All comparisons · vs Recall.ai

meetbot vs Recall.ai.

Recall.ai built the category — meeting-bot-as-a-service — and the API surface they shipped is the one most of this market now copies. Three thousand customers and counting. We mirror their HTTP API on purpose so you can switch in an afternoon, and then we charge you less per bot-hour. Recall's hosted API is closed-source; their public sample repos on GitHub ship without a LICENSE file (which is effectively all-rights-reserved). Our SDK + samples are MIT — clone, fork, ship. Below: every spec line, the math at three usage points, and a section dedicated to where they're still the better choice.

01 · tl;dr

The short version.

01Recall.ai bills $0.50/hr recording + $0.15/hr transcription + $0.05/hr-30-day storage. meetbot is a flat $0.30/hr for the bot-hour. Transcription on meetbot is BYOK today (Whisper/Deepgram/AssemblyAI on your key) and hosted ships Q3 2026 — so a like-for-like bundled comparison at GA, not today.
02Recall's hosted backend is proprietary; their public sample repos ship without a LICENSE (effectively all-rights-reserved). meetbot ships MIT-licensed SDKs, OpenAPI spec, and sample apps; the hosted backend is proprietary. Hosting: Hetzner Falkenstein (eu-central-1) by default vs US-default for Recall.
03Recall covers 6 platforms today (Meet/Teams/Zoom/Webex/Slack-Huddles/GoTo) and lists 3,000+ customers. We cover 3 (Meet/Teams/Zoom), have zero paying customers today (we're pre-launch), and ship Webex + Discord in M2. They are the safer pick for breadth and reference customers today.

Use Recall.ai if…

→You need Webex, Slack Huddles, or GoTo coverage today.
→You need a SOC 2 Type 2 report and a HIPAA BAA in the contract on day one.
→You're indifferent to license — a proprietary hosted API is fine — and price is not a buying criterion.
→You want US-region data residency by default.

Use meetbot if…

→Your unit economics break above $0.40/hr on the bot-hour — most LLM-on-meeting products do.
→Your customers ask where the data lives, and you'd like to answer eu-central-1.
→You may want to read the SDK + sample source under MIT and fork it without license-clause anxiety.
→You want flat per-minute pricing on the bot-hour and no storage line. Transcription is BYOK today — pipe per-speaker audio into Whisper/Deepgram/AssemblyAI on your own key (zero meetbot fee on that leg) until our hosted Whisper ships Q3.

02 · spec table

Side by side. No spin.

Numbers verified against the cited source on the date in the page footer. PR a correction if anything has moved.

	meetbot	Recall.ai
recording price	$0.30 / hr (flat bot-hour)	$0.50 / hr^[1]
transcription price We do not ship transcription today. Per-speaker audio + the meeting platform's native captions are emitted; transcription is BYOK on those tracks until our hosted Whisper ships Q3.	BYOK today (free pass-through); hosted Whisper +$0.10/hr at GA (Q3 2026)	+$0.15 / hr^[2]
storage price	$0 (we don't host recordings; they upload directly to your S3)	+$0.05 / hr / 30 d^[3]
bundled $/hr (today) Apples-to-apples: 1 hr recorded + transcribed + stored 30 days. Today the BYOK path on per-speaker audio is free of meetbot fees but you pay your provider directly. At GA (hosted Whisper, Q3 2026) the bundled rate becomes $0.40/hr us vs $0.65/hr them.	$0.30 bot-hour + your own transcription bill (BYOK)	$0.65
billing granularity	per minute	per minute (rounded)
free tier	first hour, no card	5 hr / mo^[4]
SDK license Recall's hosted API is closed-source. Their public sample repos on github.com/recallai ship without a LICENSE file (effectively all-rights-reserved by default copyright). Ours are MIT.	MIT SDK + samples; proprietary hosted backend	proprietary API + unlicensed sample repos
self-hostable	planned (M5, source available); not committed	no
platforms	Meet, Teams, Zoom (Webex + Discord M2)	Meet, Teams, Zoom, Webex, Slack Huddles, GoTo
data residency	eu-central (Falkenstein) only	US, EU, JP^[5]
compliance certifications	none today (SOC 2 Type 1 on M5 roadmap; no BAA template; no published uptime SLA)	SOC 2, ISO 27001, HIPAA BAA^[6]
SLA	no published SLA (will publish once status.meetbot.dev has real history)	99.9% SLA on enterprise tier
transports	webhook, websocket, RTMP	webhook, websocket, RTMP
mobile SDK Recall has a mobile SDK in market. We do not. Our embedded SDK lands Q3 2026 — ambient mic capture only (Apple/Google block VoIP audio capture from other apps).	shipping Q3 2026 (iOS first)	shipping today^[7]
customers (public)	pre-launch (zero paying customers today)	3,000+

[1]recording price: recall.ai/pricing
[2]transcription price: recall.ai/pricing
[3]storage price: recall.ai/pricing
[4]free tier: recall.ai/pricing
[5]data residency: recall.ai/pricing
[6]compliance certifications: trust.recall.ai/
[7]mobile SDK: recall.ai/product/mobile-recording-sdk

03 · pricing scenarios

The math, three ways.

