FAQ
Straight answers
Common questions about our applied-AI studio, engagement models and what we will — and will not — promise.
Is MindForge AI a course, or does it build fully autonomous AI that replaces my team?
No on both. We are an applied-AI studio that designs and builds custom AI systems for clients — we do not sell courses. We build AI to take repetitive work off people and keep humans in the loop; we do not promise to replace teams and do not guarantee accuracy, cost savings or ROI. Outcomes depend on your data, scope and adoption.
How do engagements work — project vs retainer?
Most clients start with a fixed-scope discovery sprint or proof of concept (project fee in CAD). Production AI assistant builds and automation pipelines are scoped as projects with defined deliverables and timeline. Ongoing MLOps, model evaluation and iteration typically move to a monthly retainer once the system is live. We outline project scope in a statement of work before any build begins.
What are typical CAD budgets?
Discovery sprints: C$8,000–C$15,000. Prototypes and proof-of-concept builds: C$20,000–C$45,000. Production deployments (AI assistants, workflow automation, ML models): C$35,000–C$150,000 depending on integration depth. Retainers for MLOps and support: C$6,000–C$18,000/month. We quote after understanding your data and roadmap — not from a generic price list.
How long does a project take?
Discovery sprints: 2–3 weeks. Prototypes: 4–8 weeks. Production deployment: 8–16 weeks for most AI assistant and automation builds. Timelines depend on data readiness, API access and client review cycles. We publish a roadmap at the end of discovery with realistic milestones — not optimistic guesses.
Which models and tools do you use?
We are model-agnostic. Typical stack includes OpenAI, Anthropic, Azure OpenAI, open-weight LLMs, vector databases for RAG, Python/FastAPI backends, cloud (AWS, GCP, Azure) or on-prem deployments where required. Tooling choices follow your data privacy, latency and budget constraints — not our preferences.
How do you handle data privacy and security?
We follow PIPEDA-compliant practices for Canadian client data. Data processing agreements define retention, access controls and subprocessors. Client data used for fine-tuning requires explicit consent. We support on-prem or VPC-isolated deployments for sensitive workloads. Security reviews are part of every production deployment plan.
Who owns the code, models and IP?
Custom code and configurations built under your project are yours upon final payment, as defined in our statement of work. Pre-existing MindForge tooling and frameworks remain ours. Third-party model weights and API terms follow the provider's licence. We document IP ownership clearly before build starts.
What about accuracy and human oversight?
AI systems err — hallucinations, bias and edge-case failures are expected. We design guardrails, retrieval systems with citations, confidence thresholds and human-in-the-loop review for high-stakes outputs. Model evaluation benchmarks are run before launch. We never promise zero errors or fully autonomous operation without oversight.
What do you NOT do?
We do not sell courses or training bootcamps. We do not offer AI-income schemes, crypto or AI-trading bots. We do not claim AGI or sentient AI capabilities. We do not guarantee ROI, cost savings or accuracy. We do not promise to replace your team. We do not fabricate client logos or partner badges.
Our studio delivers AI strategy, generative-AI applications, assistants, agents, automation and data work for client organizations. Results depend on data quality, scope and adoption — never guaranteed.