About Protabyte
Technical judgment, carried through to working software.
Protabyte is an AI-enabled consulting and product engineering practice led by David DeWitt. Client delivery, technical leadership, and ongoing product development inform the same hands-on approach.
David DeWitt leads Protabyte. He works with clients and builds products, combining technical judgment with AI-assisted engineering to turn complex problems into useful software.
Client work & product building
Each sharpens the other.
Delivery keeps the work grounded.
Client engagements bring real users, inherited systems, deadlines, and operational constraints into the design. Architecture, AI, and modernization decisions have to fit that environment and leave the team able to sustain the result.
Explore federal modernization experienceProduct building tests decisions over time.
Ongoing animal-health, genetics, and financial-research work requires the same decisions to hold up across successive releases: how to model the domain, preserve source meaning, handle incomplete data, and build a workflow people can inspect. Applied research adds another test: whether a persuasive explanation survives examination of the evidence.
Explore products and applied researchHow we work with AI
More capacity to build. Clear ownership of the result.
AI is part of the engineering workflow, from investigation through implementation and review. Technical direction and responsibility for delivery stay with the engineer leading the work.
01
Build the context
- AI assists
- Explore unfamiliar code, trace dependencies, and surface questions to investigate.
- Engineering judgment
- Check those findings against source code, data, and the people who use the system. Decide what matters.
- What your team receives
- A system map, constraints, and an agreed milestone.
02
Implement a complete workflow
- AI assists
- Develop implementation options, assist with changes, and draft tests and documentation.
- Engineering judgment
- Choose the architecture, inspect the changes, and connect the interface, data, permissions, and failure paths.
- What your team receives
- A working increment that completes a user’s task.
03
Check the result in context
- AI assists
- Help investigate failures, identify edge cases, and compare expected and observed behavior.
- Engineering judgment
- Run the checks, review the evidence, resolve gaps, and make the release decision.
- What your team receives
- Validated behavior, known limits, and operational handoff.
Two ways AI creates value
Use AI to engineer software. Engineer useful AI into the product.
An AI-assisted development process can support any software engagement. An AI feature is a separate product decision: it needs useful source data, an evaluation plan, appropriate permissions, and a clear path for human review.
Read how we approach AI engineeringA sample deliverable
See how a technical decision takes shape.
A recommendation, the alternatives behind it, and a first milestone with clear checks. Read an illustrative brief showing what your team can use to make the next decision.
Explore the sample briefStart a conversation
Bring the problem.
We’ll work through the next move.
A product to build, a delivery challenge, a difficult architecture decision, or a government platform that needs a path forward.