What's the timeline?
A Rescue Sprint runs 2–4 weeks for one clearly defined workflow, depending on its complexity and how much access setup is involved. We confirm the timeline together on the fit call before any work begins.
How do you handle data, privacy, and on-prem?
We can deploy on-site and on-premises so your model and data stay inside your building — including a fully offline, air-gapped option for sensitive work. When we work in your cloud accounts, access is scoped to what the workflow needs, secrets stay out of the repository, and any access can be revoked at handoff. Data-handling terms are agreed in writing before a sprint starts.
Who owns the result?
You do. You get the code, the repository, the deployment path, and the documentation — plus, for on-site builds, the model and the data it's trained on. It's an asset you own, not a subscription you rent.
What tools and stack do you use?
Primarily Python, with RAG and vector search, LLM and multi-agent orchestration, and computer vision (YOLO / TensorRT) — deployed on Docker, GCP, or on-site NVIDIA Jetson and edge hardware. We meet your workflow where it lives rather than forcing a rewrite, including existing no-code/automation setups where that's the right call.
What happens if the scope changes?
A sprint covers exactly one workflow. If we find additional workflows or open-ended app work, that's scoped and quoted separately as its own engagement rather than absorbed silently. This keeps the timeline and price predictable for both of us — with no unlimited revisions.
How does pricing work — and do you guarantee ROI?
Starting points are: AI Stack Audit from $750 (credited toward a sprint if booked within 30 days), Rescue Sprint from $3,500, and Automation Care Retainer from $1,200/month. Each engagement is quoted to its actual scope before we start. We commit to shipping a reliable system — we don't promise a specific ROI, profit, or business outcome.
What about support after launch?
The optional Automation Care Retainer (from $1,200/month, month-to-month) keeps your shipped workflow monitored and maintained, with one small improvement or fix each month and priority async support. It's optional and has no long lock-in — you own the system either way.