Skip to content
AtheronLABS

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

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

Franchise AI tools

Marketing pre-checks and answers from your manual.

Franchise head offices answer the same questions and check the same flyers again and again. A model that flags brand problems before a person reviews, and an assistant that answers from your own manual, take the repetition off your team while people still make every decision.

// what matters here

What is different in this industry.

01

Flags, not approvals

The marketing model points to likely problems and the rule they break. Your team approves every piece.

02

Trained on your own decisions

A small model fine-tuned on pieces your team has approved and returned, so it learns your brand, not a general one.

03

Answers that cite the manual

The support assistant answers only from your manual and guides, and names the page.

04

Your manual stays private

Both run on models in your own cloud account, so confidential material stays out of other systems.

// 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 appAICustom LLM

Local marketing pre-check

Hypothetical. Not a client, not a result.

// The problem

Franchisees send hundreds of local flyers, posters and social posts a season for approval, and the marketing team checks logos, colours, required wording and offers by eye.

// What we would build

A submission portal where a small model, fine-tuned on your own approved and rejected pieces, flags likely problems before a person looks: an old logo, a missing line of required wording, an offer outside the national promotion. A person on your marketing team still approves every piece. We use a custom fine-tuned model because the task is narrow, the examples are yours, and it runs privately at low cost per check.

// What is in it

  • Franchisees upload a flyer, poster or post with its run dates
  • A fine-tuned model flags logo, colour, wording and offer problems
  • Each flag shows the rule in your brand guide it comes from
  • Your marketing team approves or returns every piece
  • Approved and rejected pieces feed the next round of training
  • An evaluation set run before every new model version

// Stack

  • Fine-tuned open model
  • vLLM
  • Next.js
  • PostgreSQL
  • DigitalOcean GPU

// estimate

Build
≈ US$120,000 to US$184,000, delivered within 31 weeksCAD 171,500 to 261,600
Hosting
≈ US$4,360 a monthCAD 6,210 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
Custom LLM
Model running cost
≈ US$4,190 a monthCAD 5,960 a month

Timeline by milestone

Discovery
2.9 to 3.4 weeks
Data and evaluation plan
0.4 weeks
Design approved
2.4 to 3.9 weeks
Core features
1.9 to 2.9 weeks
Full build
1.4 to 1.9 weeks
Training
6.4 to 10.9 weeks
Testing and fixes
0.4 weeks
Evaluation and red-teaming
3.9 to 4.4 weeks
Launch
0.9 to 1.4 weeks
Production
0.9 to 1.4 weeks

How it is paid

Deposit 20%
CAD 34,060 to 52,080
Discovery 1.5%
CAD 2,554.50 to 3,906.00
Data and evaluation plan 7.9%
CAD 13,453.70 to 20,571.60
Design approved 1.5%
CAD 2,554.50 to 3,906.00
Core features 4.6%
CAD 7,833.80 to 11,978.40
Full build 3.1%
CAD 5,279.30 to 8,072.40
Training 23.9%
CAD 40,701.70 to 62,235.60
Testing and fixes 1.5%
CAD 2,554.50 to 3,906.00
Evaluation and red-teaming 15.9%
CAD 27,077.70 to 41,403.60
Launch 1.5%
CAD 2,554.50 to 3,906.00
Production 8.6%
CAD 14,645.80 to 22,394.40
Holdback, 30 days after launch (10%)
CAD 17,030 to 26,040
GPU time for training and testing, at cost
CAD 1,200. Billed up front, at cost, outside the milestones
Example projectWeb appAIOpen-source model

Franchisee support centre

Hypothetical. Not a client, not a result.

// The problem

Head office's support line answers the same questions from franchisees and their staff all day, in English, French and Spanish, about equipment, recipes, procedures and systems.

// What we would build

A support centre with articles drawn from your operations manual and equipment guides, in English, French and Spanish, and an assistant that answers from them, cites the page and hands harder questions to your support team. We run an open-source model in your own cloud account, because the manual is confidential and stays out of anyone else's systems.

// What is in it

  • Support articles written from your manual and equipment guides
  • English, French and Spanish, with native review
  • An assistant that answers only from your articles and cites the page
  • Hand-over to your support team with the conversation attached
  • An open-source model in your own cloud account
  • A report of the questions it could not answer

// Stack

  • Llama or Mistral (open-source)
  • vLLM
  • PostgreSQL with pgvector
  • Next.js
  • Crowdin

// estimate

Build
≈ US$105,000 to US$161,000, delivered within 22 weeksCAD 149,800 to 229,100
Hosting
≈ US$4,360 a monthCAD 6,210 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$4,190 a monthCAD 5,960 a month

Timeline by milestone

Discovery
2.9 to 3.4 weeks
Specification and evaluation plan
0.4 weeks
Design approved
2.4 to 3.9 weeks
Core features
1.4 to 1.9 weeks
Full build
0.9 to 1.4 weeks
Working pilot
3.9 to 6.9 weeks
Testing and fixes
0.9 to 1.9 weeks
Launch
0.9 to 1.4 weeks
Production
0.9 to 1.4 weeks

How it is paid

Deposit 20%
CAD 29,900 to 45,760
Discovery 1.8%
CAD 2,691.00 to 4,118.40
Specification and evaluation plan 10.5%
CAD 15,697.50 to 24,024.00
Design approved 1.8%
CAD 2,691.00 to 4,118.40
Core features 5.6%
CAD 8,372.00 to 12,812.80
Full build 3.7%
CAD 5,531.50 to 8,465.60
Working pilot 31.7%
CAD 47,391.50 to 72,529.60
Testing and fixes 1.8%
CAD 2,691.00 to 4,118.40
Launch 1.8%
CAD 2,691.00 to 4,118.40
Production 11.3%
CAD 16,893.50 to 25,854.40
Holdback, 30 days after launch (10%)
CAD 14,950 to 22,880
GPU time for training and testing, at cost
CAD 300. Billed up front, at cost, outside the milestones

// questions

Questions we get about this.

Can the marketing model approve a franchisee's flyer by itself?

No. It flags problems for your marketing team, who approve or return every piece.

How many past flyers does the pre-check model need?

A few hundred approved and returned pieces is a good start. We look at what you have in discovery and say if it is enough.

Which languages can the franchisee support assistant answer in?

The languages your articles are written in, such as English, French and Spanish, each reviewed by a native speaker.

// 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 for franchisors: local marketing pre-checks and support assistants | Atheron Network Labs