Software and AI engineering
Agents that run where your data is allowed to be.
Dualis Logic is a founder-led studio in Seattle and Bellevue. We build custom tool-calling models and production agent harnesses, deployed in your cloud or on hardware you own.
Sound familiar
- Your agents work in the pilot. Your security team keeps them away from the real data.
- The model guesses at your internal tools. Each wrong argument costs you a turn.
- You pay one API bill a month. It does not price a resolved ticket.
What we do
Three practice areas.
Custom Models & Token Compression
A model trained on your own tool calls, running where your data lives. From $45,000.
Agent Harness Engineering
An agent harness in your repositories. Agents do the volume; your engineers sign triage, spec, review and acceptance. From $38,000.
Compliant AI & Governance
Map the AI your staff use and where the data goes, then give them a workspace you govern. From $12,000.
Operate: we run it, or train your team to. Monthly.
The arithmetic
We price your harness with caching on.
| Uncached | Cached | |
|---|---|---|
| Tool and system tokens per turn | 6,000 | 6,000 |
| Turns per task | 15 | 15 |
| Cost per 1,000 tasks | $180 | $9 |
Claude Sonnet 5.5 list prices on 2026-10-13: $2 per million input tokens, $0.10 cached.
Turn on caching before you buy a model.
Why us
Security review is where you win.
- We have worked inside the frameworks. SOC 2, HIPAA, ISO 27001, PCI DSS and GDPR.
- You own the weights. They live in your account and keep running if we part ways.
Dualis Logic is not itself SOC 2 certified. We are the engineers who know what your auditor is going to ask.
Start with a diagnosis.
Send your turns per task and monthly API spend. The audit prices your harness three ways.