Bonfiyah

The pricing decision, in plain text

Why Pro and Pro AI are separate tiers.

Pro = the recording-volume problem solved. Pro AI = stakes-attached AI work. The split exists because the underlying economics genuinely diverge — and most users don't need the second half.

The first thing to know about Bonfiyah's pricing: we resisted the impulse to bundle.

The conventional move would be a single Pro tier at $19.99/mo that bundled everything Pro AI does today. It would have higher conversion, higher ARPU, and a simpler pricing page. It would also be wrong, because it would charge $19.99/mo to roughly 70% of users who don't need the AI inference cost burned on their behalf.

Two genuinely different problems.

The recording-volume problem is what most active recorders run into around month three. You have 60+ recordings. The Free tier's 120-minute-per-month cap is now a real constraint. You want sync between iPhone, iPad, and Mac. Persistent speakers across recordings would be useful. Search across the library is useful. An AI meeting summary on every recording is useful — and it's included in Pro, along with Speaker Memory Health, the window into how well your speaker memory knows each voice. All Pro features. All built on architecture that runs once at recording time — one bounded pass per recording, not open-ended reasoning across your whole library.

One thing that isn't a Pro feature: the widgets and the live recording Dynamic Island. Those are free for everyone, in every tier — Flag / Quiet / Stop and a capture meter without unlocking, plus eight widgets in all — four on the Home Screen and Lock Screen, three live activities, and a Control Center starter. The Pro AI data some widgets show (Daily Brief, Promise Ledger, Contradictions) needs a subscription to populate, but the widgets themselves never do.

The stakes-attached AI problem is different. You have a meeting that matters and you want a 250-word executive briefing across the project's last six recordings. You have a witness deposition and you want commitments tracked with timestamps. You have a compatibility analysis you only want to run once on this conversation, with the consent gate, with the four research frameworks. These are inference-heavy operations that cost real money to run, and they only make sense for users who are routinely doing high-stakes work.

The math.

An AI Summary on a 60-minute recording uses roughly 25,000 input tokens through the inference layer. At Bonfiyah's negotiated rates, that's about $0.04 of variable cost per generated summary — bounded, once per recording, and now included in Pro. Project Context, Compatibility, Team Dynamics, and the deep Speaker Insights pass each cost between $0.06 and $0.18 of inference. A heavy Pro AI user runs 50–80 inference operations a month — call it $5–10 of variable cost. Add infrastructure (Apple's 30%/15% cut, the upstream transcription provider line, iCloud transfer, multi-device sync infrastructure, observability) and that is what Pro AI's $19.99/mo, with its 35 hours of cloud transcription a month, is priced to cover.

Pro at $14.99/mo carries a small, predictable inference cost: the ~$0.04 per-recording AI summary, and nothing open-ended. It covers that, the App Store cut, the iCloud transfer cost, the upstream transcription provider line, the multi-device sync infrastructure, and a stable recurring contribution to the engineering work. What Pro doesn't carry is the cross-recording reasoning suite — the operations that run large models across your whole library on demand.

If we'd bundled at $19.99/mo flat, we'd be charging Pro users for cross-recording inference they're not running. The most committed-but-not-AI-heavy customers — the litigators recording witness sessions, the journalists building source archives — would be paying for inference they don't use. The pricing would feel wrong, and over time, it would cause churn.

Why one monthly price.

The price has to be high enough to absorb the long-run inference and infrastructure costs, and low enough that someone unfamiliar with the category will give it a fair test. So each plan has one price, month to month, with nothing that steps up later: Pro at $14.99/mo; Pro AI at $19.99/mo. New accounts get 14 days of Pro and Pro AI first, no card needed. The cross-recording surfaces compound, so a short look isn't enough to judge the product: give Pro AI a few weeks of real meetings to know whether it's earning its place in your work, and cancel anytime if it isn't.

The clean upgrade trigger.

Most users hit Pro before they hit Pro AI. The trigger for Pro is concrete: you record more than a couple of hours a month, or you want sync across iPhone/iPad/Mac, or you want speakers and search to work across recordings. Once any one of those is true, $14.99/mo is correctly priced.

