Frequently asked questions.

Not by default, and never without your explicit action. All document ingestion, embedding, retrieval, and generation happen on your local machine. The only network calls are (a) cloud LLM API calls if you configure them, and (b) opt-in support-form submissions where you review what's being sent before it ships.

No. We never receive your RFPs or generated proposals. Even our support form only receives content you explicitly attach, with API keys and license keys auto-redacted client-side before transmission.

Yes. Configure Local as the backend for every consumer and disable the license check-in (offline activation code). No outbound network calls will fire. You will need to download local model files (GGUF format) once from an internet-connected machine.

A rotating session log lives in your app data folder (UserState/session.log). It records every operation the app performs, locally. Nothing is transmitted unless you explicitly attach the log to a support form and hit Send.

Only what you review and approve. The default payload is your typed message plus optional log attachment. Before ship, API keys, license keys, session tokens, and authorization headers are pattern-scrubbed automatically. You see the scrubbed payload before submitting.

Windows 10 (22H2) and Windows 11. Both x64 architecture only. Windows 11 recommended for the latest GPU driver support.

To run local LLMs meaningfully: an NVIDIA GPU with at least 24 GB VRAM. RTX 3090, 4090, or 5090 all work well. For cloud-only usage (Gemini or OpenAI backends), any modern Windows machine will run the app fine — no GPU required.

Only for cloud LLM calls if you use them, and for the initial license activation (once). Everything else runs offline including local model inference, document ingestion, and DOCX generation.

The installer itself: about 3 minutes. First-run setup (initial knowledge-base ingestion, first local model download if using local): 15-30 minutes depending on how much you're loading and your internet speed.

Not on the same machine simultaneously — the app enforces single-instance per machine. Multiple users on separate machines each need their own license.

Beta pricing: 2 months free, then $500/month for the following 10 months. General availability pricing to be announced. Cloud LLM API costs (if you use Gemini or OpenAI backends) are billed directly by those providers.

Per-machine for the beta. One license activates one Windows machine. Contact us for multi-seat organizational pricing.

Yes — that's what the 2-month free beta is for. Request an invite and we'll set you up with a full-featured license.

Email support via the built-in Contact Vendor form, with 1-business-day response target. Bug fixes and feature updates ship as encrypted config updates you load into the app — no re-install needed.

Beta license renews at the same or comparable rate at end of the 12-month period. You'll be notified with plenty of lead time. No auto-charge without your consent.

Varies by RFP length and backend. Typical mid-sized industrial RFP (~50 pages): 4-15 minutes end-to-end. OpenAI GPT-4o is fastest (~5 min); Gemini and local Qwen2.5-32B are comparable (~13-15 min each).

PDF, DOCX, DOC, PPTX, PPT, XLSX, TXT, MD, and common image formats (PNG, JPG, TIFF). Images and embedded diagrams get processed by a vision model.

Yes. All exports are standard DOCX files. Edit freely in Word. When you regenerate a section in the app, the app's version overwrites the file, but you can also keep hand edits by editing the app's in-memory grid before regenerating.

Yes. Per-section regenerate is available from the proposal editor. The Ambassador (per-section writer) can be re-invoked on any single section without touching the others.

No. Prompts ship with the app in your vertical's Sales Pack. Advanced users can view and edit prompts via Settings, but the default behavior is fully turnkey.

A Sales Pack is an encrypted configuration bundle that includes: pre-calibrated LLM parameters, industry-specific prompts, extraction schemas tuned for your vertical's RFPs, and default model choices. It's what makes the app "know" your industry out of the box.

Industrial EPC (engineering, procurement, construction) is the primary calibrated vertical at launch. Additional verticals in development: defense contracting, healthcare services, environmental consulting, and legal services.

Yes. Vertical calibration is our recurring product-development work. If you're an early customer in a new vertical, we'll work with you to build the Sales Pack together, using your (redacted) past-RFP corpus as the calibration set.

Quarterly for major re-calibrations. Ad-hoc updates for prompt refinements or new-model support ship as encrypted config updates any time. Load via Settings → Load PJX, no re-install.

Yes. Every prompt slot and parameter is user-editable via Settings. Your customizations persist through vendor updates unless you explicitly reset to defaults.

Local: any GGUF-format model runnable by llama.cpp — Qwen2.5, Llama 3.x, Mistral, and others. Cloud: OpenAI (GPT-4o, GPT-4-Turbo, GPT-3.5-Turbo) and Google Gemini (2.5-Flash, 2.5-Pro, 1.5-Pro).

Only if you want to use cloud backends. Local-only operation requires no API keys. If you want cloud backends, you supply your own key — we don't proxy or bill for API usage.

Absolutely — that's the design center for the regulated-industry customer. Local Qwen2.5-32B runs the full Analysis and Proposal workflow on a single 24-32 GB GPU with no cloud calls at all.

Fall back to a smaller local model (Qwen2.5-14B or 7B) or use a cloud backend. The VRAM safety solver will warn you before an out-of-memory crash and suggest smaller alternatives.

Yes. Backend selection is per-consumer. Use Local for sensitive extraction (Analysis, RFP Chat) and Gemini for cheaper drafting (Proposal Writer), for example.

ProjEx Advisors is founded by an industry veteran with 38 years in industrial EPC — proposal management, project execution, and business development. The product is built to serve the workflows they've lived firsthand.

Beta: 1 business day for first response. Faster for critical issues (app won't launch, license won't activate). Support requests via the in-app Contact Vendor form or by email to support@projexadvisors.com.

Your data stays on your machine — nothing to migrate. Your license would remain valid for the current term. We're happy to discuss source-code escrow arrangements for enterprise customers with continuity concerns.

Yes, for enterprise customers. Contact us to discuss terms.

sales@projexadvisors.com or via the in-app Contact Vendor form. We keep a public roadmap for beta customers on request.