Three usage points: a hobbyist, a startup, and a scaled company. Formula visible per cell — copy it into a spreadsheet, plug your own numbers in.

scenario 1

Hobbyist · 10 hr / mo

10 hours of meeting recording per month.

meetbot

10 hr × $0.30 = $3.00 (bot-hour only)

$3.00/mo

Recall.ai

10 hr × ($0.50 + $0.15 + $0.05) = $7.00

$7.00/mo

Today: meetbot only bills the bot-hour; transcription (if you want it) is BYOK — pipe per-speaker audio into Whisper/Deepgram/AssemblyAI on your own key. Both providers' free tiers absorb most of this. Recall's 5-hr free tier means real cost ~$3.25; meetbot's first-hour-free means ~$2.70 plus whatever your provider charges for ~10 hrs of audio.

scenario 2

Startup · 1,000 hr / mo

1,000 hours of meeting recording per month.

meetbot

1,000 hr × $0.30 = $300 (bot-hour) + your transcription bill (BYOK)

$300+/mo

Recall.ai

1,000 hr × $0.65 = $650.00

$650/mo

$350/mo of saved bot-hour spend at this scale — minus whatever your transcription provider charges. Whisper-large-v3 self-hosted on a Hetzner GPU box (~$200/mo for an RTX 4090) handles ~20 concurrent realtime streams; Deepgram pay-go is around $0.0043/min ≈ $0.26/hr. Real bundled cost on meetbot today: roughly $300 + $200–260 = $500–560/mo. At GA (hosted Whisper Q3 2026): $400/mo flat.

scenario 3

Scale · 50,000 hr / mo

50,000 hours of meeting recording per month.

meetbot

50,000 hr × $0.30 = $15,000 (bot-hour) + transcription

$15,000+/mo

Recall.ai

50,000 hr × $0.65 = $32,500.00

$32,500/mo

At this scale you self-host Whisper-large-v3 on dedicated GPU infra (a couple of $200–400/mo Hetzner GPU boxes serve thousands of concurrent streams) and the transcription leg approaches its hardware floor. Recall negotiates volume discounts at this scale. We don't yet, but our list price on the bot-hour is already below their post-negotiation floor we've seen quoted.

04 · where they win

Where Recall.ai is the better choice.

We include this section because the alternative — pretending we win everywhere — is dishonest, and dishonest comparison pages are the reason most of them aren't worth reading.

01More platforms. Recall ships Webex, Slack Huddles, and GoTo today; we ship Meet, Teams, and Zoom and have Webex + Discord on the M2 roadmap. If Webex coverage is a hard requirement now, pick Recall.
02Compliance paper. Recall has SOC 2 Type 2, ISO 27001, and signs HIPAA BAAs on the enterprise tier. We have no certifications today; SOC 2 Type 1 is on our M5 roadmap, no BAA template yet. Procurement will ask about this.
03Multi-region residency. Recall offers US, EU, and JP regions; meetbot is eu-central-1 (Falkenstein) only until first US enterprise asks for a region. APJ residency requirements are a hard miss for us today.
04Hosted transcription, today. Recall bills $0.15/hr for transcription bundled with the bot. We don't ship transcription today — BYOK only (per-speaker audio piped into your Whisper/Deepgram/AssemblyAI key, zero meetbot fee on that leg). Our hosted Whisper option ships Q3 2026.
05Mobile SDK, today. Recall has a mobile SDK in market; ours ships Q3 2026 (iOS first). If you need to embed in an iOS or Android app right now, they're the only option.
06Volume references. They have 3,000+ public customers — HubSpot, Calendly, ClickUp et al. — which is a real signal in procurement reviews. We have zero paying customers (we're pre-launch); that's earlier-stage risk.
07OAuth pool depth. Their hosted OAuth credential management for Zoom/Google/Teams is more battle-tested. We ship the same model in M3 but the surface is younger.

05 · where we win

Where meetbot wins.

Each line links to the doc page that proves it. Numbers, not adjectives. Sourced against Recall.ai's public surface as of the date below.

