How to add an AI chatbot to your app.
A plain-language walkthrough.
The short version of how to add an AI chatbot to your app with Newly: you write what the chatbot should do in plain language, and Newly builds it into your app. No LLM API to wire up, no API keys to manage by default. The rest of this page walks through the steps, what the feature can do, and a few things to know first.
The three steps
Three steps from idea to a working chatbot.
This is how to add an AI chatbot to your app with Newly, start to finish. Three steps, and you only do the talking on two of them. The middle one is Newly’s job.
- 1
Describe the chatbot in plain language
Brief Newly the way you would brief a teammate. Who does the chatbot talk to, what should it know, and how should it sound: clipped and friendly, or thorough and formal? No code, no config form. A sentence gets you moving: "add an in-app assistant that answers questions about my recipes and suggests substitutions." The more specific you are about how it should behave, the closer the first version lands.
- 2
Let Newly build it
Newly generates the chat UI, the request handling, and the backend wiring as part of your project, and the Liquid Backend runs the AI behind it. You do not pick a provider, stand up a server, or paste a secret. This is the part that would normally take the most setup by hand.
- 3
Preview, refine, and ship the real app
Open the preview and talk to your chatbot. Want it stricter, better-informed, or in a different tone? Say so, and it changes. When it behaves, you publish to the App Store and Google Play straight from the same project. The developer fees at that point go to the platforms, not to Newly.
What this flow asks of you.
You do not need to learn an SDK, set up a proxy, or pick a streaming format. The flow asks for one thing: describing how the chatbot should behave, the same way you would brief a coworker. Newly handles the wiring underneath.
AI chat, at a glance.
API keys required by default – the Liquid Backend runs the AI for you
Plain-language prompt to describe the chatbot you want
Bring your own model key in Database → Secrets, or use the default
React Native and Expo codebase as the output, yours to edit
What it is
What goes into an AI chatbot, and what Newly handles.
A chatbot looks trivial from the outside: a text box, a list of messages, an answer that appears. The reason how to add an AI chatbot to your app is a real question is everything behind that text box. By hand you choose a model provider, build a server route so your API key never ships inside the app, format the request and the response, handle timeouts and rate limits without breaking the chat, and build the message UI with its loading and retry states. Then you maintain all of it as the provider keeps changing its API.
Newly is an AI mobile-app builder, and it takes that layer off your plate. You write what the chatbot should do, and the bundled Liquid Backend wires up the AI integration for you, with no API keys required by default. What’s left is the only part a user ever notices: what the chatbot actually says.
Newly vs hand-coding
Describe it with Newly vs build it by hand.
Both paths end with a chatbot in your app. The difference is the work to get there. With Newly you describe the feature; by hand you build and maintain each piece below yourself.
With Newly
- Plain-language prompt instead of LLM SDK docs
- Liquid Backend handles the AI integration
- No API keys required by default
- Chat UI generated as part of your app
- A real React Native and Expo codebase as the output
Hand-coding it yourself
- Pick an LLM provider and read its SDK docs
- Build a backend proxy to keep your key off the device
- Wire request and response handling and error states
- Build the chat UI, loading states, and retries yourself
- Maintain it all as the provider's API changes
Hand-coding is a reasonable choice if you want to write every line yourself from the first request. With Newly, you describe the feature and get to a working chatbot with fewer steps.
Where Newly fits
Pricing, and what you get.
Newly is paid from day one, with plans starting at $25 a month and no free tier. Cloud builds, AI generations, and managed previews cost money to run, which is why there is no free plan.
The codebase Newly generates is a genuine React Native and Expo project that you own and can edit, export, and ship. By default the Liquid Backend runs the AI for you with no keys to manage. If you prefer your own provider or a specific model, you can add a key in Database → Secrets and the feature uses it instead. Most people never open that screen.
Beyond a chatbot
Other AI features you can add.
Once you know how to add an AI chatbot to your app, you already know how to add most of the AI features below. They use the same flow: you describe the behavior, Newly builds it. A chat box is one of the more familiar forms it takes.
Conversational advisor
A chatbot pinned to your app's subject: a cooking assistant, a study buddy, a support bot for your own product. The user types, it answers, and the prompt you wrote decides how much it knows and how far it strays.
Personalized advice
AI that responds to the specific person in front of it. Someone describes their goal or situation, and the feature answers for them rather than reciting a stock paragraph. This is the shape behind most coaching and recommendation features.
Image analysis
Hand the AI a photo and let it read the image. The classic version: a user snaps their lunch and gets a calorie estimate back. Receipts, plants, skin, handwriting, the input is a picture instead of a sentence.
Journaling with AI
A journal that answers. The AI reads each entry and reflects back, summarizing or nudging or asking the next question. A plain notes screen turns into something that responds when you write in it.
The honest notes
A few things to know first.
If you are weighing how to add an AI chatbot to your app, here are four practical points worth knowing before you start.
Newly is not free
Plans start at $25 a month, and there is no free tier or trial. If that does not fit your budget, it is worth knowing up front.
API keys are optional, not the default
You can ignore keys entirely and let the Liquid Backend run the AI. The bring-your-own-key path in Database → Secrets is there for teams locked to a provider or a particular model, and almost nobody else needs it.
It is a real app, not a widget
The chatbot lands in a genuine React Native and Expo codebase, not a webview shell or a sandboxed demo. You can keep editing it and publish to the app stores when it is ready.
Store fees are separate
Publishing costs $25 once for Google Play and $99 a year for Apple. That money goes to the platforms, not to Newly, and no builder can waive it for you.
How to pick.
It comes down to one question: do you want to write the wiring yourself, or describe the behavior and let Newly build it? If writing every line is what you want, code it by hand. If you mainly want the chatbot in a working app, Newly is the faster path.
FAQ
How to add an AI chatbot to your app, in plain language.
You describe it. With Newly you write what the chatbot should do in plain language – who it talks to, what it knows, how it should respond – and Newly builds the feature into your app. No LLM API to wire up, no backend proxy, and no API keys required by default; the bundled Liquid Backend runs the AI behind the scenes, so your prompt becomes a working conversational feature inside a real React Native and Expo app. Hand-coding the same thing means picking a provider, building a server route to keep your key off the device, parsing responses, and handling errors yourself. Newly handles those steps for you. Plans start at $25 a month.
Related
Describe the chatbot, and Newly builds it.
From $25 a month: you describe the chatbot, Newly builds it into a React Native and Expo codebase, and no API keys are required by default. The whole walkthrough comes down to those steps.