Terms of Service

What we actually do

We design and build custom AI agents — software connected to your own systems that takes defined actions inside rules we agree with you, rather than answering questions and stopping there. We're not a marketplace matching you with a contractor; the people who scope and build the agent are ours, working under this agreement.

Discovery comes first

Every project starts with a discovery conversation, sometimes paid separately for larger or more ambiguous tasks. Nothing is quoted before it. If, after discovery, the honest answer is that an agent isn't the right fix for your task, we'll say so rather than scope one anyway.

Scope, in writing

A statement of work names the systems the agent touches, its decision boundaries, what counts as finished, and what's explicitly excluded. Work outside that scope is a change request, agreed and priced before we start on it — not assumed to be included because it seemed related.

Money

Project fees are quoted in US dollars, fixed once scoped, and typically invoiced against milestones. Ongoing support, once an agent is live, is billed separately as a retainer. The AI model or API usage the agent generates is billed directly to you by the underlying provider — we estimate it during scoping, but we don't mark it up or bill it ourselves. Bank fees, currency conversion, and any tax properly owed in your jurisdiction sit with you.

Timeline

An indicative timeline is part of the written scope. Timelines can move if requirements change mid-build, if a system we need access to isn't made available when expected, or if testing surfaces something the original scope didn't anticipate — we'll flag any of that as soon as we see it, not at the deadline.

Access to your systems

The agent runs on credentials you create, scoped to what the task needs. You can revoke that access at any time, for any reason, and doing so pauses the agent immediately. We don't hold standing access beyond what an active engagement requires.

What we can't promise

Here's the part we won't dress up. We can design careful decision boundaries and test against the cases we can anticipate, but we can't promise an agent will never encounter a case it handles badly — that's why every agent we build includes an escalation path, not a claim of perfection. We also can't promise a specific commercial outcome (more sales, lower costs) follows automatically from automating a task; that depends on your process and market as much as the agent itself. What we can promise is a system built to the agreed scope, tested against the cases we agreed to test for, with a clear way to catch and correct what it gets wrong.

Fixing what's broken

If the agent doesn't do what the signed scope says it should, we fix it at no extra charge. That's different from a change request — a new requirement that wasn't in the original scope is new work, priced separately.

Who owns what gets built

The agent's configuration, prompts, and any code written specifically for your project belong to you once paid for in full. Three normal exceptions: open-source components keep whatever licence they already had, the underlying AI model or platform we build on stays governed by its provider's own terms, and our general methods and know-how — not specific to your project — remain ours to use on the next one.

Confidentiality

Whatever either side learns about the other during a project — your systems and credentials, our internal methods — stays confidential and is used only for that project. We'll sign your NDA, or offer a mutual one, before any detailed discussion of your systems.

Your data, and your customers' data

You decide how your own data, and any of your customers' data the agent touches, gets processed — that responsibility stays with you. Tell us during scoping which rules apply, including whether a formal data processing agreement is needed, and we'll build around it.

Liability

Neither side answers to the other for indirect loss, lost profit, or lost data, whatever the cause. The most either party owes the other in connection with a project is capped at the fees paid under that project's statement of work. This doesn't limit liability that can't legally be limited.

Ending a project

Either side can end a project on written notice for a serious breach that isn't fixed within a reasonable time after being raised, or if the other party becomes insolvent. On the way out: fees already earned stay payable, our access to your systems is revoked, and confidentiality obligations keep running.

Using this website

Everything on aiagents.firm.in is informational. Nothing here is an offer you can simply accept, and no contract exists until a statement of work is signed. Please don't put credentials, API keys or anything sensitive into a form on this site.

Standard clauses

A signed statement of work overrides this page wherever the two disagree. Neither side can assign this agreement without the other's written consent, except to a genuine successor of the whole business. If one clause here turns out unenforceable, the rest still stands. Neither side is liable for failing to perform because of something genuinely outside their control.

Governing law

Which law applies and where disputes get heard is set out in your signed statement of work, agreed rather than assumed, since clients are spread across enough jurisdictions that no single default suits everyone. Absent an explicit choice, the law where AI Agents Firm operates applies. Either way, both sides agree to raise a disagreement directly and try to resolve it in good faith before filing anything.

What this page is

A plain description of how projects actually work, not legal advice, and not a substitute for a lawyer who knows your own jurisdiction. Wherever it conflicts with something you've signed, what you signed wins.

Contact

Questions before you sign anything: info@aiagents.firm.in, or call +91 63530 74980, Monday to Friday, 10:00–19:00 IST.