BuiltOnAir

Alli Alosa and Danny Fellars dive into a series of recent Airtable product updates. They discuss the new ability to perform unique counts on linked record fields and how this functionality specifically improves data visualization within interface charts.

The walkthrough covers several workflow improvements, including better handling of lookup and rollup configurations, field matching during connection changes, and smarter single-select filter behavior.

Finally, they touch on various user interface refinements, such as updates to the interface record builder, dashboard count formatting, and general UI cleanup to improve the user experience.

⏱ In this cut:

00:26 — Gemini and Claude connections

01:25 — Unique counts on linked records

04:32 — Lookup and rollup improvements

10:25 — Interface record builder updates

11:35 — Dashboard count formatting

13:47 — UI and experience improvements
This is a clip from S26-E01 of the BuiltOnAir Podcast.

Danny and Alli discuss the significant expansion of Airtable automation limits, which now range from 75 for free users up to 200 for enterprise accounts. They also touch on recent stability fixes, including auto-retries for script input errors.

The conversation covers new automation capabilities, such as the ability to delete records directly and dynamic date field comparisons. The hosts weigh in on the potential risks of automating deletions and share best practices for using these new features safely.

Finally, they explore the impact of MCP integration for creating and editing custom scripts and highlight several UI/UX improvements designed to streamline workflows, like improved field duplication and expanded text editing.

⏱ In this cut:

00:13 — Increased automation limits

04:22 — Deleting records via automation

06:35 — Date field comparisons in find records

07:56 — MCP integration for custom scripts

12:40 — Field duplication UI improvements

13:40 — Expanded text editing in AI fields
This is a clip from S26-E01 of the BuiltOnAir Podcast.

Rupert from DocsAutomator demonstrates an advanced e-signing suite designed for seamless Airtable integration. Unlike traditional manual workarounds, this tool provides a legally compliant way to manage business contracts with full audit trails and seven-year document retention.

The demonstration features the DocsAutomator AI agent, which can build complex legal templates through a simple chat interface. The agent identifies Airtable data structures, automatically maps fields, and generates necessary automations to handle the entire contract lifecycle.

Watch the end-to-end workflow in action, from initial template generation and data mapping to sending signing requests and receiving the final, signed document back into Airtable, complete with a compliance certificate containing IP addresses and hashes.

⏱ In this cut:

02:14 — Overview of the e-signing solution and compliance

03:10 — Generating contract templates with the AI agent

07:15 — Automatic Airtable data mapping and template creation

12:15 — Managing the end-to-end signing process

17:30 — Legal audit trails and document compliance

🔗 Resources mentioned in this episode:
• DocsAutomator: https://docsautomator.co?via=on2air

This is a clip from S25-E08 of the BuiltOnAir Podcast.

A deep dive into the technical workflow of DocsAutomator, demonstrating how to map Airtable fields and related line items to Google Doc placeholders. Rupert explains how the built-in AI agent can reverse the traditional workflow by generating templates directly from your existing Airtable data structure.

The discussion covers post-generation actions, such as saving PDFs back to Airtable, merging existing PDFs, and sending documents via Gmail or Outlook. It also explores the technical complexities of working with the Google Docs API and a breakdown of supported platforms including Notion, ClickUp, SmartSuite, and more.

⏱ In this cut:

00:22 — Mapping Airtable data to Google Doc placeholders

00:42 — Using AI to generate templates from data structures

01:35 — Mapping line items and related tables

06:06 — Post-generation actions and webhooks

09:56 — Challenges of the Google Docs API

13:21 — Supported platforms and formats

🔗 Resources mentioned in this episode:
• DocsAutomator: https://docsautomator.co?via=on2air

This is a clip from S25-E08 of the BuiltOnAir Podcast.

The team explores advanced workflows for managing AI agents, specifically how to use formulas and triggers to prevent an AI from overwriting manual human edits. They demonstrate using "last modified" fields and custom checkbox logic to create a "global stopper" for automations.

They also dive into the various output types available for field agents, including single and multiple select fields, number formats, and image analysis for tasks like extracting data from receipts.

Finally, they demonstrate the utility of web-searching for images, such as company logos, and share a practical example of using AI agents to research hotel amenities and reviews for travel planning.

⏱ In this cut:

01:05 — Protecting human edits from AI overwrites

02:51 — Using formulas to control AI triggers

11:44 — Exploring AI output types and image analysis

12:36 — Finding company logos via web search

16:23 — Real-world use case: AI for travel research
This is a clip from S25-E07 of the BuiltOnAir Podcast.

