Buildorado
Back to Blog
aifeature-launch

Introducing AI Nodes

Use AI to process form submissions, generate content, and make intelligent decisions.

Buildorado Team·February 15, 2026·6 min read

What if your workflow could read a form submission and instantly decide whether the lead is worth a sales call? Or draft a personalized follow-up email in the prospect's tone? Or route a support ticket to the right department based on the content of the message, not a dropdown selection? That is exactly what AI Nodes bring to Buildorado -- and they are available today for every plan.

AI Nodes are a new category of workflow nodes that let you plug artificial intelligence directly into your automation flows. No separate AI tools, no custom API integrations, no prompt engineering frameworks. You add an AI node to your canvas, configure it with a prompt, connect it to upstream data with variables, and let it run. The output feeds into downstream nodes just like any other data in your workflow.

Let's walk through what is available, how it works, and three concrete use cases you can build right now.

The 9 AI Node Types

Buildorado ships with nine AI node types, each designed for a specific class of task. You can find the full technical reference in the AI Nodes documentation, but here is a practical overview of what each one does and when to use it.

1. AI Agent

The most powerful AI node. An AI Agent is a conversational model with tool calling, memory, and multi-turn reasoning. It can autonomously call tools, search data, and make decisions across multiple steps.

When to use it: Complex tasks that require reasoning and multiple steps -- like analyzing a submission, looking up related data, and drafting a response that references all of it.

Providers: OpenAI (GPT-4.1, o3, o4-mini), Anthropic (Claude Opus 4.6, Sonnet 4.6), Google (Gemini)

2. Text Generation

The workhorse node for most AI workflows. Text Generation takes a prompt and produces text -- summaries, classifications, translations, extracted data, drafted emails, anything you can describe in a prompt.

When to use it: Single-turn text tasks -- classify a lead, summarize a form response, extract structured data from free text, generate an email draft.

Providers: OpenAI, Anthropic, Google, Groq

3. Text-to-Speech (TTS)

Converts text to natural-sounding speech audio. The output is an MP3 file URL that downstream nodes can use -- embed it in an email, store it in Google Drive, or pass it to a webhook.

When to use it: Accessibility features, audio confirmations, podcast-style summaries of form submissions.

Providers: OpenAI, ElevenLabs (v3)

4. Speech-to-Text (STT)

Transcribes audio files to text. Point it at an audio URL or an uploaded file, and it returns a transcript you can process with other nodes.

When to use it: Transcribing voice notes or audio recordings submitted through file upload fields.

Providers: OpenAI (Whisper), Deepgram (nova-3)

5. Image Generation

Creates images from text prompts. The output is a PNG image URL.

When to use it: Generating custom visuals based on form data -- personalized certificates, product mockups, social media graphics.

Providers: OpenAI (DALL-E 3), Stability AI (SD3.5)

6. Vision

Analyzes images -- extracts text, identifies objects, answers questions about visual content. Feed it an image URL and a question, and it returns a text analysis.

When to use it: Processing uploaded images -- reading receipts, analyzing product photos, verifying document uploads.

Providers: OpenAI (GPT-4o, GPT-4.1), Anthropic (Claude Vision)

7. OCR

Extracts structured text from images and documents. Similar to Vision but optimized specifically for text extraction.

When to use it: Reading scanned documents, extracting data from uploaded business cards, processing handwritten forms.

8. Image Edit

Modifies existing images using AI -- inpainting, outpainting, style transfer, and background removal. Feed it an image URL and an editing instruction, and it returns a modified image.

When to use it: Post-processing uploaded images -- removing backgrounds from product photos, adding watermarks, applying style transformations to user-submitted images.

Providers: OpenAI (DALL-E), Stability AI

9. Embeddings

Generates vector embeddings for semantic search and similarity matching. This is a building block for more advanced AI applications.

When to use it: Finding similar submissions, building recommendation systems, semantic search across your data.