01Lower bot-hour price ($0.30/hr vs $0.50/hr recording). At GA — once our hosted Whisper transcription lands Q3 — the bundled comparison becomes $0.40/hr (us) vs $0.65/hr (them); today, BYOK transcription on per-speaker audio nets you the same gap with zero meetbot fee on the transcription leg.
proof: /pricing
02MIT-licensed SDK + sample apps (JS today; Python, Go, Rust on the roadmap). Fork them, embed them in dual-licensed products, ship them inside closed-source apps. No ELv2, no BUSL, no all-rights-reserved-by-default-because-we-forgot-a-LICENSE-file.
proof: github.com/meetbot
03EU-hosted by default. Bot containers, control plane, and database all in Hetzner Falkenstein. Recordings upload directly to your S3 bucket — no Atlantic crossing unless you point that bucket at a non-EU region.
proof: /security
04Per-minute billing — billed by the second of meeting time, not by container wall-clock or per-bot fee. A 17-minute call costs 8.5¢.
proof: /pricing
05Public OpenAPI spec at github.com/meetbot/openapi — generate your own client, run it through your own SDK pipeline, audit the surface in CI.
proof: github.com/meetbot/openapi
06Same three-transport delivery model (webhook, WebSocket, RTMP) — your existing Recall consumer code reroutes with one URL change.
proof: /docs/webhooks

06 · migration

The whole switch. Eight lines.

Same shape, same fields, different host. Replace your Recall.ai bot-dispatch call with a meetbot one. Webhook payloads land in the same JSON shape your handler already parses.

Recall.ai (before)ts

// Recall.ai
const res = await fetch("https://us-west-2.recall.ai/api/v1/bot", {
  method: "POST",
  headers: {
    Authorization: `Token ${process.env.RECALL_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    meeting_url: "https://meet.google.com/abc-defg-hij",
    bot_name: "notes",
    transcription_options: { provider: "deepgram" },
    real_time_transcription: { destination_url: WEBHOOK_URL },
  }),
});

meetbot (after)ts

// meetbot — same shape, different host.
// Note: we don't ship transcription today — per-speaker audio
// lands in your bucket; pipe it into your provider of choice.
// Hosted Whisper option lands Q3 2026.
const res = await fetch("https://api.meetbot.dev/api/v1/bot", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.MEETBOT_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    meeting_url: "https://meet.google.com/abc-defg-hij",
    bot_name: "notes",
    // captions: passthrough of platform-native captions (Meet/Teams/Zoom)
    delivery: [{ transport: "webhook", url: WEBHOOK_URL }],
  }),
});

07 · faq

The questions we actually get.

Q.How does the price actually compare?

Today: $0.30/hr (us, bot-hour only) vs $0.50/hr (Recall, recording only). Bundled with transcription + 30-day storage Recall is $0.65/hr; we don't bundle transcription today (BYOK on per-speaker audio is free of meetbot fees, you pay your provider directly). Once our hosted Whisper ships Q3 2026 the apples-to-apples bundled comparison becomes $0.40/hr (us) vs $0.65/hr (them). The pricing scenarios above show today's math; the calculator at /pricing lets you plug in your own assumptions.

Q.How long does the migration actually take?

Most teams switch in an afternoon. The HTTP shape is intentionally compatible — we ship a recall-compat path that accepts Recall's bot-create payload 1:1 and translates internally. Webhook payloads land in the same JSON shape your handler already parses; your verification code reuses an HMAC-SHA256 with a different key.

Q.What if I need Webex or Slack Huddles?

Stay on Recall for those today. Webex lands in M2 (~July 2026); Slack Huddles is something we deliberately don't build (no public API entrypoint exists; the only path is TOS-violating browser automation, which Recall also doesn't do — they wrap the official Slack AI huddle-recap API instead). Discord lands in M2 too.

Q.Do you sign a BAA?

No. We have no BAA template today, no SOC 2, no HIPAA posture. SOC 2 Type 1 is on the M5 roadmap; the BAA template would follow that audit. If you have a HIPAA-bound deployment timeline, this is a real reason to stay on Recall today and re-evaluate us once those land.

Q.Where is the data physically stored?

Bot containers run on Hetzner Falkenstein (DE). Postgres + recording manifests live on the same host. The recording media itself goes to whatever S3-compatible bucket you nominate — if you point that at AWS Frankfurt, the entire pipeline stays in EU borders. If you point it at AWS us-east-1, only that final hop leaves the EU. We can send a DPA on request (template-stage, not a published certified document).

Q.What about the Recall.ai SLA?

Recall offers a 99.9% SLA on their enterprise tier. We don't publish an uptime number or a contractual SLA — we will once status.meetbot.dev is up with real history. If contractual uptime is a buying criterion today, that's a real reason to stay on Recall.

Q.How open is the source, exactly?

The SDK (JS today; Python, Go, Rust on the roadmap), CLI, sample apps, and OpenAPI spec are MIT — clone, fork, ship. The bot binary itself stays closed in our monorepo (apps/bot) — the same Vercel/Next.js shape. We may publish a source-available bot in M5 under a permissive license; not committed yet. By contrast, Recall's hosted API is fully proprietary and their public sample repos on GitHub ship without a LICENSE file (which under default copyright is all-rights-reserved — you cannot legally fork them).