
Preston
“Clear summary and action items automatically.”
Sandblasted aluminum




A visible recording cue
Editable Markdown files





mira_session_read --session latestmira_wiki_resolve --topic "launch-architecture"In today's 42-minute sync, the team reviewed our Q4 infrastructure readiness under a projected 5x peak load (40,000 req/s). We reached unanimous consensus on migrating ingestion from synchronous HTTP to partitioned event streams (Kafka/Redis Streams), locking the blue/green 15-second rollback SLA, and automating secret rotation. All core architecture blockers are cleared for the v2.4 launch.
| Workstream | Deliverable & Acceptance Spec | Owner | Due Date | Status |
|---|---|---|---|---|
| Data Platform | Kafka partition auto-rebalance & dead-letter queue replay harness | Alex M. (Lead) | Oct 22 | Verified |
| Core Infra | Complete dual-write shadow verification on PostgreSQL v16 read-replicas | Sarah T. (Infra) | Oct 25 | In Review |
| Security | VPC egress lockdown, IAM KMS key rotation & penetration sign-off | Dave K. (SecOps) | Oct 27 | In Progress |
| Observability | Deploy OpenTelemetry distributed tracing & PagerDuty Tier-1 escalation | Elena R. (SRE) | Oct 29 | Scheduled |

Pick up where you left off, with a Wiki that connects past conversations and decisions.
What still needs a decision before launch?
What did we decide just now?
What changed since our last conversation?
List owners and next steps from the standup
Draft a follow-up email from our meeting
Which follow-ups did I promise the client?
Which themes came up across these interviews?
Turn this interview into an article outline
Where did the interviewees disagree?
Pull together the requirements from these calls
Why did we choose this approach?
Bring last week’s decisions into today’s brief
What budget did marketing approve?
Turn my voice notes into a project outline
Make a mind map of these ideas
Write tests for the edge case we just discussed
Turn my lecture notes into a study outline
Add these constraints to the implementation plan
What did we decide about the API?
Write release notes from the product sync
Explain the architecture we discussed at the whiteboard
List action items by owner
What should we revisit at our next meeting?
Draft an implementation plan from this discussion


