FAQs
Straight answers. No buzzwords.
Everything you've wondered about private AI, spelled out clearly. Ask us anything — except for your data. That stays with you.
01 — General Overview
What is FortiqAI?
FortiqAI is a privacy-first AI platform that runs on your infrastructure — on-premises or inside a private cloud — so sensitive documents never leave your control. It combines a user-friendly React web UI with a Node.js backend and local LLMs/vector stores (e.g., Qdrant) to provide search, summarization, Q&A, and document analytics.
Who is FortiqAI for?
Organizations that must protect proprietary or regulated information: law firms, pharmaceutical R&D labs, financial institutions, healthcare providers, government teams, and any business with critical IP or compliance needs.
How is FortiqAI different from cloud-based AI services?
Cloud services send your documents to external servers; FortiqAI keeps everything inside your environment. That brings three benefits: compliance, predictability, and customization.
Do you replace our current DMS, or work alongside it?
Alongside. FortiqAI indexes and answers questions across your documents — it does not become your system of record. Your DMS stays the source of truth.
What happens if the system doesn't know the answer?
It says so. FortiqAI does not generate an answer your documents don't support — an honest "not found in your documents" beats a plausible-sounding guess.
Can the AI act on our systems?
No — by design. It retrieves, compares, extracts, and drafts for human review. It does not approve, release, or close anything on its own.
Are you SOC 2 or ISO 27001 certified?
Not currently. We say so plainly rather than implying otherwise, and we will walk your security team through the deployment architecture directly. See our Security and Compliance pages for detail.
02 — Customization & Services
Can you customize FortiqAI for our specific workflow?
Yes — FortiqAI is modular. Typical customizations include custom connectors, UI changes, automated workflows, and custom extraction types. We start with a short discovery, then deliver a scoped plan: prototype → build → test → handover.
Do you offer fully bespoke AI development?
Absolutely. If your needs go beyond adapting the product, we build custom private-AI systems from scratch — architecture, data pipelines, model fine-tuning, UI, and integrations.
What's the typical customization timeline?
Small changes: ~1–3 weeks. Medium scope: ~3–8 weeks. Large/bespoke: 2–4+ months. We start with a short discovery and deliver a fixed-scope or time-and-materials plan.
03 — Deployment & Technical Details
What does a typical deployment look like?
Containerized packages (Docker / Helm) or VM appliances. Discovery → installation → ingest/connect documents → indexing and model setup → validation and training → go-live.
What hardware and sizing should we plan for?
Typical guidance: a modern multi-core CPU, 32 GB RAM minimum, NVMe SSD storage, and a GPU (NVIDIA T4/A10/A100 class) to accelerate inference for larger workloads.
Do we need an internet connection?
No — FortiqAI can run fully offline once installed. Internet is only needed for optional updates or external integrations.
How do updates and security patches work?
Signed update packages and a documented process: manual patching (air-gapped), scheduled secure update windows, or one-time connections under customer control.
What integrations and data sources can FortiqAI connect to?
PostgreSQL, Qdrant (vector store), SharePoint / network shares, S3-compatible storage, internal DMS, and REST/GraphQL APIs, plus standard file ingestion (PDF, DOCX, PPTX, CSV, OCR).
Where is my data stored and who can access it?
All data remains on your infrastructure, with encryption at rest, TLS in-transit, role-based access control, SSO integration (SAML/LDAP/OIDC), and full audit logs.
Which language models do you use, and can we choose?
FortiqAI is model-agnostic — we deploy open-weight models that run entirely on your hardware, defaulting to permissively licensed families (Apache 2.0 and similar) since they clear enterprise legal review fastest. Your counsel can approve the specific choice.
Who owns the index and the deployment if we stop working with you?
You do. The deployment, the index, and everything in it belong to your organization — including if you choose to stop working with us. We design for exit and portability from day one, not lock-in.
How do you handle documents in Hindi, Gujarati, or mixed-language files?
Multi-language and mixed-language document sets are handled during ingestion and indexing. Exact model behavior for regional languages is confirmed against your actual document set during the Readiness Assessment — we do not promise a blanket capability without testing it on your documents first.
Can it read our scanned archives?
Yes, via OCR. Quality depends on scan resolution and document condition — a Readiness Assessment on a sample of your archive tells you honestly what retrieval quality to expect before you commit to a pilot.
04 — Pricing & Support
What support and training options do you offer?
Support tiers range from standard email support to enterprise SLAs with 24/7 coverage and a dedicated support engineer, plus admin/user training and onboarding materials.
How is pricing structured?
License/instance fee, customization development, optional support & maintenance, and your own infrastructure costs — with discovery + fixed-scope quotes for one-off projects.
What does a pilot cost and how long does it take?
A scoped pilot typically runs 6–10 weeks at a fixed price agreed before you start. See Pricing & Engagement for the full ladder.
What GPU do we need?
Most first deployments run on a single GPU — see Deployment & Sizing for the honest sizing guidance, tier by tier.
How do we get started?
Contact & NDA (if required) → discovery call → discovery workshop → proposal & SOW → optional pilot → deployment & training → ongoing support.
Still Have Questions?
Talk to the team building it.
No sales script — just a straight answer about your specific documents and environment.