ChatGPT‑like productivity — hosted privately in Canada, with zero data leakage.
Give your firm drafting, summarization, and research superpowers without risking confidentiality. Your data stays in AWS ca‑central‑1, in a single‑tenant environment aligned with PIPEDA and provincial law society expectations. No training on your inputs. Ever.
Draft & Summarize
High‑quality first drafts, case summaries, memos, and emails in minutes. Paste or upload (optional) to accelerate routine work.
Private by Design
Dedicated VPC, per‑tenant compute, private storage, strict deletion policies. Logs limited to operations only.
Simple Interface
Clean, familiar chat UI (Chatbot‑UI / OpenWebUI based) with role presets for common legal tasks.
Why firms choose WebGuru Private LLM
- Data residency in Canada: All compute and storage in AWS Montreal (ca‑central‑1).
- Single‑tenant isolation: Each firm gets its own environment—no co‑mingling of data or workloads.
- PIPEDA‑aligned controls: Encryption, access limits, audit‑friendly configuration, and deletion on request.
- No data reuse: Your prompts/files are never used to train or tune models.
- Predictable cost: Flat monthly pricing, no per‑token surprises for typical usage.
- Fast value: Paid pilot to prove ROI within two weeks, then continue or walk away.
What you get on day one
- Private chat interface with access control (email allow‑list or SSO).
- Modern open models (e.g., Llama‑3.x) served via vLLM/Ollama.
- Secure file handling with strict deletion policy.
- Optional: basic RAG over firm documents (phase‑in).
Security & Compliance Hooks
- Residency: AWS ca‑central‑1 only.
- Encryption: TLS in transit, AES‑256 at rest.
- Isolation: Dedicated VPC, private subnets, per‑tenant storage buckets.
- Identity: SSO or per‑user accounts, least‑privilege access.
- Logging: Minimal operational logs, no prompt/body logging by default.
- Deletion: Admin can trigger secure deletion of chats/files; automatic retention windows configurable.
- Review readiness: Settings align with PIPEDA and provincial law society confidentiality obligations.
Architecture (Pilot)
Llama‑3.x → vLLM/Ollama → Chatbot‑UI in a single‑tenant stack. All components confined to your tenant’s VPC and storage with TLS terminators and WAF option.
How it works
- Discovery (30 mins): Understand matters, workflows, and access needs.
- Pilot setup (3–5 days): Stand up your isolated stack in ca‑central‑1 and provision users.
- Proof week (up to 2 weeks): Real drafting/summarizing tasks with success metrics.
- Decide: Continue monthly or export and delete everything.
Pricing
Pilot
- Isolated environment in ca‑central‑1
- Up to 10 users
- Success checklist + ROI snapshot
Standard
- Single‑tenant hosting + monitoring
- Ongoing security updates
- Fair‑use model inference included*
Founding Client
- Everything in Standard
- Priority input on roadmap
- Referral credit program
* For atypical high‑volume workloads we’ll size dedicated GPUs/CPUs together.
FAQ
How is this different from public ChatGPT or Gemini?
Public tools are multi‑tenant and may move or process data outside Canada. Our stack is single‑tenant in Canada, with strict controls and no training on your data.
Do you support document search (RAG)?
Yes, we can phase in basic RAG over firm documents once the pilot proves value. This remains within your tenant and region.
Which models do you run?
Modern open models like Llama‑3.x via vLLM/Ollama. We’ll benchmark against your tasks and can swap models without moving data.
What about audits and deletion?
We provide admin‑level controls for secure deletion and export. Logs are minimal and operational only; prompt bodies aren’t stored by default.
Contact
Email: varun.mehrotra@webguru.ca · Toronto, Canada