Recorder + local AI

Record, then listen back

Notes from supported meetings
| Question | OneMira C1 | Traditional recorders | Meeting AI apps |
|---|---|---|---|
| What can I record? | Meetings, interviews, lectures & headphone calls (no bot) | Nearby audio, usually kept as recordings | Meetings and calls; support varies |
| Who writes the notes? | Built-in offline AI (or optional ChatGPT, Claude, Gemini & Cursor) | You or separate software | Built-in AI; integrations vary |
| What can I keep? | Auto-organized Wiki, mind maps & editable local files | Your audio recordings | Notes in the app; export options vary |
| Can I keep the recording? | Yes, stereo WAV (your choice in Settings) | Kept until you delete it | Depends on the app and plan |
| Is there a monthly fee? | $0/mo · No 300-min cap | Usually none for basic recording | Free and paid plans vary |
awesome transcription!
damn bro this is cool
Been looking for something like this for a while. I just wanted a simple device for local transcription without paying for yet another subscription. So far, so good.
Didn’t expect it to feel this well made.
I just got mine but I already love it.
Mostly using it for two- or three-person conversations in my office. It's been a pretty painless addition to the setup.
Didn't expect to say that about a recorder, but yeah. I use it constantly.
Tiny, clean, and doesn't take over the table. Been using it for one-on-ones and project chats. So far, so good.
Meetings in, notes out. That's what I bought it for. It's been solid.
We make a lot of decisions verbally that never make it into the doc. This has been handy for catching the context behind those decisions.
OneMira C1 is an 85 g sandblasted aluminum wireless pocket AI voice recorder and note taker. It transcribes in 99 languages, live or later, and automatically organizes your conversations, people, projects and decisions across meetings into notes, mind maps and a living knowledge base with zero subscription fees.
Press the center button once to capture in-person meetings, client interviews, lectures, online calls—even while wearing AirPods or Bluetooth headphones without a meeting bot—or ideas on the go. Engineers, consultants, executives and researchers can generate summaries, decisions, action items and client commitments offline with built-in local AI, plug in an API key (including Google Gemini, OpenAI and Anthropic), or bring spoken context directly into ChatGPT, Claude and Cursor via Mira CLI.
C1 keeps your phone in your pocket without call interruptions, and gives recording a dedicated one-press control with a quiet rotating turntable. Unlike cloud-bound card recorders that upload your private audio to their cloud and charge monthly minute plans after a 300-minute cap, OneMira runs live transcription and built-in offline AI with zero monthly fees, no meeting bot, and automatic organization across meetings so notes never pile up in a drawer.
No. Local transcription in 99 languages, built-in offline AI, cross-meeting organization and access to your local files are included with $0 monthly fees, no 300-minute monthly cap and no forced cloud uploads.
Practical capacity depends on battery, storage and your computer. Optional third-party AI services have their own fees and limits.
Use OneMira Desktop on a Mac (macOS 12 or later) or a Windows computer (Windows 10 or Windows 11) for transcription, speaker labeling and AI notes. Running the AI on your computer instead of a thin phone app is what keeps transcription 100% offline with $0 monthly subscription fees. C1 connects by Bluetooth 5.4 (LE Audio) or USB-C, and has no Wi-Fi or headphone jack. OneMira Desktop does not require a OneMira account.
Yes. Press the center button once to start recording (the turntable rotates quietly while active) and press once again to stop, or schedule a start and stop time. C1 records standalone onto 32 GB of onboard encrypted storage for up to 30 hours on a charge—no live phone or computer connection needed while recording. When you're back at your desk, reconnect over Bluetooth 5.4 or USB-C to transfer recordings to your Mac or Windows computer for local transcription and AI summaries.
OneMira ships C1 to the United States and other supported countries. Estimated delivery is 1–3 days in the United States and 3–5 days for other supported destinations.
Available shipping methods, rates and taxes collected by OneMira are shown at checkout. For international deliveries, applicable customs duties, VAT and import tariffs are the recipient's responsibility unless checkout specifies otherwise. Carrier logistics, customs clearance, weather and remote destinations can affect delivery estimates.
You may request a return within 30 days of delivery for a refund of the purchase price. C1 must be in good, functional condition, with its original packaging, cable, accessories and documentation.
Email dtc@onemira.com with your order number for return authorization, then ship the product within 14 days of receiving the instructions. You pay return shipping unless the item arrived damaged, defective or incorrect. After inspection, refunds are issued to the original payment method within 5–10 business days.
Your transcripts, summaries and Wiki notes are stored directly on your computer as open, editable Markdown (.md) files—never uploaded to a shared OneMira cloud workspace. You choose in Settings whether to keep the original recordings as standard stereo .WAV files on your computer, with direct USB-C or Bluetooth transfer and no subscription required to access or export your files.
Content you actively choose to send to an optional cloud model, local-network model or external agent follows that service's data terms.
Yes. After the one-time software and model downloads, speech transcription, speaker labeling and built-in AI generation (notes, action items and mind maps) run 100% offline on your Mac or Windows computer—even with Wi-Fi turned off and with no OneMira account required.
Those initial downloads need internet access. Optional cloud models need a provider connection; local-network models need access to their host.
Place C1 within about 5 m of the speakers—such as on a desk or conference table for 2–4 people in a room, or beside a laptop speaker. Two MEMS microphones, an on-chip neural engine and hardware noise cancellation help capture clear voices, while local processing automatically separates and labels different speakers in the transcript.
Results depend on distance, background noise, overlapping speech and room acoustics.
Yes, on supported computers and with the documented setup. Whether you are on Zoom, Microsoft Teams, Google Meet or Slack Huddles—even while wearing AirPods or Bluetooth headphones—C1 can combine your nearby voice with far-end call audio without sending an awkward bot into the meeting or occupying your computer's microphone or audio inputs.
On Mac, far-end capture requires macOS 14.2 or later and the documented Core Audio process-tap setup.
The 1250mAh lithium battery (2x–4x larger than typical 300–500mAh phone-card recorders) provides up to 30 hours of continuous recording, up to 90 days of standby, and fast ~2-hour USB-C charging. C1 includes 32 GB of persistent context storage, enough for 10+ years of typical context use.
That estimate applies to text context, not audio-recording capacity. You choose whether to retain original stereo WAV audio in Settings, and retained recordings use storage separately.
OneMira transcribes speech in 99 languages—including English, Spanish, Italian, German, French, Portuguese, Dutch, Russian, Polish, Chinese, Japanese, Korean, Arabic and Hindi—with local speaker labeling. Availability and accuracy can vary by language, accent, speaker and recording conditions.
Yes. Use your own API key with model services from nearly all major official providers and platforms—including Google Gemini, OpenAI (ChatGPT), Anthropic (Claude), DeepSeek, OpenRouter and local Ollama / LM Studio—or let compatible agents such as ChatGPT, Claude or Cursor work directly with your meeting context through Mira CLI or MCP.
These are optional; Desktop's built-in local AI works independently offline after setup. External services have their own credentials, data terms, limits and fees.
Yes. C1 connects to one computer at a time and syncs in both directions in real time. Complete the initial sync requested by OneMira when you switch computers. Agents on the active computer use its local context through files, CLI or MCP.
You do. C1 itself does not open your repository or edit project files. A connected AI agent works within the access and permissions you give it. You choose what context to share and which proposed changes to review or accept.
Recording laws vary by location and situation, so you are responsible for checking and obtaining any consent the law requires. The safest practical default is to tell participants before capture begins and get their agreement; seek qualified legal advice when the rule is unclear.
C1 alone does not make a workflow GDPR- or HIPAA-compliant. Local transcription can reduce transfer to a vendor-operated transcription cloud, but compliance still depends on consent, retention, access controls, your computer environment, organization, jurisdiction and any third-party AI service you use.
For hardware wireless compliance, C1 is certified under FCC ID: 2BXIX-C1.
OneMira C1 includes a 1-year limited warranty against manufacturing defects, starting on the delivery date. It does not cover damage caused by misuse, accidents or unauthorized modification.
To start a claim, email dtc@onemira.com with your order number and a description of the issue.
Email dtc@onemira.com for help with orders, returns, shipping, warranty, setup or product questions. Evaluating C1 for your sales, consulting, legal or engineering team? Many organizations start by trialing a single C1 covered by our 30-day return policy, or email dtc@onemira.com for 5–20+ unit team pricing, enterprise IT/security Q&A and distribution inquiries. Include your order number when relevant. The current support target is a reply within 1 business day, measured in Hong Kong time (UTC+8).