Ticket office
No meter, no tokens, no end-of-month surprise. Solo is free because your machine does the work — we start charging when your team does.
In short
Luge is flat-priced: Solo free (your key or your local models), Team at $29/seat/month, managed sovereign modules per team (Notetaker +$45, AI Assistant +$89, Voice agent +$149) on Canadian infrastructure. Never billed per token — the only metered line is voice overage ($29/h beyond the included 8 h). Partner grid under NDA.
Free
One human. Actually free.
The honest limit: one human, and inference runs on your key or your machines.
Download$29/seat/mo
Billed annually · $35 month-to-month · from the 2nd human
Priced per seat — never per token. Inference stays yours (BYO-AI), or add a managed sovereign module below.
Start Solo, grow laterThe semi-turnkey option
Don't feel like managing GPUs or API keys? We provide inference on Canadian infrastructure managed by us — we can evolve it under the hood without your price moving. Modules add onto the Team plan, per team (one active advisor included; +$20/+$40/+$70 per additional active advisor depending on the module). Each level includes the previous one.
+$45/team/mo
Listens, transcribes and summarizes your meetings. 20 h of transcription/month.
+$89/team/mo
Everything in Notetaker + agentic work: forms, follow-ups, file search.
+$149/team/mo
Everything in Assistant + voice: it calls and answers. 8 voice hours/month, then $29/h.
Models included: Qwen3.5/3.6, Ornith 1.0, Cohere Transcribe — open source, hosted in Canada. Small and mid-size models that excel at everyday office work thanks to our small-verified-steps workflows. Not frontier magic — reliable work. The honest details are in the local models guide; need bigger? BYOK Claude/GPT, as an explicit choice.
Resellers and integrators: Luge resells under your brand — full white-label, from the app name down to the audio server. The partner grid (platform, software seats, inference) is shared on request, under a non-disclosure agreement. Our first vertical partner: Tchat N Sign (financial services).
Enterprise: dedicated sovereign inference tenant (full isolation), custom deployments — by quote.
The day you invite a colleague. Your Solo workspace becomes a Team workspace — agents, channels and documents become shareable, nothing is lost. First month of the 2nd seat free.
The day someone asks "where is this data processed?" — or the day managing API keys gets old. Switch on a managed module: same app, same agents, inference in Canada, flat price.
The day you want to resell it to YOUR clients. Full brand pack, SDK, wholesale grid under NDA. Write to us.
Straight answers
You connect your own Anthropic, OpenAI or Azure key — or run small models entirely on your machine. Luge charges for the employee, not the electricity in its brain.
Small and mid-size open-source models — the class that runs on a very good workstation, not frontier models: Qwen3.5/3.6, Ornith 1.0 (agentic coding), Cohere Transcribe for voice. Open source so everything is verifiable, hosted in Canada for Law 25 and PIPEDA. That model class is exactly why our workflows run in small verified steps — a small model doing one precise job is reliable. Need frontier-level reasoning? Plug in your Claude or GPT key (BYOK) — but that data leaves for the provider, as your explicit choice. The full honest breakdown is in our local models guide.
Canadian inference infrastructure, managed by us. We can evolve the supplier under the hood — your price, your contract and the Canadian residency of your data don’t move.
On Solo with local models: essentially nothing for inference and transcription. With your own key: only the requests you send to your provider, under your own agreement. With managed sovereign modules: your requests run on Canadian infrastructure. The visual breakdown lives on the privacy page.
Because the expensive part — inference — happens on your hardware, on your key, or on small models orchestrated into short steps. We don’t pay unpredictable per-token bills, so we don’t charge you that way either. The only metered line: voice overage beyond the included 8 hours.
No. Download a small local model in the app and everything runs on-device — or take a managed sovereign module and we handle it all. A key just gives you access to bigger brains when a task calls for one.
Luge is designed to make your compliance easier: local processing, personal-information detection, audit logs, and an option to run inference on infrastructure hosted in Canada. Compliance with Law 25 remains an organizational responsibility — no software can "guarantee" it.
Luge’s first vertical deployment is operated by Tchat N Sign, whose organization holds SOC 2 Type II, ISO 27001 and ISO 27701 certifications.
Yes. The partner grid (platform, software seats, inference) is shared on request, under a non-disclosure agreement. Write to us: luge@nsolutions.ai.
Your workspace switches to the Team plan; agents, channels and documents become shareable. Nothing is deleted, nothing resets. First month of the second seat is free.
Recording happens on your device, like any screen recorder. Consent rules for recording conversations still apply in your jurisdiction — informing participants stays your responsibility.
Month-to-month, cancel anytime. Your local data stays on your machine — it’s yours. For a hosted workspace, export before you close it.
Compliance note: Luge is designed to make your compliance easier (Law 25, PIPEDA) — local processing, PI detection, audit logs, Canadian infrastructure as an option. No software can "guarantee" your compliance: it remains an organizational responsibility.