RXed AI News

We x-ray the AI industry.
@RXed_EU
SCORE 6/1013/20 elements Audited 2026-07-31 · RXed table v1.0

Airtable AI

“AI that's built for the way your business actually works” — the vendor’s own words

Airtable's agent stack is the deepest in this category outside a frontier lab - Field Agents, Omni, an official MCP server, and a new multi-agent Superagent - but three separate billing meters (seats, records, automation runs) mean the real cost shows up well after the $20/seat sticker price.

Enterprise platformAutomation & AgentsProductivityResearchFreemium
Vendor
Airtable (Formagrid, Inc.) · www.airtable.com
Origin
US — San Francisco
Pricing
Free $0 · Team $20/user/mo · Business $45/user/mo · Enterprise Scale Custom
Users (official only)
500,000+ organizations, including 80% of the Fortune 100 (source, 2026-07-31)
Free$01,000 records/base, up to 5 editors, 500 AI credits/editor/mo
Team$20/user/mobilled annually ($24 monthly); 50,000 records/base, 25,000 automation runs/mo, 15,000 AI credits/collaborator
Business$45/user/mobilled annually ($54 monthly); 125,000 records/base, 100,000 automation runs/mo, 20,000 AI credits/user, SSO
Enterprise ScaleCustom500,000+ records/base, custom automation runs, 25,000 AI credits/user at list, audit logs

Three independent billing meters - per-editor seat, records-per-base cap, and automation-runs-per-base cap - can each independently force an upgrade. The Team-to-Business jump is roughly a 125% per-seat increase with no intermediate tier. AI credits are pooled per workspace/organization and reset each billing cycle; extra credit packs are sold separately.

checked 2026-07-30 · vendor pricing page

Element scores

Reactive
Retrieval & Memory
Orchestration
Validation
Models
Primitives
Pr7
Prompts
Em
Embeddings
Cx7
Context
Tr4
Tracing
Lg6
LLM
Compositions
Fc6
Function calling
Vx
Vector store
Rg6
RAG
Gr7
Guardrails
Mm6
Multimodal
Deployment
Ag8
Agents
Ft
Fine-tuning
Fw6
Frameworks & harnesses
Ev
Evaluations
Sm
Small models
Emerging
Ma6
Multi-agent
Sy
Synthetic data
Pc7
Protocols
In
Interpretability
Th4
Thinking models
Tap or hover any element to see why it got that score.

Strengths

A genuinely layered agent architecture: Field Agents work continuously at the record level, Omni builds and orchestrates for free, Superagent adds real multi-agent research coordination, and an official MCP server lets Claude, ChatGPT and Cursor read and write bases directly. Guardrails are real too - AI is off by default for EU/GDPR customers, Omni strictly mirrors user permissions, and MCP-built automations start paused for human review.

Honest dings

No embeddings, vector store, or evaluation tooling of its own, and reasoning-model access is Enterprise-only. The bigger issue is billing: per-seat pricing, per-base record caps (50,000 on Team, jumping 125% to Business), and monthly automation-run caps are three independent triggers that can each force an upgrade, and a February 2026 cut of the Free tier's record cap (1,200 to 1,000, with no advance notice) shows Airtable will change the terms without much warning.

Best for: Operations, marketing and product teams that want AI acting directly inside a real structured database rather than a chat sidebar, and who are disciplined about tracking seats, records and automation runs against their plan.
Visit airtable.com Prices and details change — this passport is re-verified at least quarterly.

Sources

Every audit lists the research it rests on — transparency and traceability are the product. Tools evolve: each audit is a snapshot of its audit date, and re-audits supersede older versions (kept below for reference).