Nick from Public Editor launched his AI-Native SaaS Product 2x faster & increased chances of raising series-A funding by 500%
- Industry
- Education | Media
- Type
- Startup
- Client
- Public Editor
- Technology
- Web
- Location
- Silicon Valley, USA
Increase in investor interest for Series-A
Faster article analysis turnaround
Decrease in Cloud architecture cost
Increase in Engineering speed
Context & Goal
Nick, founder of Public Editor, a startup based out of Silicon Valley, had hired a software design and engineering team a year ago. He was hoping to turn his research project into an AI-Native software product to help fight misinformation and educate people which he can commercialise and build a startup around.
Business Challenge
Nick's problem was
- Inexperienced software design & engineering team
- Team had no background in building software for startups with their specialized needs
- He felt stuck & frustrated with his current team, losing control over his product
- He believed that there ought to be a better way to build the product
The Plan
Nick hired Carte Blanche to
- Audit broken product to identify gaps
- Redesign user-experience (UX) to bring clarity to confusing and broken flows
- Gamify product experience for adoption
- Fix clunky code for scalability
- Bring control and clarity through product management
Transformation
Before
- 12-months & thousands of $ spent, product was still half-finished causing Nick to lose further opportunities to raise funding
- Missed deadlines made Nick look bad to potential funders and partners
- Feeling lost, flailing, out of control, at the mercy of an engineering team that was not transparent, and not always competent
- Cringy, confusing and broken product
After
- Product MVP launched successfully within 6 months
- 5x more investors interested for next round of funding
- 12 Bedrock AI agents annotating any news article end-to-end in one pass
- Credibility analysis turnaround cut from a few days to a few minutes
- Users love Nick's product interface and experience
- Engineering speed increased by 2x & AWS architecture cost reduced by 84%
- Product control & visibility improved by 3x
Deliverables
Why AI Annotation Was Hard to Get Right
A single “rate this article” prompt produces confident but inconsistent, unexplainable scores, unusable for a product built on trustworthy media criticism.
- Full-article analysis exceeded reliable context limits
- Malformed JSON broke the production pipeline
- Generic judgement ignored the team's annotation rubric
- Model-generated scores weren't reproducible or defensible
We didn't prompt harder. We split the analysis into specialised, verifiable steps:
- Specialised agents per dimension
- Rubric-grounded evaluation
- Deterministic, reliable scoring
- Traceable passage-level explanations
A few days of manual annotation became a few minutes of automated analysis.
Core Technologies Used
React
Next.js
Expo
TypeScript
Django
AWS
Docker
Redis
PostgreSQL
DynamoDB
NginX
Tailwind
RadixUI
D3
Vitest
Figma
IllustratorProduct Features Developed
AI Article Annotation
- 12 Claude agents per article
- Weighted thesis & argument extraction
- Sourcing, reasoning, language, evidence & probability analysis
- Rubric-based evaluation against the Public Editor’s dictionary
- Deterministic credibility scoring with passage-level reasoning
- One-click ingestion from browser extension & Reddit
Text Manipulation
- Filter, Search, Sort functionality
- Text scraping & cleaning
- NLP-based textual analysis
- Document parser
- Text highlighting & categorization
Other Product Features
- Guided training paths & user progress tracking
- Gamified learning paths
- Social media sharing
- Data import and export
- User-onboarding
- Auto static-site-generator
- User progress & usage analytics dashboards
- Charts and graphs
Integrations
Amazon Bedrock
Claude
Stripe
Facebook sign-in
Google sign-in
Google AnalyticsBrowser Extensions
Chrome
Firefox
Safari
Edge
Book Your Free 1-Hour
Product Consultation Meeting
- 2 Months
Product MVP launched in 8 weeks
- 200%
Increase in Engineering speed
- 300%
Increase in product control and visibility
- 84%
Decrease in Cloud architecture cost
- 400%
Increase in data accessibility
- 66%
Decrease in management cost due to automation
