Airtable AI: Omni, Field Agents, and the Credit Meter
On June 24, 2025, Airtable relaunched as an “AI-native app platform” — CEO Howie Liu called it a “refounding moment.” A year on, that’s real and generally available, not a roadmap. But “Airtable AI” is three distinct products, it runs on a credit meter most people don’t price out until they hit it, and everything it does is anchored to one place: a base. Here’s what each piece actually does, what it costs in credits, how far the native automation reaches, and where it stops.
The three pieces of Airtable AI
Airtable’s AI isn’t one assistant. It’s three layers, all GA, described on airtable.com/product/ai:
Omni is the flagship — “your expert AI app builder, data analyst, and web researcher.” From a natural-language prompt it builds and iterates apps, tables, interfaces, and automations, and it answers questions about your data conversationally. Omni is the successor to 2024’s Cobuilder (some third-party write-ups still list them as two products — they aren’t; Cobuilder folded into Omni at the relaunch).
Field Agents are “AI-powered researchers, analysts, and content creators that perform workflow tasks at scale” — running inside records across a base to enrich leads, generate content, extract data from transcripts, triage feedback by sentiment, do web research, and translate. You spin one up by asking Omni to configure it.
Field-level Airtable AI applies AI to individual fields: text generation, summarization, classification (it can auto-fill single- and multi-select options), extraction from attachments and the web, sentiment analysis, translation, numerical computation, and linked-record suggestion. It runs per record — manually or automatically when input values change — and won’t overwrite a cell you edited by hand. You choose the underlying model from multiple providers (OpenAI and Anthropic among them), subject to your plan and admin settings.
Taken together this is a genuine AI platform, not a chatbot bolted onto a spreadsheet. If your work already lives in Airtable, Omni and Field Agents do real work inside it.
The catch is the credit meter
Airtable moved AI off the old per-seat add-on and onto bundled, non-rolling credits. Per Airtable’s billing docs, each plan includes a monthly allotment:
| Plan | Monthly AI credits |
|---|---|
| Free | 500 per user (Editor and above) |
| Team | 15,000 per billable collaborator |
| Business | 20,000 per paid user |
| Enterprise Scale | 25,000 per paid user |
Two details do the damage. First, credits don’t roll over — Airtable states plainly that unused credits “expire at the end of the month.” Second, the useful actions spend real quantities: a data-analysis response from Omni runs about 10 credits, so the 500-credit free tier is a few dozen questions, not a daily habit. (Building apps and agents with Omni is free and burns no credits — it’s the analysis and field work that meters.)
When you run out, you buy packs — 10,000 for $20/month, 100,000 for $200/month, up to 400,000 for $800 — or wait for the reset. None of this is a reason to avoid Airtable AI; it’s a reason to price the meter before you build a required process on top of it, because the person who hits the wall is whoever happens to click next.
Airtable Automations: the native builder
Separate from the AI, Airtable ships a mature no-code automation builder. It’s trigger-and-action, and the menu is broad.
Triggers: record enters a view, record created, record updated, record matches conditions, form submitted, at a scheduled time, webhook received, button clicked, plus Google Workspace and Outlook triggers.
Actions: create, update, and find records; run a script (custom JavaScript — the escape hatch that calls external APIs and branches logic, though it’s not available on Free); send email; post to Slack or Teams; Google Workspace actions (Gmail, Calendar, and more); a document automator; and connectors for Jira, Salesforce, GitHub, Twilio SMS, Facebook Pages, and Hootsuite. There’s also a “Generate with AI” action that drops model output into an automation step.
The limit that matters for planning is the monthly run count — one run per trigger fire, including failed attempts, regardless of how many actions the automation has:
| Plan | Monthly automation runs | Run history |
|---|---|---|
| Free | 100 | 2 weeks |
| Team | 25,000 | 6 months |
| Business | 100,000 | 1 year |
| Enterprise Scale | 500,000 | 3 years |
There are structural caps too — on the order of 50 automations per base and 25 actions per automation — so very large orchestrations get split across bases. For most teams the run count is the ceiling to watch: a webhook-triggered automation on a busy base can chew through the Free tier’s 100 runs in an afternoon.
And it all orbits the base
Here’s the structural thing, and it isn’t a knock — it’s just where the center of gravity sits.
Airtable’s AI reasons over records that are already inside a base. Omni analyzes your tables; Field Agents work your rows; field AI fills your cells. And the Automations’ outward reach — the email, the Calendar invite, the Slack post — are outputs of a base automation: they fire from a base event (a record created, a form submitted, a schedule, a webhook). So the honest version isn’t “Airtable can’t touch other apps.” It plainly can, as automation outputs and connected actions.
