A technician holding a phone with the Dezide app open

About Dezide

25 years of guided troubleshooting

We started with a hard question: can you capture what your best troubleshooter knows, and hand it to everyone else? We've been answering it since 2001.

The origin

It began as a research project

We met in the late 1990s as graduate students in a machine learning department, back when “AI” meant math, not marketing. Hewlett-Packard's R&D group was looking for exactly that expertise, and out of it came BATS, the Bayesian Automated Troubleshooting System.

The premise was simple to state and hard to build: expert troubleshooting knowledge shouldn't live in one person's head. It should be captured in a structured form and made available to everyone, everywhere.

BATS

Bayesian Automated Troubleshooting System: the HP research project Dezide grew out of.

The mechanism

We made Bayesian networks practical for the real world

Bayesian networks were powerful in theory and unusable in practice. They couldn't handle the messiness of an actual repair. We researched and patented a set of innovations that changed that, breaking the barriers that had kept the math in the lab.

For the first time, troubleshooting knowledge could be captured the way it actually works: as causes, with their probabilities, weighed against the time and cost of each step. That combination is what turns a model into a decision: the best next step, not just a list of possibilities.

1

Causes

What can actually go wrong, mapped as a structured model.

2

Probabilities

How likely each cause is, updated with every answer.

3

Time

How long a step takes, so the fastest path wins.

4

Cost

What a step costs, so the cheapest route to the fix wins too.

2001

We spun out of HP to put the technology in everyone's hands

By 2001 the research was ready to become a product. After a lot of preparation, and a lot of coffee, we spun off from Hewlett-Packard as an independent company, built and run by people who genuinely care about causal AI, troubleshooting, and knowledge management.

The goal hasn't changed since: take what we invented and make it generally available, to any organization that needs to fix complex things reliably. Their machines treat cancer patients, and every minute of downtime lands on someone's treatment schedule.

2001

2026

Spun out of HP

Trusted across industries