Legal

Privacy Policy

Last updated: 2026-08-17 · Effective: on publication

This policy explains what Overbuilt collects, why, where it is kept, and what we will and will not do with it. It is written to be read, not to be survived.

Who we are

Overbuilt is operated by Overbuilt Technologies LLC, a limited liability company registered in the United States.

For any question about this policy, or to ask what we hold about you, contact James McConnell · operations@overbuilt.ai · 678-276-6587.

Written correspondence can be sent to our registered address:

Overbuilt Technologies LLC
2810 N Church St PMB 910677
Wilmington, Delaware 19802-4447
United States

What we collect

When you book a walkthrough

The booking form collects your name, business name, email address, phone number, industry, and team size, along with the meeting time you select and your timezone. We use this to prepare for and hold that call.

When you become a client

To configure and run agents we are granted access to systems you choose to connect. Depending on the tier and the agents you buy, that can include email, calendar, files, a CRM, and accounting or card-payment data. We only request access to what the agents you have bought actually need.

Automatically

No cookies, no third-party tracking. This site sets no cookies, runs no advertising pixels, no tag managers, no behavioral tracking and no third-party analytics service. Nothing is stored on your device and nothing about your visit is shared with another company.

We do count page views, on our own server. When you open a page, it sends us two things: the path of the page, and the host name of the site that referred you if there was one. That is the whole payload. We record it as a daily tally.

What that specifically does not include:

• No cookie, no localStorage, no identifier of any kind written to your device.
• No IP address stored, and none hashed or derived into an identifier.
• No device or browser fingerprint.
• No per-visitor record. We keep counts per day, not rows per person, so there is no profile to build, sell or hand over.
• No full referring URL — only the bare host, because a full URL can carry a search query and a host name cannot.
• No cross-site tracking. The counter runs on our own infrastructure and knows nothing about where you go next.

We honor the Do Not Track setting in your browser. If it is on, we count nothing at all.

These tallies are kept for up to 400 days and then deleted. We use them for one purpose: seeing which pages people actually read so we can improve them.

Separately, our hosting providers keep standard technical logs — IP addresses, timestamps, error records — for security and diagnostics, as any web host does. We do not use those logs for analytics, profiling or marketing, and we do not combine them with anything you submit.

What we do with it

• Configure, run and support the agents you have bought.
• Produce the usage and cost reporting included in your tier.
• Contact you about your engagement.
• Develop, train, evaluate and improve our models and agents.

We do not sell personal information, and we do not disclose your identity or your confidential information. We do not reproduce your data in work delivered to another client.

Model providers and training

Running an agent means sending relevant content to a model provider. Which content depends entirely on the job the agent has been given. We use OpenAI and Anthropic, both through their commercial APIs. The full list of third parties involved in delivering the product is published at overbuilt.ai/subprocessors.

Do you train on our data?

Yes, and we would rather say so plainly than bury it. We use data processed through the service to improve the agents we build — their configuration, the prompt and evaluation libraries behind them, and the correction patterns that stop a mistake recurring. That is how an agent gets better at the specific work a business like yours actually does.

To be precise about what that is not: we do not train foundation models. The underlying models are OpenAI's and Anthropic's, reached through their commercial APIs, and we do not operate model-training infrastructure of our own. What improves is the layer we build and own on top of them.

What that does not mean:

• We do not sell your data, and we do not disclose your identity or confidential information.
• We do not reproduce your data in work delivered to another client.
• Confidentiality is contractual, not a promise on a web page — we sign a non-disclosure agreement with every client, and the client agreement sets out the data licence and its limits in writing before anything is connected.

Where data is stored

Client operational data lives in three places, and it is worth naming all three rather than implying there is only one:

• On a dedicated device housed with you. The primary working copy sits on your own premises, on hardware you hold.
• On Overbuilt's hosted infrastructure. Agent configuration, run history and the records agents build on are held on our own hosted services, so work runs on schedule and survives a device failure. Those providers are named on the Subprocessors page.
• On an external backup drive kept with your device.

The original statement of this — Credentials and integration tokens are held locally on that device rather than pooled in shared cloud infrastructure.

Overbuilt's own public surfaces — this site, the booking service and the internal dashboard — run on established hosting providers under their standard security controls.

How long we keep it

• Booking-form submissions from people who never become clients: one year.
• Client operational data during an engagement: one year, and for the duration of the contract where the agent's work builds on its own history.
• After cancellation: up to one year.

We reserve the right to retain data necessary for legal, tax and accounting obligations beyond those windows.

Your choices

• You can revoke any integration's access at any time, from the system that granted it.
• You can ask what we hold about you, ask for a copy, or ask us to delete it.
• Revoking access stops the agents that depend on it. That is the intended behavior, not a fault.

Send requests to James McConnell · operations@overbuilt.ai.

We respond to data requests as promptly as we reasonably can.

Export and deletion

On cancellation you receive an export of your data. We reserve the right to retain what is necessary for legal, tax and accounting purposes.

Changes

If this policy changes materially we will update the date above and notify active clients directly rather than relying on you to re-read the page.