●About

We build AI into companies, and co-found new ones.

Descent is a team of AI builders. As an operating partner we build AI into how established companies run; and where deep domain expertise meets a real opportunity, we co-found a new AI-native company and share in what we build. We proved the model inside GoTyme, a $1.5bn digital bank, and we're building with partners today.

Building with

01Inside GoTyme

We proved it inside a $1.5bn bank.

Before Descent, we led GoTyme's AI transformation end to end. A $1.5bn digital bank went from almost no AI to deeply AI-native: 20 million customers across two continents, an AI team grown from 3 to 37 in a year, 1,600+ people trained to work with AI, and more than 70 agentic systems in production. It's the playbook we now bring to the companies we build with.

Start at the top

The first AI workshops were for the board and executives. When leadership sets the expectation, the whole company follows.

A standard for everyone

Every employee had to meet a company-wide AI standard, including building and shipping their own working agent.

Agents into production

A specialist team took the worst back-office bottlenecks and shipped agents that ran them, live and in production.

Embedded in every unit

Those teams moved into every business unit, until working with AI was simply how the whole company ran.

02Why Descent

Why we built Descent.

Andrew Pfaff

Founder

ex-GoTyme · AI & Automation Pod Lead

LinkedIn ↗

David Sithole

Founder

ex-GoTyme · Lead AI Engineer

LinkedIn ↗

Nikos Pasion

Founder

ex-GoTyme · Senior AI Engineer

LinkedIn ↗

What stayed with us was the distance. The gap between people who use AI well and everyone else has become enormous. The top 1% of engineers now ship 46x more code than the median, and Anthropic, worth hundreds of billions, runs its entire growth marketing with one person. It started in software, and it's spreading fast.

The companies that pull ahead will get two things right: people who can direct teams of AI agents, and data and systems underneath that compound as the models improve. Descent exists to build both, inside companies as an operating partner, and from the ground up in the ventures we co-found with domain experts.

The name

Why we're called Descent.

Gradient descent is how almost every modern AI model learns. With each step, the model tunes itself to make its errors a little smaller, moving downhill toward the point where the gradient flattens to zero. It is the quiet mechanism underneath the whole field, and the idea we are named for. How it works ↗

Let's build.

Whether you want AI built into your business or a new company built alongside you, we should talk.

Start a conversation → See open roles