Scheduled work is no longer the dividing line — ChatGPT and Claude both run tasks on a timer without anyone watching. The line that still holds is event triggers. Nothing kicks off because something happened in one of your apps; it kicks off because the clock said so, which means anything time-sensitive waits for the next run.
What Airtable doesn’t do is act on events that originate outside a base. Its triggers are base-centric; nothing in Airtable watches your inbox as a first-class event source, runs your calendar day to day, or reasons across your whole workday. And it all assumes the prerequisite most people skate past — your data has to be in an Airtable base first. For a team that already runs on Airtable, that’s free. For someone whose work lives in email, a calendar, and a dozen apps, “build the base” is the actual cost.
Where Carly fits
Carly starts from the other end. Instead of making you faster inside a base, it acts across the apps your work already lives in — email, calendar, tasks, Slack, and ~260 more — triggered by real events rather than a prompt typed into a table.
- No base to build. Carly connects Airtable natively, or via a scoped personal access token as a bring-your-own-key connection — but it doesn’t require your work to move into Airtable first. It reads and writes the base and your inbox, calendar, and the rest.
- Event-triggered, not just prompt-triggered. A Carly workflow can start when an email arrives, on a schedule, or on a webhook — updating an Airtable record, drafting the reply, and posting to Slack as one flow, unattended. That inbound-email trigger is the one Airtable’s own automations don’t have.
- The pricing shape is different. Non-AI, Zapier-style workflow steps run free and unlimited; AI agents start at $35/month. The plumbing that moves data between Airtable and your other apps isn’t drawing down a monthly credit balance the way an Omni response does.
The two aren’t really competing for the same job. Airtable AI is the best way to reason over data you’ve structured into a base, and its Automations are a strong engine for work that starts in that base. Carly is for the work that starts outside it — in your inbox or calendar — and has to cross several apps to finish. Plenty of teams run both: Airtable as the system of record, Carly as the thing that acts on it and everything around it.
Mapping the wider field? Best AI workflow automation tools covers the category, and best Airtable integrations has the connection patterns in more detail.
FAQ
What is Airtable AI? Airtable’s built-in AI, generally available since the June 2025 “AI-native” relaunch. It has three pieces: Omni (an agentic assistant that builds apps and analyzes data), Field Agents (AI that runs workflow tasks across records at scale), and field-level Airtable AI (summarize, classify, extract, translate, and generate inside fields).
What is Airtable Omni? Omni is Airtable’s flagship AI — an app builder, data analyst, and web researcher driven by natural language. Building and iterating apps and agents with Omni is free; conversational data-analysis responses cost roughly 10 credits each. It replaced the earlier Cobuilder.
How much does Airtable AI cost in credits? AI is bundled as monthly credits that don’t roll over: 500 per user on Free, 15,000 per collaborator on Team, 20,000 per paid user on Business, and 25,000 on Enterprise Scale. You can buy extra packs — for example, 10,000 credits for $20/month or 100,000 for $200/month.
What are Airtable’s automation limits? Automation runs are capped monthly: 100 on Free, 25,000 on Team, 100,000 on Business, and 500,000 on Enterprise Scale. A run is counted each time a trigger fires, including failures, no matter how many actions the automation contains. The “run a script” action is unavailable on the Free plan.
Can Airtable AI act in my email or calendar? Airtable Automations can send email and create calendar events through Google Workspace and Outlook actions, and post to Slack or Teams — but as outputs of a base automation. The AI itself is anchored to records inside a base and doesn’t triage your inbox or act on events that start outside Airtable; that cross-app, event-driven work is where a tool like Carly fits.
Does Airtable have an API? Yes. Airtable exposes a public REST API that any individual can use with a scoped personal access token, created in account settings (personal access tokens replaced legacy API keys, which were shut off in February 2024). That’s how Carly connects to Airtable as a bring-your-own-key integration.
More: Best Airtable integrations · Airtable MCP · Airtable alternatives · Best AI workflow automation tools
Ready to automate your busywork?
Carly schedules, researches, and briefs you—so you can focus on what matters.
See what people say
"Before Carly, I relied on a Calendly link, but the whole process felt impersonal and not very professional. Carly changed that by handling all the back-and-forth, so I'm no longer stuck in endless email threads trying to line up schedules.
Now Carly reaches out to candidates, shares my real-time availability, lets them pick a slot, then sends a Zoom link and drops it straight into my calendar. She sends reminders to both of us before each call, which has significantly reduced no-shows and last-minute confusion.
On top of scheduling, Carly acts like a full executive assistant, sending me my schedule the night before so I can prepare for each call. It reminds me of the old x.ai assistant, but Carly is noticeably smarter, faster, and better suited to my healthcare recruitment business."
