Working with Calfy
What does Calfy actually do?
We design, build, and run custom AI agents and automations that work inside the systems your business already uses — your CRM, inbox, calendar, help desk, or database. That means software that does the work, not a chatbot that talks about it. See the full range of what we build for the shapes a project typically takes.
Are you a reseller or an agency that builds on top of an AI tool?
We are a build agency. We design and write the system around your workflow rather than configuring a template product and reselling it under our name. Where an off-the-shelf tool genuinely covers your case, we will tell you that on the call instead of building something you do not need — see how a custom AI agent differs from an agent-builder platform.
What does the first call involve?
Thirty minutes, free, no obligation. Bring the workflow that costs your team the most hours and a rough sense of the systems it touches. We ask what happens today, where it breaks down, and what a system would need to see to do it well. You leave knowing whether it is worth building at all — that honesty is the point of the call, not a sales pitch dressed as one.
Cost and commitment
How does pricing work?
We do not publish a rate card, because two requests that sound identical — "an AI agent for support," say — can need completely different builds. Instead we scope the workflow first, then give you a clear price agreed before any build work starts, in writing, so you know exactly what you are paying for. The pricing page walks through what actually drives the number.
Is there a minimum project size?
No hard minimum, but there is a practical one: the workflow needs enough volume and enough repeated shape to be worth automating. A task that happens twice a month rarely clears that bar. We will tell you honestly on the first call if your case does not, rather than take the project anyway.
What happens if we want to stop?
You can stop after the first call if the case does not hold up, or rescope once you have seen the written plan if priorities shift. If a system is already live and you no longer need it, we scope winding it down the same way we scoped building it — plainly, with a clear price, before any work starts. We do not keep billing for a system nobody asked to keep changing.
Timeline and delivery
How long do projects take?
Most systems are live in weeks, not quarters. We deliberately start with the narrowest useful version of a system rather than the full ambition, because a small system running well in production teaches us more than a large one still being tested. The process page lays out the four stages — Discover, Design, Build, Run — and roughly how long each takes.
Who owns the system and the code once it is built?
You do. What we build is yours — the code, the configuration, the logic that encodes how your business makes decisions. That is a deliberate design choice: it is the difference between a custom build and renting space on someone else's platform, where the logic behind your workflow lives on their roadmap instead of yours.
Technology and integration
Will it work with our existing software?
In almost every case, yes. Modern tools connect through their APIs; older systems that were never built to talk to anything usually still have a database, a scheduled export, or an email route we can work through. Legacy systems are the normal case, not the exception — bring the list of what you run to the first call and we will tell you what connects and how.
What happens when the AI gets something wrong?
It is built to fail safely rather than to never fail. Actions with real consequence — refunds, account changes, anything touching money — sit behind approval rules. Every decision is logged, and anything outside the system's defined scope routes to a person with the context already assembled, instead of guessing past its competence. We also test against your real historical cases before launch specifically to surface the failure modes while it is still safe to find them.
Who maintains it after launch?
We do. Launch is not the handover point where we disappear — it is where the system starts generating the signal that makes it better. We monitor what it handles, tune it as real usage reveals edges the original scope missed, and keep it working as the tools it connects to change their APIs underneath it. How your data is handled through all of that is covered plainly on our security page rather than buried in a contract appendix.