Knowledge Base
Everything you need to know about our AI engine, document formatting, and security protocols.
Core Architecture & Output
Unlike rudimentary systems that output flat Markdown text, CareerCraft Uganda utilizes a proprietary semantic parser that generates highly structured Rich-JSON. This JSON is then mathematically mapped into OpenXML coordinates, creating a native Word document with perfectly styled tables, bullet points, headers, and pagination that strictly adheres to corporate standards.
Yes. Our presentation generator does not just dump text onto a slide. It programmatically injects dynamic corporate theme colors and utilizes true PowerPoint Slide Masters. The output file is a native .pptx that is 100% editable in Microsoft Office or Google Slides.
Extremely. The XLSX compiler is instructed to map currency arrays directly into Excel's accounting format, lock header rows, and apply standard styling like zebra-striping. This is not CSV; it is fully featured Excel architecture.
Yes. When querying the Academic Research engine, the LLM pipeline is strictly prompted to utilize APA 7th Edition (or Harvard) formatting, injecting theoretical frameworks and automatically compiling a 'References' block with 10-15 formatted citations.
Security & Confidentiality
Never. We enforce a zero-retention policy across our LLM infrastructure. Prompts are processed strictly in-memory during generation and are instantaneously purged. We do not - and will not - train models on your proprietary data.
All user accounts and document references are stored in our Supabase clusters utilizing military-grade AES-256 encryption. Our database rules strictly enforce Row Level Security (RLS) meaning it is mathematically impossible for another tenant to query your records.
Yes, 100% of data transmission - from your browser to our API layer - is secured via TLS 1.3 encryption protocols. Interception is structurally prevented.
Billing & Subscriptions
Instead of locking you into a subscription, our Pay-As-You-Go system charges you exactly for what you need. You simply purchase a specific volume of pages. These credits never expire and are only deducted upon successful document generation.
Absolutely. When you upgrade, your account immediately unlocks the new tier limits. Any remaining time on your previous tier is automatically prorated as a credit toward your new invoice.
If you pay in UGX, the price you see is exactly the price you pay. For USD payments, a standard 15% cross-border processing markup is applied to cover international gateway fees.
Because AI generation incurs hard compute costs on our servers, we generally do not offer refunds for completed generations. However, if a systemic bug prevents your document from compiling, our support team will manually credit your account.
AI Limitations & Best Practices
While our models are prompted heavily for high accuracy and zero-hallucination output, LLMs are inherently probabilistic. You should always review academic statistics, financial projections, and legal clauses generated by the engine before distributing them.
Currently, our streaming engine supports generating continuous documents up to 50 pages in a single request. For larger outputs, we recommend generating chapters sequentially using our structured templates.