Frequently asked questions

Caltor is an AI engineering and software firm based in San Francisco. We design and build production AI systems for enterprises: agents, retrieval systems, copilots, decision systems and the software around them. We also do the earlier work of deciding what to build, through use-case discovery and target architecture. What we do not do is strategy decks without a build behind them, or people sold by the head to work a queue under your managers. Every engagement is scoped to a system that ships, with evaluation, security review and observability built in from the start.

It starts with a conversation, usually with one of the founders. You describe the problem; we come back with an architecture and a scoped fee. Most engagements open with a discovery phase: a use-case audit and a target architecture that ends in a costed build plan. That first phase is deliberately small. It lets both sides test the working relationship before committing to a production build, and its output is useful to you even if you build with someone else.

Fees are scoped and agreed in the first conversation, based on the engagement shape and what the system needs to do. We do not publish prices, because two systems with the same name rarely cost the same to build well. What we will always give you is a number before you commit, attached to a defined scope. There are no hourly surprises: if scope changes, we re-scope together and agree the new number before work continues.

You do. Every contract assigns the work product to the client: code, prompts, documentation and infrastructure configuration. We keep only our pre-existing tools and general methods, and those are identified in the agreement so there is no ambiguity later. Nothing we build for you is reused for another client, and nothing is held back to create dependence on us. Handover documentation is part of the deliverable, so your team can operate what it legally owns.

Confidentiality starts with a mutual NDA before we see anything sensitive. During delivery, your data stays in your environment wherever the architecture allows; we work inside your cloud accounts and identity systems rather than copying data out. Security is engineering scope, not an afterthought: threat modeling, access controls and audit trails are in the plan from the start, and we build to the compliance frameworks your industry requires. Your security team is welcome in our reviews at any point.

Every engagement is founder-led from first contact: the partner who scopes your work is your counterpart through delivery. The engineers on the work are senior, which is how we keep genuinely senior people on every engagement at our size, and reviews and decisions are scheduled around your working day. Contractually it is simple: you sign with one US entity, and everything the team produces is assigned to you at source through that contract.

We do not publish delivery schedules, because two systems with the same name rarely take the same work. What we do commit to is the shape of it: the first workflow is scoped to reach production ahead of the rest of the build rather than arriving at the end of it, and you see working software as it lands instead of at a reveal. The schedule itself is set with you in writing before you sign, and if a date you need is not achievable we say so then rather than later.

A large consultancy sells you a team; the partners who won the work rarely build it. A staffing firm sells you people, and the outcome remains your problem. We sell a working system, and where the problem will not hold still long enough to be scoped as one, we place senior engineers inside your organization who answer to us rather than to your line managers. The founders who scope your engagement are accountable for delivering it, and the engineers on the work are senior, not a leverage pyramid. We are small by design, which means we take fewer engagements and cannot hide behind a brand. The system either works in production or we have failed visibly.

The system keeps evolving, and so does the ground under it: models change, usage grows, new capabilities become worth adding. Our Maintenance engagement covers uptime and SLA monitoring, model and prompt upkeep verified by the evaluation suite, senior engineering time reserved for the system, and a standing roadmap review. If you prefer to run the system yourself, handover is built into every build: documentation, runbooks and training for your team are deliverables, not extras. Either way, you are not dependent on us to keep the lights on.

Financial services, automotive, industrial, healthcare, and enterprise technology organizations with complex software estates. These are sectors where the founders have built and operated systems directly, including regulated clinical platforms and security tooling aligned to SOC 2, ISO 27001, HIPAA and PCI-DSS. The common thread is regulation, integration depth and real consequences for wrong answers. If your problem sits outside these sectors but shares those properties, talk to us. If we are not the right firm for it, we will tell you.