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

AI for grant writing

An assistant that reuses your own past proposals.

Grant writing repeats the same history and outcomes in a new shape for each funder. An assistant grounded in your own past proposals gives staff a cited starting point for every answer. A frontier model is the right route here: the material is not sensitive, the volume is small and it costs least to run.

Start a project See the example projects

// what matters here

What is different in this industry.

Your words, reused well

Answers are built from your own past proposals and reports, so they sound like your organisation and say what you have already said accurately.

Every passage has a source

Each answer cites the document it came from, so staff can check facts and figures before anything goes to a funder.

No personal data involved

The library holds program documents and reports, not beneficiary or donor records, so nothing sensitive is sent to the model provider.

Staff stay the authors

The assistant gives a starting point. Staff edit, decide and submit, and funders get an application the organisation stands behind.

// 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 appAIFrontier model

Grant writing assistant

Hypothetical. Not a client, not a result.

// The problem

Staff spend days on each grant application rewriting history, program descriptions and outcomes they have already written many times.

// What we would build

An assistant that answers a funder's questions from the organisation's own past proposals, annual reports and program documents, citing the source for each passage, on a frontier model because the material is not sensitive, the volume is small and it is the least costly route.

// What is in it

Past proposals, annual reports and program descriptions indexed
First versions of answers to a funder's questions, citing the documents used
Word limits and each funder's terms kept in view
No beneficiary or donor personal data in the library or the prompts
Staff review and edit every answer before it is submitted

// Stack

Frontier model API
PostgreSQL with pgvector
Next.js
DigitalOcean

// estimate

Build: ≈ US$80,300 to US$123,000, delivered within 22 weeksCAD 114,300 to 174,800
Hosting: ≈ US$218 a monthCAD 310 a month
Support: ≈ US$1,000 a monthCAD 1,425 a month

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

AI route: Frontier model
Model running cost: ≈ US$42 a monthCAD 60 a month

Timeline by milestone

Discovery: 2.9 to 3.4 weeks
Specification and evaluation plan: 0.4 weeks
Design approved: 2.4 to 3.4 weeks
Core features: 1.4 to 2.4 weeks
Full build: 0.9 to 1.4 weeks
Working pilot: 3.9 to 6.4 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 22,860 to 34,960
Discovery 2%: CAD 2,286 to 3,496
Specification and evaluation plan 10.2%: CAD 11,658.60 to 17,829.60
Design approved 2%: CAD 2,286 to 3,496
Core features 6.2%: CAD 7,086.60 to 10,837.60
Full build 4.1%: CAD 4,686.30 to 7,166.80
Working pilot 30.7%: CAD 35,090.10 to 53,663.60
Testing and fixes 2%: CAD 2,286 to 3,496
Launch 2%: CAD 2,286 to 3,496
Production 10.8%: CAD 12,344.40 to 18,878.40
Holdback, 30 days after launch (10%): CAD 11,430 to 17,480

Start from this example

// questions

Questions we get about this.

Do funders accept applications written with AI?

Policies differ between funders, so check each one. The assistant works from your own material and staff remain the authors, which is what most policies ask for.

Why a frontier model rather than open source?

The documents are not sensitive and the use is light, so a frontier model is cheaper and quicker than hosting your own. If that changes, the assistant can move to an open-source model.

What does a grant assistant cost to run each month?

At a charity's volume, model usage is small; hosting is the larger part. The estimate shows both, separately from the build.

Back to the industry: Nonprofit About the service: AI agents and models

// 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.

Start a project Book a call