How AI Nodes Work in Practice

Every AI node follows the same configuration pattern, which keeps things consistent regardless of which type you are using:

  1. Choose a provider -- Select the AI provider (OpenAI, Anthropic, Google, etc.)
  2. Pick a model -- Choose the specific model (GPT-4.1, Claude Sonnet 4.6, etc.)
  3. Select a credential -- Pick the API key you have configured in Settings > Credentials
  4. Write your prompt -- Use {{variables}} to inject data from upstream nodes
  5. Set parameters -- Temperature, max tokens, and other model-specific settings

The AI node processes the input and produces output variables that downstream nodes can reference:

{{aiNode.text}}          // Text generation output
{{aiNode.audioUrl}}      // TTS audio URL
{{aiNode.transcript}}    // STT transcript
{{aiNode.imageUrl}}      // Image generation URL

This means you can chain AI nodes with branch nodes, action nodes, and other AI nodes to build sophisticated pipelines. The output of one AI node becomes the input of the next.

Use Case 1: Automated Lead Scoring with AI

Traditional lead scoring uses rigid rules: if budget is above a threshold and company size is above another threshold, assign a score. This works for structured fields, but it completely ignores the richest signal in your form -- the free-text fields.

Here is how to build AI-powered lead scoring in Buildorado:

The workflow:

Contact Form → AI Text Generation (Score Lead) → Branch (Route by Score)
  → Hot Lead: Email + Slack + CRM
  → Warm Lead: Email + Google Sheet
  → Cold Lead: Email

Step 1: Create a form with standard fields (name, email, company size, budget) plus a textarea: "Describe your project and timeline."

Step 2: Add a Text Generation node after the form. Configure the prompt:

You are a lead qualification expert. Based on the following form submission,
assign a lead score from 0-100 and a category (hot, warm, or cold).

Name: {{contactForm.fullName}}
Email: {{contactForm.email}}
Company Size: {{contactForm.companySize}}
Budget: {{contactForm.budget}}
Project Description: {{contactForm.projectDescription}}

Respond in JSON format:
{"score": <number>, "category": "<hot|warm|cold>", "reason": "<brief explanation>"}

Step 3: Add a Branch node that reads the AI output and routes based on the category. The AI evaluates signals that rules cannot capture -- urgency in language, specificity of requirements, mentions of competitors, timeline pressure.

The result: Leads that write "We need this deployed by next quarter and have executive buy-in" get flagged as hot, even if their stated budget is modest. A rule-based system would miss that entirely.

Use Case 2: Auto-Generated Follow-Up Emails

Personalized follow-up emails convert dramatically better than generic ones, but writing a unique email for every submission does not scale. AI Text Generation solves this.

The workflow:

Application Form → AI Text Generation (Draft Email) → Send Email Action

Step 1: Build your application or inquiry form with whatever fields are relevant.

Step 2: Add a Text Generation node with this prompt:

Write a professional, warm follow-up email to {{applicationForm.firstName}}.
They applied for: {{applicationForm.position}}
Their background: {{applicationForm.experience}}
Their message: {{applicationForm.coverLetter}}

The email should:
- Thank them specifically for something they mentioned
- Reference their relevant experience
- Explain the next steps in our process
- Be 3-4 paragraphs, conversational but professional
- Sign off as "The [Your Company] Team"

Step 3: Connect the AI output to a Send Email action node. Use {{draftEmail.text}} as the email body.

The result: Every applicant receives an email that feels personally written, references details from their submission, and arrives within seconds of submitting the form. Your team never has to write a routine acknowledgment email again.

Pro tip: Set the temperature parameter to 0.7 for emails -- high enough to produce varied, natural language, but low enough to stay on-topic and professional.

Use Case 3: Intelligent Form Routing

Support tickets, feedback forms, and internal requests often include a "Category" dropdown that users either ignore, misuse, or spend too long deliberating over. AI can classify the submission automatically based on the content.

The workflow:

Support Form → AI Text Generation (Classify) → Branch (Route by Department)
  → Billing: Slack #billing + Jira ticket
  → Technical: Slack #engineering + Jira ticket (high priority)
  → General: Email to support team

Step 1: Create a support form with just a subject line and a description field. No category dropdown needed -- the AI handles classification.

Step 2: Add a Text Generation node:

Classify this support request into exactly one category:
billing, technical, account, feature_request, or general.

Subject: {{supportForm.subject}}
Description: {{supportForm.description}}

Also assess the urgency (low, medium, high) based on the language used.

