Airtable spent a decade as the friendly spreadsheet-database hybrid that marketing teams used instead of begging engineering for a tool. On June 24, 2025, the company described itself as "refounded" around AI, launching Omni, an agentic assistant that builds working apps from a conversation. CEO Howie Liu framed the bet in one line: "When the cost of making apps goes to zero, how many more apps will we see made?"
A year on, the question worth answering is narrower and more useful: does Omni build apps you would actually put in front of a team, and what does the tier you need really cost? Both answers are more mixed than the marketing suggests.
What Omni actually does
Omni is not a chatbot bolted onto a database. According to Airtable's app-building page, it generates the three layers of an Airtable app together — tables, interfaces, and automations — and lets you see the data and logic it produced rather than hiding them behind a generated black box. That matters. Plenty of AI app builders emit code you cannot inspect or hand off. Omni's output is ordinary Airtable, which any existing Airtable user can open and edit by hand.
Airtable also says Omni switches between building apps and working with data: it will run analysis across large record sets, extract structure from unstructured content, and answer questions about what it finds. Permissions are respected, so the agent should not surface records a given user could not already see.
Omni followed Cobuilder, Airtable's earlier natural-language app generator. The lineage is worth noting because it explains the product's shape — Airtable arrived at conversational building through templating, not through code generation, which is why the results look like well-organized bases rather than bespoke software.
The real pricing, tier by tier
This is where reviews of Airtable most often go wrong, because the tier names changed and third-party summaries lag behind. Airtable's own pricing page and its plans documentation currently list four tiers:
- Free — $0. Capped at 5 collaborators with Editor or Creator permissions, 1,000 records per base, 1GB of attachment storage per base, and 1,000 API calls per workspace per month.
- Team — $20 per user per month, billed annually. 50,000 records per base, 20GB of attachment storage per base, 100,000 API calls per workspace per month.
- Business — $45 per user per month, billed annually. 125,000 records per base, 100GB of attachment storage per base, unlimited API calls per workspace per month.
- Enterprise Scale — custom pricing through the sales team.
Monthly billing costs more than the annual rates above. There is no separate AI add-on to purchase; AI credits are bundled into every plan, including Free.
AI credits are the number that decides your tier
Record limits get the attention, but for anyone building with Omni the credit allocation is the real constraint. Airtable's plans documentation lists 500 AI credits per editor on Free, 15,000 per billable collaborator on Team, 20,000 per paid user on Business, and 25,000 per paid user on Enterprise Scale at list price.
The jump from 500 to 15,000 is the meaningful one. Free-tier credits are enough to see what Omni does and not much else — a few app generations and some analysis runs will exhaust them. Team is the first tier where conversational building is a habit rather than a demo. The step from Team to Business more than doubles the price for a 33% credit increase, which means Business is a purchase you make for record capacity, unlimited API calls, and admin controls, not for AI headroom.
One billing detail saves real money and is easy to miss: on Team, commenters are billable, while on Business they are not. Read-only collaborators, form submitters, and share-link viewers are never charged on either plan. Teams with a large review-and-approve population sometimes find Business cheaper in practice than Team, despite the higher sticker price.
Integrations and governance
Airtable lists pre-built connectors for GitHub, Twilio, Zendesk, Salesforce, Box, Outlook Calendar, Google Calendar, Snowflake, Jira, YouTube, Slack, and email. For AI governance, admins can enable or disable AI at the workspace level and choose among model providers including OpenAI, Anthropic, and Meta, with an Amazon Bedrock option available. Airtable states that model providers never retain customer data and that customer data is not used for model training — a claim worth confirming in writing during procurement, but a clear one as published.
Where it breaks down
Hands-on testing tells a rougher story than the launch material. In a detailed June 2025 walkthrough, reviewer Nikhil R found a genuinely polished interface and one automation — a Slack integration — that worked correctly on the first attempt. He also hit a wall of limitations: Omni could not delete rows or remove duplicates, silently ignored a request to create a Kanban view, failed to build a logo-fetching automation and substituted placeholder images instead, and required several attempts before a rule modification took effect. Long tasks produced timeout errors reading "Assistant took too long to respond," and deleting a field then requesting a summary crashed the session badly enough to need a page refresh. His summary: not "ready to be called an 'App builder' yet."
Those findings are from mid-2025 and Airtable has shipped since, so treat them as a picture of the launch build rather than today's. The pattern they reveal is durable, though. Omni is strong at first-draft generation and weak at iterative editing — exactly the phase where most real app work happens.
Broader review data lines up. Airtable holds a 4.6 out of 5 average on G2 across thousands of reviews, with customization and integrations rated highest and ease of use and support responsiveness noticeably lower. Recurring complaints center on performance degradation with large datasets and the per-base API rate ceiling, which teams running concurrent automations and integrations hit sooner than expected.
Who should buy it
Airtable says over 500,000 organizations use the platform, and it cites agency Code and Theory saving more than 10,000 hours. Take vendor case studies for what they are, but the underlying fit is real: operations, marketing, and content teams that need structured internal tools fast and have no engineering budget to spend on them.
Start on Free to confirm Omni understands your domain, then move to Team, which is the honest entry point for sustained AI use. Reserve Business for record volume, unlimited API access, or a commenter-heavy org chart. If you need pixel-level UI control, offline access, or complex logic over hundreds of thousands of rows, Omni will produce a convincing prototype and then quietly hand the hard part back to you.