Frequently asked questions

What is Ariftly?

Ariftly is a software product company building an AI-native Forward-Deployed Engineer for complex enterprise workflows. Ariftly Forge is the workflow-facing product experience. Ariftly Engine provides the developing intelligence and execution capabilities, while Ariftly Runtime is designed for private, customer-controlled execution. Forward-Deployed Engineering helps customers configure and deploy the product around their systems and workflows.

What workflows are a good fit?

The best starting point is a bounded workflow that crosses systems, contains repeatable technical work, has a clear owner, and produces a result that can be verified.

Oracle SQL → Snowflake is the current local V1 example. Databricks and legacy ETL modernization are near-term directions. ERP migration, EDI onboarding, integration and technical operations are broader future workflow directions. See workflow solutions for the current scope and roadmap distinctions.

What does AI automate, and what still needs human approval?

AI can assist with inventory, analysis, mapping, transformation planning and candidate changes. Deterministic checks verify agreed properties. People review unsupported cases, ambiguous business rules and consequential changes.

Critical work is completed through validation and evidence — not simply because an AI model says it is done.

Does Ariftly replace our existing enterprise tools?

No. Ariftly is designed to work with the vendor, enterprise and open-source tools teams already trust. The workflow determines which tools are used, how they are accessed and where exceptions require additional handling.

Can it run privately or use customer-controlled models?

Yes, private and customer-controlled execution is a core product direction. The current local SQL V1 includes a deterministic path that does not require a model call. Model providers are explicitly configured, and deployment, model access and data boundaries are defined with each customer.

Does the workflow improve over time?

The product is designed to learn from outcomes, exceptions, validation results and human corrections. Potential improvements are evaluated and reviewed before being introduced into production workflows.

What data is retained, and who owns the outputs?

Retention, access, deletion, configuration ownership and delivered artifacts are defined in the written engagement or deployment terms. Those boundaries should be clear before production access is granted.

How are discovery, deployment and pricing handled?

Pricing is quote-based. A typical engagement starts with workflow discovery, followed by a scoped assessment where needed, then a pilot or production proposal. Pricing depends on workflow complexity, systems involved, deployment environment, validation requirements, data volume, human-review requirements and security constraints. See pricing and deployment.