The product
Is this just ChatGPT or Claude with our documents?+
No. A general AI assistant will write a confident answer to almost any question, whether or not your company has ever said it. Rainharbor's engine drafts only from answers and documents your company has already approved, shows the source behind every answer, and writes nothing when your material doesn't support one. A person at your company approves every answer before it goes anywhere. The engine can even use the same AI models underneath. The difference is the rules around them.
What happens when the engine can't find an answer?+
It doesn't draft one. The question goes to your reviewer with the reason it wasn't answered, for example that nothing relevant was found in your approved material. Your reviewer writes the answer, and once approved it can be added to your library, so the same question is easier next time.
Does it learn from or train on our data?+
No. Rainharbor never uses your material to train an AI model, and nothing we do changes the model itself. What grows is your library of approved answers, which belongs to you. If you connect an AI model provider you already contract with, how that provider handles your data is governed by your own contract with them.
Can we bring our existing answer library, including from another tool?+
Yes. Your existing answer library comes in as a spreadsheet, Word, or comma-separated file, so if your current tool can export to a spreadsheet, your library can come with you. Questionnaires you've already completed can come in too, and the engine tells you exactly what is missing or out of date rather than quietly skipping it.
Which questionnaire formats does it handle?+
Spreadsheet and Word questionnaires come back with every approved answer filled in place and the file otherwise as it was. Portable document format (PDF) questionnaires with readable text can be taken in, and the answers come back as a separate file. Scanned images aren't supported, and questionnaires need to be in English. If a buyer uses a web portal, copy the questions into a spreadsheet, and the approved answers come back ready to enter.
Does it answer the whole request for proposal, or only parts of it?+
It's built for the security and compliance sections, which are usually the longest and most repetitive part of a bid. It can also draft any other question your approved material covers, such as company background. Anything your material doesn't cover, like pricing, delivery plans, or contract terms, goes to your team to answer, so nothing is guessed.
Data and security
Where does our data go?+
Nowhere outside environments you control. The engine runs inside your own cloud account, and by default so does the AI model it uses. Rainharbor stores none of your data on its own systems, and the engine sends none of your content back to us. If you choose to connect an AI model provider you already contract with, each question and the approved material behind it go to that provider, with your written approval.
Can anyone at Rainharbor see our material?+
Only your named Rainharbor delivery lead, during setup from installation through acceptance, working inside your environment and under your access rules. Our product engineering team never has access to your material. Nobody at Rainharbor has standing access to your installation, and updates are started from inside your environment, never pushed from outside.
Which AI model does it use?+
Your choice. By default, the engine uses an AI model offered inside your own cloud account, so nothing leaves it. If your company already contracts directly with Anthropic or OpenAI, the engine can use that instead. Either way, the same rules apply: approved sources only, a source on every answer, and a person approves everything.
Can our security team review Rainharbor?+
Because the engine runs inside your own cloud account, your security team reviews it where it lives, under your own controls. Rainharbor holds none of your data and has no standing access to your installation. The engine's logs record activity and timing, never the content of your questions, answers, or documents. We'll answer your security questionnaire about us, just as you answer your customers'.
Buying and owning
What does "you own it" mean in practice?+
The engine installed in your cloud account is yours to keep and run, with no subscription that can switch it off. Your answer library and every approved answer belong to you. Rainharbor keeps the rights to the underlying code, but the installed engine is yours to run for as long as you choose, and it doesn't depend on us to keep working.
What happens after acceptance? Who maintains it?+
Your engine keeps running as installed. For companies that want it kept current, ongoing support covers updates and security fixes, and adjustments when your AI model provider changes something. Any update is installed from inside your environment, never pushed from outside.
How is this different from tools we already use?+
Most questionnaire tools are subscriptions: your library lives on the vendor's systems, the AI runs on their terms, and when you stop paying, you lose access. Rainharbor installs the engine inside your own cloud account, with the AI model you choose, and the engine and your library are yours to keep. If your current tool can export your library, you can start from it.
How much does it cost, and how long does it take?+
It depends on your needs. The engine is a one-time purchase you own, not a subscription. We'll go through cost and timing with you after a short conversation about how your team handles questionnaires today, and set the timeline together at the start.