Who I am
Edvin Fasth, AI engineer and CEO of sailsrep.ai in Stockholm. I build AI-native software for configure-price-quote and guided selling: systems that turn product knowledge into an AI sales assistant that guides customers and reps to the right product — and delivers a complete, valid proposal.
Core expertise
Hybrid AI architecture: large language models combined with a symbolic constraint solver (built on Google OR-Tools), so the LLM handles language, reasoning, and world knowledge while the solver enforces what can and cannot be configured. LLM-agnostic engineering on LangChain — swap the provider, keep the product. Agent design: tools, permissions, ground truth, and feedback loops that keep generation honest. Shipping AI to production: from demo to daily use by real salespeople.
Audience
Founders, sales leaders, and engineers who are curious — or skeptical — about AI agents in B2B sales. Manufacturers who were priced out of traditional CPQ. Anyone who has seen an impressive AI demo and wonders what it takes to make it survive real work.
Perspective
The model is the least interesting part of an AI product. The interesting parts are the boundaries: what the AI is allowed to do, what checks its answers, and what feeds verified state back as ground truth. Fluency is cheap now; verifiability is the scarce asset. And speed changes everything — when an AI-native system goes live in two weeks instead of two years, a project stops being a project and becomes an experiment.
ThoughtOS
