vibstr AI call costs around $0.01. Your $1 trial budget covers roughly 100 typical interactions spread across the 14 days. Most users use far fewer.
Why so cheap per call: we strip context aggressively. The per-item sidebar sends only the item's own description + 3-5 sibling items + last 5 activity entries — not the whole project. The plan parser uses the deterministic path first, only falling through to AI when the body isn't canonical-format. The semantic matcher gets last-shot after closes #NN already did the deterministic work.
Where your money would go if we weren't careful:
- Sending the full project history to every AI call (we don't — we scope per-item)
- Using Sonnet 4 instead of Haiku 4.5 (we don't — Haiku is fine for structured extraction)
- Running AI before the deterministic parser (we don't — the canonical plan format +
closes #NNparser are free) - Looping calls unnecessarily (we don't — tool-use loop has a 5-iteration hard cap at the Edge Function level)
You can verify all this live: every AI call surfaces a toast showing input + output tokens + cost. Cumulative spend lives in Settings → ACCOUNT → AI provider with a per-call breakdown. If you spot us being wasteful with your money, that's a bug — open an issue.
How cancel works
Cancel anytime via Stripe Customer Portal (Settings → ACCOUNT → Subscription → Manage in Stripe Portal). Cancel-at-period-end is the default — you keep Pro features until the end of your current billing period, then you drop to Free. No retention emails. No "are you sure" loops. No "wait, here's 50% off if you stay" guilt trip.
If you delete your account, your data exports cleanly as JSON or CSV any time before the deletion. See /how-we-ship/best-practices/ for the export recipes.
Common questions
Why $2.99 specifically? Cheap enough that "let me try the paid tier for a month" is a low-friction decision. Honest enough that we cover hosting + our (small) AI compute cost during trials. We don't pretend $2.99 makes us rich — it doesn't. We're aiming for sustainable, not extractive.
What if Anthropic raises their prices? Your bill goes up. Ours doesn't — we don't sit in the middle. The flip side: if Anthropic drops their prices (Haiku 4.5 was cheaper than Sonnet 4), your AI costs drop immediately. We're not insulating you from the AI market — we're just not adding a margin on it.
Last updated: 2026-05-27 (1.20.2).
Source of truth: this page. If anything in the app contradicts this, the app is wrong — open an issue.