For Canadian law firms · 3 to 50 lawyers

AI for your firm without the privacy headache.

You already trust Microsoft 365 and Clio because the contracts, encryption, and compliance posture are settled. We've done the same work for AI — built on hardware we own in Canada, in a form that fits the diligence framework your firm already uses. You get the productivity. We carry the diligence load.

See how it works
The situation you're already in

The reason you haven't adopted AI yet isn't capability.

With Microsoft 365 and Clio, the privacy work is settled. Encryption, contracts, residency, vendor accountability — your firm has a framework, and those vendors fit it. LLMs break the framework. Different threat model, different terms, different vendor. Evaluating that is itself a major piece of work, and most managing partners don't have time to do it. So associates anonymize and paste, partners pretend it isn't happening, or the firm just falls behind. — What we keep hearing on intro calls
01
Harvey starts at ~$390,000 CAD a year.

USD $1,200/seat with a 20-seat minimum and a 12-month commitment. Built for AmLaw 100. The diligence is done — but the price is for firms 10x your size.

02
Consumer tools live in Virginia.

ChatGPT, Claude, Gemini — every prompt and every uploaded document hits US infrastructure. Their TOS doesn't fit your existing privacy framework, so adopting them officially means doing all the diligence work yourself.

03
Generic SMB AI doesn't know law.

Lindy and Artisan are sales tools wearing different hats. They don't speak Clio, don't understand limitation periods, can't draft a Statement of Claim — and they're hosted in the US.

04
So mid-sized Canadian firms wait.

And the firms across the street that figured this out start producing work product 4× faster. That gap compounds quarterly.

What you're actually buying

The privacy work, done.

When a managing partner evaluates a new vendor, they're checking it against a framework: signed agreement, encryption, residency, accountability, audit trail. We've shaped the entire offering to slot into that framework — same questions, same answers, same shape.

Same diligence shape as Microsoft 365 and Clio.

You don't have to invent a new evaluation framework for AI. The deliverables that satisfy your existing privacy posture for your cloud vendors satisfy ours, with one structural upgrade: instead of "data resides in Canadian regions of US-controlled clouds," the data lives on hardware we own outright in a Toronto colocation facility.

One vendor. One contract. One audit log. One Canadian phone number when you have a question.

  • i
    Standard-form DPA

    In the same shape as your Microsoft 365 and Clio agreements. We share it before the pilot signs — your privacy lawyer can read it in 15 minutes.

  • ii
    Architecture diagram for compliance

    One page. Names every system, every data flow, every storage location. Reviewable in 20 minutes by a non-technical compliance officer or your insurer.

  • iii
    Audit log per matter

    Every prompt, every draft, every retrieval — logged against the matter and reviewable. No black box. Same audit posture you already expect from your other vendors.

  • iv
    Single vendor of record

    No third-party LLM API calls, no model providers in the data path, no "subprocessor list" with twelve names. One Canadian company on the contract.

  • v
    Aligned with law society tech-competence rules

    Built around the FLSC Model Code on technological competence and the provincial bar guidance. We bring the documentation; your privacy lawyer confirms the fit.

How it works

Not a chatbot. An operator wired into the systems your firm already runs on.

The bot lives inside Clio, Microsoft 365, and OneDrive — the systems your firm already pays for. No new platform to log into, no document migration, no IT project. Same way one of your paralegals would work, except it's available at 11 p.m. on a Tuesday and it remembers every file your firm has ever opened.

i

Drafts in your firm's voice

Wills, deeds, separation agreements, retainer letters, demand letters, statements of adjustments — drafted from your matter file in the voice of the lawyer who'll sign it. Learned from your firm's actual prior work product. You edit; it learns.

ii

Reads your file, drafts the summary

Ingests opposing counsel's productions, OCRs scanned material, identifies the pages that actually matter, flags inconsistencies with prior testimony in the file, drafts the memo to lead counsel.

iii

Lives in Clio, Outlook, and OneDrive

Reads matters, contacts, calendars, and email. Drafts time entries. Files emails to the right matter folder. Watches limitation periods. Wired into the systems your firm already pays for, not a new platform to log into.

iv

Indexes every matter your firm has ever handled

Every retainer, every motion, every settlement letter, every will your firm has filed becomes searchable institutional memory. New associate asks "have we done this kind of file before?" — they get an answer in seconds, not days.

The part competitors can't copy

It gets measurably better at your firm every week.

Most legal AI is the same product for every firm. After 90 days at your firm, ours isn't.

01 / OBSERVE

Every edit becomes signal

When a lawyer rewrites a paragraph, the bot logs the diff. When a draft gets accepted unchanged, that's signal too. Every interaction is training data.

02 / ANALYZE

Weekly post-run review

Every Friday, we run analysis on the week's outputs. What got rejected, why, and what pattern is emerging. Bad memory gets pruned. Good patterns get reinforced.

03 / TUNE

Quarterly firm-specific retune

Every three months, we retrain the bot on your firm's accumulated work product. After a year, it writes more like your firm than a new associate would after 18 months.

04 / REPORT

You see what changed

Every Monday morning, the managing partner gets a one-page report: what the bot got right, what got edited, what it learned. No black box. No "trust us."

Other vendors sell you a model. We sell you a bot that knows your firm — and gets sharper at it every week the partners use it.
What it does today, what's coming

We're not promising you a research robot. Yet.

Most legal AI tools sell themselves as everything-machines. We'd rather tell you exactly what the bot does today and exactly what's on the roadmap — so you can make a real decision instead of a hopeful one.

