How to Integrate Retool with DeepAI

DeepAI provides a free tier that includes access to all API endpoints with rate limiting — no credit card is required for initial access. To get your API key, navigate to deepai.org and create a free account if you do not already have one. After signing in, click on your profile icon or navigate to the Dashboard section. Your API key is displayed on the Dashboard or under Account Settings. DeepAI's API keys are long alphanumeric strings. Copy the key value. Unlike some API platforms, DeepAI API keys do not expire and do not require periodic rotation. The API key is passed as a request header named api-key (lowercase, hyphenated) on every request — not in an Authorization: Bearer header. This is an important distinction when configuring the Retool resource, as you will add it as a custom header rather than selecting the Bearer Token authentication type. Immediately store your DeepAI API key in a Retool Configuration Variable: in Retool, navigate to Settings → Configuration Variables, click Add Variable, name it DEEPAI_API_KEY, paste the key value, and check the 'Is Secret' checkbox to prevent it from appearing in query logs or being accessible on the frontend. The free tier allows up to 5,000 requests per month with standard latency. For higher volumes, DeepAI offers paid plans at monthly flat rates. Note the specific API limitations for your tier — image generation endpoints typically consume more credits per call than text analysis endpoints.

In Retool, navigate to the Resources tab and click Add Resource. Select REST API from the resource type list. Name the resource 'DeepAI API'. Set the Base URL to https://api.deepai.org. For authentication, do NOT select 'Bearer Token' — DeepAI uses a custom header name. Instead, scroll to the Headers section and add a custom header: Name → api-key, Value → {{ retoolContext.configVars.DEEPAI_API_KEY }}. Add a second header: Content-Type → application/json. This Content-Type header applies to JSON body requests; for image upload requests that use multipart form data, you will need to override or remove this header in individual queries. Click Save Resource. To test the resource, create a query in a new Retool app: set Method to POST and Path to /1.0/text-to-image. In the Body section, select Form Data type and add a field: text → a test prompt like 'sunset over mountains'. Click Preview. A successful response returns a JSON object with an output_url field containing the URL of the generated image. If you receive a 403 error, the api-key header is missing or the key value is incorrect. If you receive a 200 response with the output_url, your resource is correctly configured and ready for building queries.

Create queries for DeepAI's text processing endpoints. Create a query named 'summarizeText'. Set Method to POST and Path to /1.0/summarization. For the Body, select Form Data and add a field: text → {{ textInputArea.value }}. The DeepAI summarization API accepts plain text and returns a JSON object with an output field containing the summarized text. Create a transformer that extracts the output field and trims whitespace. Create a second query named 'analyzeSentiment'. Set Method to POST and Path to /1.0/sentiment-analysis. Body: Form Data with field text → {{ textInputArea.value }}. DeepAI's sentiment analysis returns a JSON object with an output field containing a string like 'Positive', 'Negative', or 'Neutral', and sometimes additional detail about detected sentiment categories. Create a transformer that parses the output field and adds a color property based on the sentiment value: 'green' for Positive, 'red' for Negative, 'gray' for Neutral. These color values can be used in Retool's Tag component to show color-coded sentiment labels. Note that both the summarization and sentiment analysis endpoints accept text input via form data, not JSON — ensure your query body type is set to Form Data (multipart/form-data), not JSON. Using JSON body format with DeepAI's text endpoints is a common configuration mistake.

1// Transformer: parse and enrich DeepAI sentiment response
2const output = data.output || '';
3const sentimentText = output.trim();
4
5const SENTIMENT_COLORS = {
6  'Positive': 'green',
7  'Negative': 'red',
8  'Neutral': 'gray',
9  'Mixed': 'yellow'
10};
11
12// Find the primary sentiment label
13const detectedSentiment = Object.keys(SENTIMENT_COLORS).find(s =>
14  sentimentText.toLowerCase().includes(s.toLowerCase())
15) || 'Unknown';
16
17return {
18  raw: sentimentText,
19  sentiment: detectedSentiment,
20  color: SENTIMENT_COLORS[detectedSentiment] || 'gray',
21  is_positive: detectedSentiment === 'Positive',
22  is_negative: detectedSentiment === 'Negative'
23};

Create queries for DeepAI's image-based endpoints. Create a query named 'generateImage'. Set Method to POST and Path to /1.0/text2img (DeepAI's standard image generation endpoint — verify the current endpoint path in DeepAI's documentation as it may be /1.0/text-to-image or /1.0/stable-diffusion depending on which model you want). Body: Form Data with field text → {{ imagePromptInput.value }}. The response includes an output_url field containing a direct URL to the generated image hosted on DeepAI's CDN. Create a query named 'upscaleImage'. Set Method to POST and Path to /1.0/waifu2x. Body: Form Data with field image_url → {{ imageUrlInput.value }} (the URL of an image to upscale) or upload a file using the image field type in Form Data for direct upload. The upscaled image URL is also returned in output_url. For image colorization, create a query named 'colorizeImage' with Path /1.0/colorizer and the same image_url parameter. Create a JavaScript query named 'saveGenerationToHistory'. This runs after generateImage succeeds and inserts a record into Retool Database with the prompt text, generated image URL, endpoint used, and current timestamp — building a history log of all AI generations. Configure the generateImage query's On Success event handler to trigger saveGenerationToHistory automatically.

1-- SQL: Insert generation into history table in Retool Database
2INSERT INTO deepai_generations (
3  prompt,
4  output_url,
5  endpoint,
6  generated_at
7) VALUES (
8  {{ imagePromptInput.value }},
9  {{ generateImage.data.output_url }},
10  'text2img',
11  NOW()
12);

Build the complete multi-function AI tool panel in Retool using a Tab component to organize different AI capabilities. Create a Tab component with three or more tabs: 'Text Analysis', 'Image Generator', and 'Image Tools'. In the Text Analysis tab: add a Text Area component named 'textInputArea' with placeholder text and a max length note, a row of two buttons ('Summarize' and 'Analyze Sentiment'), and two result containers below. The Summarize result container shows the summarizeText.data output in a formatted Text component that becomes visible after the query runs. The Sentiment result container shows a Tag component with color bound to {{ analyzeSentiment.data.color }} and label bound to {{ analyzeSentiment.data.sentiment }}, plus the raw output below. Wire the Summarize button's onClick event to trigger summarizeText, and the Analyze Sentiment button to trigger analyzeSentiment. In the Image Generator tab: add a Text Input named 'imagePromptInput' with placeholder 'Describe the image you want to generate', a Generate button that triggers generateImage, an Image component with src bound to {{ generateImage.data.output_url }}, and a Table below showing generation history from Retool Database. Add a copy-to-clipboard button next to the image using Retool's copy icon action. In the Image Tools tab: add a Text Input for image URL input, two buttons ('Upscale' and 'Colorize'), and a before/after image comparison showing the original URL and the processed result URL. Add loading states to all buttons using Retool's loading property bound to the corresponding query's isFetching state.