AI-native SaaS andproduct engineering
I designed and built Krumzi, an AI-native SaaS where AI is the core product experience. I help founders and product teams build AI-native products, or add substantial AI workflows to the SaaS they already have. Remote B2B, Europe and US.
THIN INTEGRATION VS AI-NATIVE
Why it mattersThin LLM integration
- A text box that calls a model API
- Output is a blob of text or a flat image
- Users copy, paste and fix by hand
- AI sits next to the product
AI-native product
- AI makes real decisions inside a product system
- Output is structured data the product understands
- Users keep editing what the AI made
- AI is the core product experience
The loop I design for
Prompt
The user says what they want, in their own words.
AI decision / design system
The AI makes decisions inside rules the product defines.
Structured, editable output
Data the product understands, not a dead end.
User editing loop
Every piece stays editable, so users stay in control.
CAPABILITIES
6 areasAI-native product architecture
Products designed around what the AI does, with the data model, UI and pricing built to match.
Structured AI output
AI that produces data your product understands, like Krumzi's editable design documents, instead of text blobs or flat images.
The editing loop
UX for reviewing, editing and regenerating what the AI made, so users stay in control.
AI workflows in existing SaaS
Multi-step AI pipelines added to a product you already have, as shipped for Churchable.
Usage limits and cost control
Metering, quotas and plan limits so AI costs scale with revenue.
MCP and assistant integrations
Letting assistants like Claude, ChatGPT and Cursor work with your product through MCP.
RELEVANT WORK
Case studies- 01KrumziBuilt an AI-native design platform where a prompt becomes a real, structured design that stays fully editable, not a flat generated image.Founder & Product EngineerAI-native / Structured design output / Browser editor / MCP integration
- 02ChurchableJoined an existing SaaS and shipped an AI-powered workflow that turns long sermon videos into suggested short-form clips.Freelance Product EngineerExisting codebase / AI video / Remotion editor / Media optimization
WHY WORK WITH ME
Credibility- W1I built an AI-native SaaS from zero: Krumzi's AI design system, editor and infrastructure.
- W2I make product and UX decisions about AI, not only API calls.
- W3Proven inside client products too: Churchable's sermon-to-shorts AI workflow.
- Experience
- 7+ years, senior full-stack
- Current
- Full-Stack Engineer at Novoresume
- Core stack
- React, Next.js, TypeScript, Node.js
- Data / infra
- Supabase, PostgreSQL, AWS, Vercel
- Recognition
- Official Remotion Expert ↗
- Enterprise
- WHO, Metro Digital, HSBC, Thales France, Schneider Electric
FAQ
B2B · Time zones · CodebasesQ1What's the difference between adding AI and AI-native?
Adding AI usually means a text box that calls a model. AI-native means the AI makes real decisions inside your product and returns structured output the product understands, so users can keep working with it.
Q2Can you add AI workflows to our existing product?
Yes. Churchable is an example: an AI workflow that turns long sermon videos into suggested clips, built inside their existing Next.js app.
Q3Do you train models?
No. My strength is AI-native product engineering: designing products around AI and integrating models and APIs well. I'm not an ML researcher.
Q4How do you control AI costs?
Usage limits, quotas and plan-based access, like the ones built into Krumzi.
CONTACT
Reply within one business dayBuilding an AI-nativeSaaS product?
I work with SaaS companies and product teams on new products, complex features, AI-native experiences, and React / Next.js / TypeScript development. I’m available for remote B2B contract work, including long-term engagements with teams in Europe and the US.
hello@andreiterteci.com↗