Alli Alosa provides a deep dive into AI field agents within Airtable, explaining how they function as mini-agents within your database tables. She discusses the evolution of field agent outputs and how users can now work in tandem with the AI within a single field.

The discussion covers practical prompt engineering, including how to reference other fields and how to enable web search tools. Crucially, Alli explains how to manage triggers to avoid wasting credits. She demonstrates why automatic generation can be problematic and shows how to use scheduling, manual commands, and custom trigger simulations to maintain full control over AI usage.

⏱ In this cut:

01:06 — What are AI field agents?

03:30 — Writing effective prompts and using web search

09:44 — Avoiding wasted credits with automatic generation

11:12 — Managing triggers: schedules and manual commands

15:38 — Simulating automation to trigger agents on command
This is a clip from S25-E07 of the BuiltOnAir Podcast.

Kamille Parks demonstrates how to use the Model Context Protocol (MCP) to manage Airtable automations via Claude. She walks through updating an existing automation by adding placeholder email addresses to a notification step, showing how the AI can modify specific parameters while preserving existing context.

The demo also covers management tasks like deleting an automation and using the "undo" feature to revert actions. By leveraging action IDs tracked through the MCP, the AI can effectively restore what was previously deleted.

Finally, the segment explores creating new scheduled automations from scratch and exporting all base automations as JSON for auditing and configuration reviews.

⏱ In this cut:

01:58 — Updating an existing automation

06:52 — Deleting and undoing actions

08:36 — Creating new scheduled automations

16:11 — Exporting automations as JSON

📝 Read this clip's post: https://builtonair.com/s25-e06-ai-powered-airtable-automations-updating-and-reverting-with-mcp

This is a clip from S25-E06 of the BuiltOnAir Podcast.

Alli Alosa demonstrates how to use the 'Insert as JSON' feature in Airtable automations to reliably handle incoming data from external webhooks, such as Fillout. She explains why this method is more robust than traditional field mapping, especially when dealing with inconsistent data.

The discussion covers a practical use case involving event feedback collection. By passing the entire payload as a JSON object, you can prevent automations from failing when certain form fields are left empty or missing from the payload.

Finally, the team explores how to use this JSON data within a script action. Instead of manually mapping every individual field as an input, you can pass the entire object into a script, allowing for much more scalable and maintainable automation logic.

⏱ In this cut:

01:03 — Using feedback collection as a use case

12:38 — The benefits of using Insert as JSON

13:37 — Solving errors caused by empty field values

17:00 — Passing JSON data into a script action

25:09 — Conditional filters and advanced integrations

📝 Read this clip's post: https://builtonair.com/s25-e06-airtable-automations-using-insert-as-json-for-webhook-data

This is a clip from S25-E06 of the BuiltOnAir Podcast.

In this well timed episode, we break down the massive news that Airtable is being acquired by the Italian company Bending Spoons. We dive into the details of the valuation, the shock to the community, and what this means for the future of the platform.

We explore Bending Spoons' track record with previous acquisitions like Evernote and StreamYard, discussing their reputation for restructuring pricing and product tiers. We also leverage the Hyperagent Analysis of Airtable Acquisition to examine how the company might operate the platform based on their previous acquisitions.

Finally, we weigh the risks of losing institutional knowledge against the potential for fresh investment and long-term stability under a publicly traded company.

🔗 Resources mentioned in this episode:

• Hyperagent Analysis of Airtable Acquisition: https://hyperagent.com/s/h_jHCWOJ6x2uTs-vQ-Wi6g

📝 Read the full episode post and show notes: https://builtonair.com/s25-e05-airtable-acquisition-by-bending-spoons

In this episode, Kamille demonstrates the power and limitations of using Claude and the Model Context Protocol (MCP) to build an Airtable base from scratch. She walks through a live experiment where she prompts Claude to create a complex meal planner, including tables for recipes, ingredients, and a shopping list, as well as the necessary automations and interfaces.

The team also dives into recent Airtable updates, including the new Slack integration and the ability to manage automations via MCP. They discuss the current limitations of AI-driven base building—such as the inability to edit existing interfaces or handle script steps—and compare the experiences of using Claude versus Airtable's Omni.

📝 Read the full episode post and show notes: https://builtonair.com/s25-e04-building-airtable-bases-with-claude-and-mcp