Shipping today

Drafting and matter intelligence

The v1 product is built around the work that takes the most paralegal hours and the least judgment to automate well:

  • Drafting wills, deeds, separation agreements, retainer letters, demand letters, statements of adjustments
  • Summarizing discovery, productions, medical records, prior testimony
  • Indexing and searching every prior matter your firm has handled
  • Watching limitation periods, drafting client communications, prepping intakes
  • Learning your firm's voice from every lawyer edit
How we compare

The category that didn't exist for firms your size — until now.

There's the giant tier built for AmLaw 100. There's the consumer tier hosted in the US. There's nothing in between for Canadian firms with 3–50 lawyers. We built the in-between.

 
Harvey
Consumer AI
(ChatGPT / Claude)
Floor pricing
USD $288K / yr min
(~$390K CAD)
$25 / seat / mo
Data residency
US-hosted
US-hosted
Built for legal work
Yes
No
Built around Clio + M365
Limited
No
Learns your firm's voice
Generic across firms
No
Humans on call
Enterprise CSM
Submit a ticket
Why we can do this

"If this is so obvious, why isn't Harvey doing it?"

It's the right question to ask. The answer is structural — and it's why a 17-year-old Canadian managed-hosting company is better positioned to serve your firm than a $15-billion AI startup.

$0

Not venture-funded

The big AI agent companies have raised billions of dollars at multi-billion-dollar valuations. They cannot profitably serve a 12-lawyer firm — the math forces them upmarket. We can. We've been profitable for 17 years.

17yrs

Already running infra

We own the colo. We own the racks. Adding GPU capacity is incremental cost on a stack we already operate. The privacy story isn't marketing — it's our actual architecture.

1:1

Humans, not tickets

We've been keeping Canadian businesses' production systems alive since 2009. When something breaks at 11 p.m., you call the same on-call team that's kept your hosting up — not an offshore queue.

We don't sell AI. We sell a job done — on hardware we own, in Canada, with humans on call.
The ask

Don't buy. Pilot.

We're not asking you to commit to anything today. We're asking for 90 days with one workflow. If it isn't saving your paralegals 10+ hours a week by day 90, we shut it off.

  • Week 1–2We integrate with your practice management and document systems. Tune on 50 of your prior matters.
  • Week 3–4The bot starts drafting. Your lawyers edit. Every edit feeds the learning loop.
  • Week 12You see the report. Hours saved. Drafts accepted. Time-to-first-good-draft trend. You decide.
90-day pilot

One workflow. One firm. One flat fee.

$750/mo
Pilot rate. Standard pricing $2,500/mo flat after pilot.
  • Integration with your Clio and Microsoft 365
  • Custom tune on 50 of your prior matters
  • One workflow (e.g. wills, real estate closings, separation agreements)
  • Weekly learning report to managing partner
  • Direct line to our on-call ops team
  • Cancel any time after 90 days. No questions.
Honest answers

What managing partners actually ask us.

$2,500 a month seems like a lot.
It's about 21 paralegal hours, across the whole firm, per month — at roughly $120/hr loaded. We typically see that level of time savings within the first month. A single Land Title migration file or a complex estate intake gets you most of the way there. Compared to Harvey at roughly $390,000 CAD a year for a firm your size (USD $1,200/seat × 20-seat minimum), the math isn't close.
How is the privacy posture actually different from Microsoft 365 or Clio?
Same shape, on Canadian hardware we own outright. You sign one DPA, in the same form as the M365 and Clio agreements you already have on file. Single vendor of record. Single audit log. We give you an architecture diagram a compliance officer can review in 20 minutes. The structural difference: with M365 and Clio, your data is on US-controlled clouds with a Canadian region option; with us, it's on physical servers we own in a Toronto colocation facility, full stop. No third-party LLM API in the data path.
What if your AI gets something wrong?
The bot drafts. Lawyers sign. Every output has a full audit trail — what was generated, what was edited, what was sent. The lawyer is always in the loop, always the one whose name goes on the letter. We treat the bot as a tireless junior associate, not an autonomous decision-maker. We're working through the right insurance and indemnity arrangements as part of every pilot — happy to walk through the specifics on a call.
What if vmfarms goes out of business?
We've been profitable for 17 years running managed hosting for Canadian businesses. The bot is one product line, not the company. If we ever wind it down, we give you 12 months' notice, full export of your tuned model and matter index, and migration support. We're not a startup with 18 months of runway.
Does the bot do legal research? Does it know the law?
Honestly: no, not in v1. The bot is built around drafting and matter intelligence — taking the work product your firm has already produced and the facts in your matter file, and turning that into solid first drafts. For legal research and opinion work, your lawyers do that today the same way they always have. We're working on a v2 that adds licensed legal database access for caselaw research with proper citations, but we'd rather ship a focused tool that does drafting well than an everything-machine that hallucinates the law.
How do you actually keep my data on Canadian soil?
Our GPU servers are in a Toronto colocation facility we've operated out of for years. Your matter data never touches AWS, Azure, GCP, OpenAI, Anthropic, or any third-party API. The model runs on our hardware; the document index lives on our hardware; the inference happens on our hardware. We can give your IT and your bar association compliance officer a full architecture diagram.
What does the integration actually look like on day one?
We connect to Clio, Microsoft 365 (Outlook), and OneDrive — the stack most small Canadian firms already run on. We import (with your consent) a representative set of prior matters for tuning. We set up audit logging, role-based access, and a partner-review queue. Most pilots are running and producing first drafts within two to four weeks of signing.