FAQs

AI Autocomplete is the next evolution of autocomplete, that unlocks 10X faster, more powerful products by helping users discover and complete actions in real-time.

Old autocomplete predicts the next keyword, based on historical data.

AI Autocomplete predicts the action the user is trying to take, in real time.

AI Autocomplete is a tightly integrated action engine: a proprietary backend system that understands context, intent, and product logic, then uses that to generate ultra-fast, accurate predictions in real time as users type. AI Autocomplete is designed to return suggestions up to 10X faster, and 5X cheaper than a standard LLM call. With continued improvements over time, our aim is for the vast majority of predictions to return in roughly 200–300 milliseconds.

Yes. AI Autocomplete is designed to work with any natural language input field where users type to complete an action, search for something, create content, or give instructions.

AI Autocomplete is designed to return suggestions up to 5X faster than a standard LLM call. With continued improvements over time, our aim is for the vast majority of predictions to return in roughly 200–300 milliseconds.

AI Autocomplete is built for any product with a natural language text box where users type to get something done. This includes search, commerce, media generation, software generation, customer service, productivity tools, internal enterprise workflows, and more.

Yes. You can use the API directly and build your own custom UI on top of it. We provide the SDK to make it easier to get set up quickly, with a ready-made UI that is fully customizable and easy to test instantly.

Yes. The SDK is built to be 100% flexible: you can use the included UI, customize it, replace it entirely, or skip it and build your own experience directly on top of the API. The SDK will also become open source in the near future for even greater flexibility.

The SDK is designed to be extremely easy to integrate and can often be added in minutes. More advanced enterprise integrations may take longer depending on the level of customization required.

AI Autocomplete supports nearly all major platforms. If you do not see your platform listed on the website, reach out and we can help provide the right integration path.

Yes. AI Autocomplete is designed to work across all types of products, including deep enterprise workflows and internal tools.

Yes. For large enough products on the Enterprise plan, we support custom integrations such as product catalogs, custom hosting setups, private deployments, and other enterprise-specific requirements. Reach out to us if you need a custom setup.

Common use cases include search, product discovery, customer support, content creation, workflow creation, document editing, software generation, and internal tools. Any product where users type natural language to get something done can benefit.

Yes. AI Autocomplete can work anywhere users type to create, search, edit, automate, or take action.

Yes. AI Autocomplete can help users write more complete and useful inputs across emails, messages, notes, forms, workflows, and other text-based actions.

Yes. You can control what types of suggestions appear through the suggestion logic in the platform.

Yes. You can define rules, guardrails, and specific behaviors for how AI Autocomplete should make suggestions in your product.

Yes. AI Autocomplete can use app-specific context, metadata, and product logic to make suggestions more relevant to your users.

100%. AI Autocomplete is designed strictly with privacy and security in mind. It does not store the specific string that the client chooses to pass to AI Autocomplete, and for large enough products on the Enterprise plan, we can also support sensitive field masking, private cloud deployment, and custom privacy requirements; reach out to us if your product has specific needs.

Yes. AI Autocomplete is designed to support enterprise security requirements, including privacy controls, custom deployment options, and security review processes. For large enough products on the Enterprise plan, reach out to us and we can work through your requirements.

For large enough products on the Enterprise plan, we can support custom deployment models, including private cloud or customer-specific environments. Reach out to us if you need this type of setup.

AI Autocomplete is priced per prediction, at the same price as old autocomplete ($0.0003 / prediction). This pricing is intentionally cost effective, so there is no concern to upgrade to AI Autocomplete.

You can adjust the debounce logic to determine how frequently the system calls a new prediction. However, please note: given longer debounce logic means a slower experience, we recommend that you prioritize the best possible product experience, which can mean ~10 predictions / query.

Yes. For large enough products on the Enterprise plan, AI Autocomplete can support private cloud deployment options. Reach out to us if your team needs a private cloud setup.

Yes. For large enough products on the Enterprise plan, we support enterprise deployment models for customers that need custom hosting, security, privacy, or infrastructure requirements. Reach out to us to discuss your requirements.

Yes. We recommend testing AI Autocomplete in production with a small percentage of users first so you can get real usage data quickly before expanding the rollout.

You can track any metric on your side by configuring SDK or custom UI logging. We provide metrics such as session count, latency, predictions per session, engagement rate, and word lift.

Word lift measures whether users provide more complete inputs after AI Autocomplete is added. It helps show how AI Autocomplete improves input quality and user engagement.

Yes. AI Autocomplete can be tested with a subset of users so you can compare engagement, input quality, conversion, or other product-specific metrics.

Yes. AI Autocomplete is designed to support enterprise security and compliance requirements, including SOC 2 review processes.

Yes. For large enough products on the Enterprise plan, we can support security reviews as part of the evaluation process.

Yes. For large enough products on the Enterprise plan, legal and security teams can review the architecture, deployment model, privacy controls, and security details as part of their evaluation.