Respond in JSON: {"category": "<category>", "urgency": "<urgency>"}

Step 3: Add a Branch node with conditions for each category. Route to the appropriate team channels and ticketing systems.

The result: Support requests are instantly routed to the right team with the right priority level. Users do not have to guess which category their issue falls into, and your team does not waste time re-routing misclassified tickets.

Bring Your Own Key (BYOK)

All AI nodes in Buildorado use a bring-your-own-key model. You add your API keys from OpenAI, Anthropic, Google, or other providers in Settings > Credentials, and select them when configuring AI nodes. This gives you several advantages:

Full cost control. You pay the AI providers directly at their published rates. Buildorado does not mark up API costs or add per-token fees. You can monitor usage and set spending limits directly in your provider dashboard.

Choose your models. Different tasks benefit from different models. Use a fast, cheap model (GPT-4o-mini, Claude Sonnet) for simple classification, and a powerful model (GPT-4.1, Claude Opus) for complex reasoning. You can mix providers within the same workflow.

Data privacy. Your API keys are encrypted with AWS KMS and never leave Buildorado's infrastructure. Form data is sent directly from Buildorado's servers to the AI provider -- it never passes through third-party intermediaries.

No vendor lock-in. If you want to switch from OpenAI to Anthropic, just update the provider and model in the node configuration. The rest of your workflow stays the same.

To set up your first credential, go to Settings > Credentials, click Add Credential, choose your provider, and paste your API key. It is available across all your workflows immediately.

Getting Started with AI Nodes

Adding AI to your workflows takes just a few minutes:

  1. Get an API key from your preferred provider (OpenAI, Anthropic, or Google AI)
  2. Add the credential in Buildorado under Settings > Credentials
  3. Open your workflow in the visual builder
  4. Drag an AI node from the sidebar onto the canvas
  5. Configure the prompt using {{variables}} to inject form data
  6. Connect the output to downstream action or branch nodes
  7. Preview and test with sample data before publishing

If you have never used AI in a workflow before, start with the Text Generation node -- it is the most versatile and easiest to reason about. Once you are comfortable with prompts and variable injection, explore the Agent node for more complex multi-step tasks.

Tips for Writing Effective Prompts

The quality of your AI node's output depends heavily on the prompt. Here are practices that consistently produce better results:

Be specific about the output format. If you need JSON, say so explicitly and provide the exact schema. If you need a classification from a fixed set of categories, list every valid category.

Include examples. Showing the model one or two examples of ideal output dramatically improves consistency.

Use the right temperature. Low temperature (0.1-0.3) for factual extraction and classification. Medium temperature (0.5-0.7) for emails and creative content. High temperature (0.8-1.0) for brainstorming and varied output.

Inject context with variables. The more relevant data you pass into the prompt via {{variables}}, the better the output. Do not make the model guess -- give it the data.

Test with real data. Prompts that work great with your test input might fail with edge cases from real users. Preview your workflow with varied inputs before publishing.

What's Next

AI Nodes are live for all Buildorado plans today. Here are resources to help you get started:

  • AI Nodes Documentation -- Full reference for all 9 node types, providers, configuration, and output variables
  • Workflows -- Understand how nodes, edges, and variables work together
  • Branch Nodes -- Route workflows based on AI output with conditional logic
  • Action Nodes -- Connect AI output to emails, Slack, CRMs, and 30 integrations
  • Conditional Logic -- Build sophisticated routing rules with 35+ operators
  • Templates -- Browse pre-built workflows that use AI nodes

If you are new to Buildorado, start with our Getting Started guide to build your first workflow in five minutes. For tips on structuring complex automations, read our Workflow Automation Best Practices. And if you want to combine AI with multi-step forms, our guide on multi-step forms with conditional logic shows how to build adaptive form flows that feed directly into AI nodes.

For a comparison of form builders with AI features, check out our Typeform alternatives roundup and Buildorado vs. Tally vs. Fillout.

We are excited to see what you build with AI Nodes. If you create something interesting, share it with us on Twitter or in the community Discord -- we love featuring creative use cases from the community.

Introducing AI Nodes | Buildorado Blog | Buildorado