AI Receptionist for Law Firms in Toronto: A Multilingual Intake Routing Template

Most guides to legal answering services treat language as an add-on. In Toronto it is the first fork in the call. Get the language wrong and the rest of your intake never starts. Get it right and the caller reaches a booked consult, or a fast referral, without waiting for Monday.

What follows is general information for Ontario licensees, not legal advice. Check every rule below against your own practice and the Law Society's current guidance before you switch anything on.

Toronto at a glance: why routing matters here

Two public datasets shape how a Toronto intake line should be built. One counts the firms. The other describes who is calling them.

• Firms are small. Statistics Canada's Canadian Business Counts for June 2026 list 6,179 legal services locations with employees in the Toronto census metropolitan area. Of those, 4,761 had 1 to 4 employees and another 750 had 5 to 9, so roughly 89% had fewer than 10 people. Counts are by location and exclude firms with no employees. Source: Statistics Canada, Table 33-10-1176-01, released August 14, 2026.
• Callers are multilingual. In the 2021 Census, among Toronto-area residents who spoke mainly a language other than English or French at home, the most common were Mandarin (13%), Yue or Cantonese (11%), Punjabi (10%) and Urdu (6%). Source: Statistics Canada, The Daily, August 17, 2022.
• Inside the city itself, 1,244,555 people, or 44.9% of residents, reported speaking at least one non-official language at home. The top non-English home languages were Mandarin, Cantonese, Tagalog, Spanish and Tamil. Source: City of Toronto, 2021 Census Backgrounder: Language.
• Your regulator is down the street. The Law Society of Ontario is based at Osgoode Hall, 130 Queen Street West, which also houses the Court of Appeal for Ontario and the Superior Court of Justice. Source: Law Society of Ontario, Plan your visit.

Put those together and the picture is clear. Most Toronto firms do not have a front desk team, let alone one that covers four or five languages after hours. That is the gap an AI receptionist is meant to fill, under your supervision.

Who this is for

• Solo lawyers and paralegals in Toronto, Mississauga, Brampton, Markham, Richmond Hill, Vaughan or Scarborough who miss calls while in court or with clients.
• Small firms of 2 to 9 people that share one receptionist, or none, and rely on voicemail after 5 p.m.
• Practices with a multilingual client base, such as immigration, family, real estate, wills and estates, and criminal defence, where the first caller is often a relative speaking for someone else.
• Firms that already have a script on paper and want it followed the same way at 2 a.m. as at 2 p.m.

It is not a fit if you want software to assess the merits of a case, quote fees for a contested matter, or decide who you can act for. Those stay with you.

How intake works, step by step

1. Answer. The AI picks up on the first ring, says it is an automated assistant for your firm, gives the recording notice you approved, and asks how it can help.
2. Screen. It confirms the language, asks whether the caller is a current client, and matches the matter to a practice area you handle. If you do not handle it, it says so and offers the referral list you supplied.
3. Gather conflict-check details. It collects the caller's full name and contact details, the full names of the other parties, and any key date. It does not take the story. It does not decide whether there is a conflict. You or your staff run the check.
4. Book. If the matter fits and no urgency rule fires, it offers open consult times from your calendar, then sends a confirmation. If an urgency rule fires, it follows the action you set: transfer, page the on-call lawyer, or give a prepared message.

The template: a multilingual intake call flow

Use this table as a starting draft. The right-hand column is where the lawyer signs off. Nothing in the middle column should go live until that column is filled in.

Language handling in a Toronto setting

Start with the languages your clients actually use. The census figures above suggest Mandarin, Cantonese, Punjabi, Urdu, Tagalog, Spanish and Tamil are common in the area, but your own call log beats any census table. Two cautions apply.

• Screening is not interpretation. An AI that greets and screens in Cantonese is a convenience. It is not a qualified interpreter for the consult itself. Decide in advance when you will book a human interpreter.
• Names are evidence for your conflict check. Have the AI spell back every name and record it in both the caller's script and Latin letters where possible. A misspelled party name is how conflicts slip through.

Urgency rules by practice area

These are the four patterns Toronto small firms most often ask about. Each rule names a trigger the AI listens for and an action you choose. The AI applies the rule; it does not judge the legal stakes.

Keep the trigger list short and literal. Broad triggers like "anything serious" lead to every call being flagged, and then nothing is.

Sample routing rules block for lawyer approval

This is the plain-language rule set we ask a lawyer to mark up before any build starts. It is written to be read by a person first and configured second. Replace every bracketed item.

# Intake routing rules: [Firm name], Toronto
# Approved by: [Lawyer name, LSO number]   Date: [YYYY-MM-DD]
# Review again: every 90 days or after any script change

greeting:
  identify_as: "automated assistant for [Firm name]"
  recording_notice: "This call is recorded and transcribed so we can
    follow up on your enquiry. Say 'no recording' to continue without it."
  never_say: ["best", "guaranteed", "you will win", "you have a case"]

languages:
  supported: [English, Mandarin, Cantonese, Punjabi]   # edit to match your callers
  other_language: offer_callback_with_interpreter
  summary_language: English (note original language)

practice_areas:
  accept: [immigration, real_estate, family, wills_estates]
  refer_out: [criminal]    # or move to accept and set on_call below
  refer_list: [your approved referral list]

conflict_check_fields:
  - caller_full_name (spell back)
  - phone, email
  - other_party_full_names (spell back)
  - matter_type
  - key_date
  stop_detail_line: "Thank you. Please save the details for your
    consultation so we can complete a quick internal check first."

urgency:
  immigration: deadline_within_days: 15 -> flag_urgent, earliest_slot
  real_estate: closing_within_days: 1 -> transfer_or_text [clerk]
  family: caller_unsafe -> say "Please call 911 now" ; court_date -> flag_urgent
  criminal: in_custody -> page [on-call lawyer] ; else read approved_message

hard_limits:
  - no legal advice, no opinion on merits or outcome
  - no statement on whether a conflict exists
  - no fee quotes beyond the published consult fee
  - human handoff offered on request at any point

booking:
  calendar: [calendar name]
  consult_length_minutes: [30]
  confirmations: booking details only, no marketing content

Notice what is missing. There is no field for the facts of the case and no rule that lets the AI pick a lawyer based on merits. Those omissions are the point.

Compliance notes

The table below points to the official source for each issue. It is a checklist, not a legal opinion.

For the full Ontario intake picture, including how to split conflict details from the narrative, read our guide to law firm intake automation in Ontario.

Pricing

These are our published prices. The AI receptionist and intake start at Foundation. Starter covers the website and automated basics only.

All prices are in USD. CAD quotes are available on request. To estimate what unanswered calls might be costing your firm, try the missed-call cost calculator for law firms.