AI Engineering Insights
Notes from the engineers at Nortik. What we learn shipping production AI systems for real clients: the decisions that held up, the costs nobody scoped, and the things we'd do differently next time.
What clinicians actually want from AI is two hours of their day back
Not diagnosis engines. The AI that clinicians adopt takes over documentation, inbox triage, and prior authorization, the work that happens after the patient leaves the room.
Ognjen GataloRead post
The martech stack has 15,000 tools. AI earns a slot by deleting handoffs, not adding features
Marketing teams don't need another AI-powered tool. The use cases that stick, from campaign assembly to audience queries to creative testing, remove handoffs between people, not clicks between screens.
Ognjen GataloAugust 19, 20265 min readRead post
Fraud, KYC, and disputes: the unglamorous AI that pays for itself in fintech
Fintech AI headlines chase robo-advice and chatbots. The returns live in the back office: KYC document review, transaction enrichment, dispute handling, and fraud ops, with the audit trail regulators expect.
Ognjen GataloAugust 12, 20266 min readRead post
Why AI prototypes fail in production (and how to ship one that doesn't)
Most AI prototypes stall somewhere between the demo and the deploy. The problem is rarely the model. It's the evals, guardrails, and infrastructure nobody scoped.
Ognjen GataloAugust 5, 20267 min readRead post
The eval suite is the product: testing LLM features like an engineer
Prompt changes without a test harness are blind bets. How we build eval suites from real traffic, and why they outlive every model you'll ever swap in.
Ognjen GataloJuly 21, 20265 min readRead post
What an AI agent actually costs to run in production
Token bills are the visible line item. The real costs hide in the architecture: retries, tool calls, human review loops. A breakdown from live systems.
Ognjen GataloJuly 8, 20266 min readRead post