Airtable AI

  • Airtable AI — Full Profile

    1. One-line positioning

    Airtable AI = generative/agentic AI layer inside Airtable’s no-code relational database (Founded 2012, SF; Howie Liu/Andrew Ofstad/Emmett Nicholas; 500k+ orgs) that turns plain-English into working bases (Cobuilder), chats over your records (Omni), runs row-level AI field types (summarize/extract/categorize/translate/generate), drops AI steps into automations, and executes multi-step Field Agents/Hyperagent with compute + Slack deploy — billed as pooled AI credits (500/editor Free, 15k/paid-seat Team, 20k Business, 25k Enterprise Scale list) on top of per-seat pricing ($0 / $20 / $45 / custom annual), with top-up packs $20 per 10k credits. Chatting in Omni is free; running AI fields/agents/automations burns credits. It is the system-of-record AI layer, not a video/voice/music tool — distinct from Notion AI (doc-first) and Retool (code-first).

    2. Core features

    Capability Free ($0) Team ($20/seat ann) Business ($45/seat ann) Enterprise Scale (Custom)
    Seat billing 5 editors max per billable collaborator ≥Commenter per billable collaborator ≥Editor per paid user, list
    AI credits (pooled/mo) 500 per editor​ 15,000 per paid user​ 20,000 per paid user​ 25,000 per paid user​ list
    Records/base 1,000 50,000 125,000 500,000
    Automations/mo 100 25,000 100,000 500,000
    Cobuilder (prompt→base) ✅ (AI Labs) ✅ ✅ ✅
    Omni chat-over-data ✅ no credits for chat​ ✅ ✅ ✅
    AI Fields (summarize/extract/classify/translate/generate) ✅ limited ✅ ✅ ✅
    AI Automation steps ✅ limited ✅ ✅ ✅
    Field Agents / Hyperagent ❌ limited ✅ Hyperagent ✅ + private models
    Formula generation assistant ✅ ✅ ✅ ✅
    AI Admin controls ❌ ❌ ✅ ✅
    SSO/SAML, audit, DLP, SCIM ❌ ❌ ✅ ✅
    API (REST/Meta API) 1,000 calls/mo 100k/mo 100k/mo custom
    Top-up AI credits ❌ (must upgrade) $20/10k, $40/20k, $100/50k… same sales-led

    Credit burn: variable per action by input tokens + model + output tokens (e.g. summarize a 2k-char field ≠ extract entity from 200 chars); Omni asking is free, Omni deep-research/document-analysis burns; AI Fields burn on recompute (toggle off to stop).

    3. Pricing — Freemium (seat-limited, 500 cr/editor) + 3 seat tiers + credit top-ups

    • Free: $0, 5 editors max, 1,000 records/base, 100 automations/mo, 500 AI credits per editor/month​ (pooled in workspace), watermark-branded forms, 2-week history. Cobuilder/Omni/AI Fields usable but shallow; serious enrichment hits the 500-cr wall fast.
    • Team: $24/mo monthly / $20/seat/mo billed annually​ — 15,000 AI cr per paid user/mo, 50k records/base, 25k automations, 20 GB, Gantt/timeline, Interface Designer, standard sync. Floor for a team doing AI enrichment.
    • Business: $54/mo monthly / $45/seat/mo billed annually​ — 20,000 AI cr/user/mo, 125k records/base, 100k automations, 100 GB, Hyperagent + AI Admin controls + SAML SSO + two-way sync + premium integrations (Salesforce/HubSpot). Dept-scale.
    • Enterprise Scale: custom, 25,000 AI cr/user/mo list, 500k records/base, 500k automations, 1 TB, App Library/HyperDB/Enterprise Hub, audit+DLP, private AI model processing, SCIM, DPA.
    • AI credit top-ups​ (self-serve, pooled): 10k=$20/mo ($200/yr), 20k=$40, 50k=$100, 100k=$200, 200k=$400, 300k=$600, 400k=$800. Sales-led plans buy via rep. Credits reset per billing cycle (Free/Team/self-serve Business) or calendar month (sales-led Business/Enterprise).
    • CPT: pricing_type = freemium, starting_price = $0, free_tier = "500 AI credits/editor/mo (pooled), 5 editors, 1000 records/base, 100 automations, Omni chat free, AI Fields limited, no Hyperagent/SSO", note “seat $20/$45 ann + pooled cr 15k/20k/25k per paid user; top-up $20/10k; cr variable by tokens/model; Omni chat free, AI Fields/agents/automations burn; viewer/portal users don’t get allocation but spend pooled”

    4. Access Type (v2 spec)

    • access_type: web-app, desktop, api
    • access_display: 🌐 Web App · 🖥️ Mac + Windows desktop · 🔌 REST API + Meta API (Airtable Metadata API, automations webhook, AI Fields callable via API; Cobuilder/Omni/Hyperagent surface in-app + Slack)
    • Distinct from all 21 prior cards (none had desktop clients except none — Luma/Kling had mobile, Airtable adds desktop + enterprise API).

    5. Who should use it

    1. Ops/revops teams whose source of truth is already Airtable​ — AI Fields classify 10k leads overnight, no Zapier+OpenAI glue
    2. Content/marketing ops​ — brief→Cobuilder base→AI summary of reader comments→auto-categorize assets
    3. PM/bug-triage​ — Omni “summarize all P1 issues this sprint” over linked tables, no export to ChatGPT
    4. Customer-support routing​ — new ticket→AI classify intent→route→draft reply (automation step)
    5. Agencies building client portals​ — Interface + AI summary cards + Portals add-on ($120–150/15 guests)
    6. Enterprises wanting governed agents​ — Hyperagent with compute env + Slack deploy + DLP/audit on Enterprise Scale

    6. Who should NOT use it

    1. Video/avatar/music/voice generation​ → Veo/Luma/Kling/HeyGen/Suno/Murf; Airtable AI has zero media output
    2. Doc/wiki-first AI​ → Notion AI (cheaper per seat, better writing); Airtable wins on structured rows, loses on prose
    3. Free-tier production enrichment​ — 500 cr/editor evaporates on 1k-record bulk summarize; not a plan
    4. Pure code internal tools​ → Retool/ToolJet/Bubble; Airtable Interface Designer is block-limited, not custom UI
    5. Non-Airtable shops​ — credit meter + seat price stacked means migrating just for “AI” rarely pays; use OpenAI API + your DB
    6. Predictable flat AI cost​ — variable credit burn by tokens makes heavy AI-field recompute spikes hard to forecast (top-up packs soften, not solve)
    7. ASR/transcribe-only​ → Whisper; Airtable ingests text, doesn’t transcribe audio natively

     

Do Not Sell or Share My Personal Information Cookie Settings