Literature assistant on a private model
Hypothetical. Not a client, not a result.
// The problem
A biotech's scientists spend hours each week searching papers, patents and their own internal reports, and the company will not paste unpublished work into a public chatbot.
// What we would build
A literature assistant on an open-source model that runs on GPUs we set up for you, so nothing leaves your control. It searches the papers you choose, your internal reports and your notebook exports, answers with the passages it used, and says when it does not know. It helps your scientists read; it does not make scientific or medical judgements.
// What is in it
- An open-source model served on GPUs set up for you, never a public chatbot
- Search across papers, patents, internal reports and notebook exports
- Answers that quote and link the passages they used
- Access by role, so confidential programmes stay with their teams
- A set of your scientists' real questions, run before every change
// Stack
- Python
- Open-source LLM
- PostgreSQL with pgvector
- Next.js
- DigitalOcean GPU Droplets
// estimate
- Build
- ≈ US$97,300 to US$149,000, delivered within 20 weeksCAD 138,500 to 211,600
- 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
- 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 3.4 weeks
- Core features
- 1.4 to 1.9 weeks
- Full build
- 0.9 to 1.4 weeks
- Working pilot
- 2.9 to 5.4 weeks
- Testing and fixes
- 0.9 to 1.4 weeks
- Launch
- 0.9 to 1.4 weeks
- Production
- 0.9 to 1.4 weeks
Outside our hands, and added to the calendar
- Test access to your ELN or LIMS
- 1 to 4 weeks
How it is paid
- Deposit 20%
- CAD 27,640 to 42,260
- Discovery 2%
- CAD 2,764 to 4,226
- Specification and evaluation plan 10.3%
- CAD 14,234.60 to 21,763.90
- Design approved 2%
- CAD 2,764 to 4,226
- Core features 6%
- CAD 8,292 to 12,678
- Full build 4%
- CAD 5,528 to 8,452
- Working pilot 31.1%
- CAD 42,980.20 to 65,714.30
- Testing and fixes 2%
- CAD 2,764 to 4,226
- Launch 2%
- CAD 2,764 to 4,226
- Production 10.6%
- CAD 14,649.20 to 22,397.80
- Holdback, 30 days after launch (10%)
- CAD 13,820 to 21,130
- GPU time for training and testing, at cost
- CAD 300. Billed up front, at cost, outside the milestones