The trigger for Pro AI is different. It's not "I'd like more AI features." It's "I have a specific class of work where being right about a meeting matters more than $20 a month." Witness preparation. Investor commitments. Per-client diagnostic reads. Cross-project executive briefings. The work where the result of getting it wrong is worse than the cost of running the inference.

Most Bonfiyah customers stay on Pro indefinitely and that is correct. About 20% add Pro AI; some keep it; many cycle in for stakes-attached projects and back out for the quiet months. We charge by the month so this is frictionless. And the Free tier — 120 minutes of recording a month — is the no-commitment way to try Bonfiyah before you ever pay, and new accounts get 14 days of Pro and Pro AI, no card needed.

What this means for you.

If you record meetings regularly and want them to be searchable, summarised, and synced across devices: Pro. $14.99/mo. The vast majority of active recorders correctly land here.

If your work has a layer where AI getting it right materially matters — high-value meetings, legal proceedings, per-client diagnostic work, decision-making across projects — Pro AI. $19.99/mo. Month to month, so you can find out on your specific work whether the AI features carry their cost.

The Free tier is the full app for up to 120 minutes a month on a single device, indefinitely. We resisted gimping it on features beyond what scales with infrastructure cost. If you're a casual recorder with a few meetings a month and one device, you should never need to upgrade — and we'd rather have you happy on Free than annoyed on Pro.

One line on cloud transcription.

Recording on Pro and Pro AI is uncapped — record as long as you like. The one line worth stating plainly: cloud AI transcription with speaker separation is included for 25 hours of recording a month on Pro and 35 hours on Pro AI, and for up to 10 hours in any 24-hour period, per account. Past either, recording and transcription keep working on-device (Apple Speech) so you're never blocked — full cloud transcription comes back when the month resets on the 1st (UTC), or at the start of the next 24-hour period for the daily ceiling. This is the same Silent Fallback behaviour that kicks in when the upstream provider has a bad moment: the capture never stops, the cloud speaker-diarization pass just waits for the allowance to reset.

Pro AI ships twenty cross-recording surfaces — Proactive Notifications, AI Project Context, Compatibility Analysis, Team Dynamics, MBTI Compatibility (Keirsey-Bates pair scoring derived from Big Five via Costa & McCrae 1992), Ray Dalio Pointillist Painting (10-dimension team heatmap + 16 PrinciplesYou archetypes), Mach-IV (Machiavellianism inline card per Christie & Geis 1970, conservatively scored), Daily Brief (one morning roll-up, capped at 1 critical ping/day, optional email-with-deep-links — v3.0.447), Weekly Report (your whole week in one summary — meetings, promises kept/broken, people, decisions, topics), Bonfire Pulse (weekly self-portrait through the 10 Principles of Burning Man, Larry Harvey 2004 — unaffiliated, fair-use; v3.1.0), Founder Insights (weekly six-dimension mirror on operator posture — Vision Clarity, Conviction, Customer Obsession, Capital Discipline, Hiring Posture, Operating Cadence; v3.0.764), Speaker Memory Health (a voice memory that learns over time and self-heals — Strong / Still learning / Needs attention, voice-samples learned, mixed-voiceprint and confusable-with-someone flags; computed from voice biometrics only — now also included with Pro), Promise Tracker, Contradictions, Pre-Brief, People Memory, Story Mode, Speaker Themes, Project Actions, and the new in-app Help AI (conversational assistant on every screen — your-data-grounded answers with citations, structured tool proposals before any action runs; v3.0.900). These surfaces use templated, deterministic outputs where stability matters, and AI inference where insight matters.

And finally: Bonfiyah Friends. Once you're on Pro or Pro AI, every paid signup you refer earns both sides a free month. No cap →

See the full pricing breakdown with every feature on every tier.

Bonfiyah

Stay near the fire

Bonfiyah notes — feature deep-dives, real-world recipes, the occasional pricing-philosophy explainer like this one.

No spam. We use ConvertKit. See our privacy policy.