Folio VI · Questions
Straight answers.
Working together
Q01What does Byte Zentrix mean?
Byte Zentrix is a name we made up on purpose, from three parts: byte, the smallest unit of information; zen, calm and clarity; and matrix, the grid where everything connects. Together it is the studio's promise: calm, connected order in scattered data, with every answer our AI systems give traced back to the exact document, page and line it came from.
Q02Who will actually build my project?
Hammad Mustafa, the founder. Byte Zentrix is a founder-led studio: you talk to the engineer who designs and builds your system. There are no account managers and no agency markup.
Q03Do you work with clients in the US, UK and Europe?
Yes. The studio is based in Karachi (UTC+5), which overlaps with US mornings and UK afternoons. Calls happen on Zoom or Google Meet, with weekly demos and async updates in between.
Q04How do projects start?
With a paid one-week Discovery Sprint: requirements, architecture and a working prototype on a sample of your documents. The fee is credited to the build if you continue, so the risk stays small on both sides.
Q05Will I own the code?
Yes. You get the full repository, deployment access, documentation and a recorded walkthrough. Nothing is locked to the studio.
Pricing
Q01How much does a document AI assistant cost?
A Document AI Assistant typically costs $1,500–4,000 depending on the number of sources, integrations and accuracy requirements. You get a fixed quote after the Discovery Sprint.
Q02How do payments work?
Milestone-based: a deposit to start, then payments tied to demos you have approved. Invoices are sent through Payoneer or Wise.
Q03Do you charge hourly?
No. Projects are fixed-scope and fixed-price, so you know the cost before work starts. Ongoing work runs on a monthly Care Plan.
Technical
Q01What is RAG, in plain English?
Retrieval-Augmented Generation. Before the AI answers, it searches your documents for the relevant passages and answers using only those. That's what makes citations possible: every answer comes from a specific page.
Q02What happens when the answer isn't in my documents?
The assistant says so. We build a refusal rule into every system: if the retrieved sources don't support an answer, it replies "that isn't in the documents" instead of inventing one.
Q03Which AI models and tools do you use?
Whatever fits your constraints: OpenAI or other frontier model APIs, or open-source models; pgvector, Pinecone or Qdrant for search; FastAPI or Node for the backend; n8n for workflows. We recommend based on your existing stack, not on hype.
Q04How do you measure accuracy?
With an evaluation set: real questions from your team, each paired with the correct answer and source. Every build is scored against it before handoff, and again every month on the Care Plan.
Data & security
Q01Will my documents be used to train AI models?
No. We use providers with zero-data-retention options, or self-hosted open-source models when data can't leave your infrastructure. Your documents stay in storage you control.
Q02Can you sign an NDA?
Yes, before you share any documents. A simple mutual NDA is standard at the start of every Discovery Sprint.
Q03Can it run on our own servers?
Yes. Systems can be deployed to your AWS, GCP or Azure account, or on-premise with open-source models if needed.