Skip to content
AtheronLABS

You're visiting from the United States. Prices are shown in US dollars. Not right?

labs@atheron:~/industries/municipalities/ai$ train agent --grounded

AI for municipal 311

An assistant for residents, with staff behind it.

Most 311 calls are questions with a published answer: garbage days, parking rules, how to book a field. An assistant that answers from your own pages, and hands everything else to staff, frees the call centre for the calls that need a person.

// what matters here

What is different in this industry.

01

Answers from your published pages

It answers only from bylaws, schedules and service pages you have published, and links the page so a resident can check.

02

Requests, not decisions

It can create a 311 request from a resident's description, which staff accept or reroute. Enforcement and approvals stay with people.

03

Hosted in Canada, in your account

An open-source model runs in a Canadian region in the municipality's own cloud account, so residents' questions stay under your control.

04

Tested before every change

An evaluation set of real resident questions is run before each change, and staff review the results before launch.

// example projects

Priced examples for this service.

Each one is hypothetical, labelled as such, and priced live from our rate card. Open any of them in the estimator and make it yours.

Example projectWeb appMobile appAIOpen-source model

311 requests with an assistant

Hypothetical. Not a client, not a result.

// The problem

A city takes 311 requests by phone, email and social media, logs them in a spreadsheet, and its call centre answers the same questions about garbage days and parking all day.

// What we would build

A 311 web and mobile app where residents report issues with a photo and a location, routing to the right crew, status updates back to the resident, and an assistant on an open-source model hosted in Canada that answers from the city's published pages and creates requests for staff to accept.

// What is in it

  • Requests with photo and map location, from the web or the app
  • Routing to departments and crews, with status sent back to the resident
  • Field crews close requests from their phones, with a photo
  • An assistant that answers from published pages and links its source
  • An open-source model served in a Canadian region in the city's own account

// Stack

  • Next.js
  • React Native
  • PostgreSQL with PostGIS
  • Llama or Mistral (open-source)
  • vLLM
  • AWS Canada

// estimate

Build
≈ US$216,000 to US$330,000, delivered within 41 weeksCAD 307,600 to 470,200
Hosting
≈ US$7,020 a monthCAD 10,000 a month
Support
≈ US$2,500 a monthCAD 3,565 a month

Prices in your currency are estimates from today's Bank of Canada rate. All invoicing is in CAD or USD.

AI route
Open-source model
Model running cost
≈ US$6,530 a monthCAD 9,300 a month

Timeline by milestone

Discovery
3.5 to 4 weeks
Specification and evaluation plan
0.5 weeks
Design approved
8.5 to 14.5 weeks
Core features
1.5 to 3 weeks
Full build
1 to 1.5 weeks
First build on devices
3 to 5 weeks
Working pilot
5.5 to 9.5 weeks
Testing and fixes
0.5 to 1 weeks
Store submission
1 to 1.5 weeks
Retention rules checked with your records manager
1.5 to 3 weeks
Launch
1 to 1.5 weeks
Production
1 to 1.5 weeks

Outside our hands, and added to the calendar

App store review
1 week
GIS and property data access
1 to 4 weeks

How it is paid

Deposit 20%
CAD 61,460 to 93,980
Discovery 4%
CAD 12,292 to 18,796
Specification and evaluation plan 6.1%
CAD 18,745.30 to 28,663.90
Design approved 4%
CAD 12,292 to 18,796
Core features 3.3%
CAD 10,140.90 to 15,506.70
Full build 2.2%
CAD 6,760.60 to 10,337.80
First build on devices 8.7%
CAD 26,735.10 to 40,881.30
Working pilot 18.5%
CAD 56,850.50 to 86,931.50
Testing and fixes 1.1%
CAD 3,380.30 to 5,168.90
Store submission 2.9%
CAD 8,911.70 to 13,627.10
Retention rules checked with your records manager 8.7%
CAD 26,735.10 to 40,881.30
Launch 4%
CAD 12,292 to 18,796
Production 6.5%
CAD 19,974.50 to 30,543.50
Holdback, 30 days after launch (10%)
CAD 30,730 to 46,990
GPU time for training and testing, at cost
CAD 300. Billed up front, at cost, outside the milestones

// questions

Questions we get about this.

What does the 311 assistant do when it does not know?

It says so, and offers to create a request or pass the resident to staff. It does not guess at bylaw interpretations.

Does it work in French?

Yes, when your published pages exist in French. It answers in the language it is asked in, from pages in that language.

Do we need a privacy impact assessment?

Often, yes. We provide the description of what is collected, where it is stored and for how long, for your privacy team to assess.

// next

Not quite your project?

Tell us what you have in mind. We will come back to you with a range and the questions that would narrow it. Or book a call and talk it through.

AI assistant for municipal 311 and resident questions | Atheron Network Labs