About
Why Latentflows exists and how I work.
Short version
I’m Rasmus Sikk. I’ve spent a decade inside B2B SaaS revenue systems and two years building AI products, and Latentflows is those two disciplines in one person.
Rasmus Sikk, founder
How I got here
Enterprise software consulting
I started in a CRM-focused team at a large software consulting firm, working on Salesforce implementations for enterprise customers alongside data science and UX teams.
That early exposure taught me to read a revenue org through its actual system state rather than its org chart or strategy deck. It’s still the first thing I do on any engagement.
Salesforce consulting agency (six years)
I founded and ran a Salesforce consulting agency focused on B2B SaaS companies between $5M and $100M ARR, grew it to eight people, and shipped dozens of engagements across CRM architecture, CPQ, billing, integrations, reporting, and data cleanup. Along the way I certified as a Salesforce System Architect, the architect-tier credential.
Six years of operating inside Sales, RevOps, Finance, and CS at that many companies gave me pattern recognition for how scaling B2B SaaS revenue orgs actually break. The same three to five things fail, in roughly the same order, at roughly the same ARR, and I can usually see them before they are described to me.
The Q2C product that didn’t land
I tried to turn the agency’s IP into a Salesforce-native quote-to-cash product covering CPQ, CLM, billing, and document generation, and spent six months building it before doing proper customer discovery. It didn’t land.
That was an expensive lesson about distribution and validation, and it’s the reason every Latentflows engagement now starts with a named sponsor, a named audience, and a named target metric before any real work begins.
Say Less (two years)
I built and shipped Say Less, a multimodal search product running over 4M fashion items scraped from Zalando, ASOS, Farfetch, and 15 other retailers. The stack included large-scale scraping infrastructure, LLM workflows on unstructured text and image data, embeddings and vector search, a conversational UX, and both a React web app and a React Native iOS app live on the Apple App Store.
Running production AI end to end, from infrastructure through models to shipped apps, is where the applied AI engineering capability behind Latentflows comes from. The tempo and stakes of revenue work are different, but the engineering muscle transfers directly.
Latentflows
Latentflows is what happens when both disciplines live in one person. Most revenue orgs end up hiring two vendors for this work: one who knows the data and the process, and one who can ship the AI. They argue with each other, and the engagement drags.
At Latentflows I do both inside a single scoped project. One person understands the revenue system, builds the foundation, and ships the AI on top, which is how every lesson on this page compounds into the current offer.
How I work
- Distribution-first. Every deliverable has a named audience and a named metric, because I’ve learned the expensive way that building in isolation kills companies.
- Builder, not strategist. You own the strategy for your function, and I build the system that executes it.
- Scoped and measured. Every engagement has a fixed scope, a fixed timeline, and named target metrics. Open-ended consulting misaligns incentives, because the longer it runs the more the consultant earns regardless of whether the metric actually moves.
- Foundation before agents. Audit first, data and process second, context layer third, and agents last, because most pilots fail when this order gets inverted.
Get in touch
If anything here resonates, I’d be glad to hear from you. A short intro call is usually the easiest way to start, and email works equally well if that’s more your speed.