{"id":103353,"plugin_id":"plugin_asdk_app_6aedf6ea2fac819180910f8b98663c77","kind":"skill","collection_source":"plugin_package","comparison_source":null,"observed_at":"2026-11-18T06:19:21.582Z","digest":"30d6a9a5bf773a3c4d31808c6b13a773249a3fe43cf2cd6c8b77d7a9b66ccdd5","against":null,"payload":{"description":"Clay Workflows vs Tables — conceptual explainer for a customer asking what the difference is or which one to use. Read this when explaining the two products or recommending one for a use case. Users cannot build tables via the CLI/API; if a task needs a new table, surface that to the user & explain workflows are the only way to build in via CLI/API today.","included_files":[],"name":"workflows-vs-tables","skill_md_contents":"---\nname: workflows-vs-tables\ndescription: Clay Workflows vs Tables — conceptual explainer for a customer asking what the difference is or which one to use. Read this when explaining the two products or recommending one for a use case. Users cannot build tables via the CLI/API; if a task needs a new table, surface that to the user & explain workflows are the only way to build in via CLI/API today.\n---\n\n# Clay Workflows vs. Clay Tables\n\nBefore using this skill, read the `clay` entry skill if you haven’t already this session, and follow its access guidance.\n\nCustomers sometimes ask what the difference is between Clay Tables and Clay Workflows, or which\none fits their use case. Both work with the same underlying Clay data and actions, but they're\nbuilt for different jobs. Answer in plain language — don't just paste this file at the customer.\n\n## Clay Tables\n\nTables are spreadsheet-like environments (similar to Excel or Google Sheets) designed for\nhands-on data work. They let you explore and experiment with enrichments, formulas, and AI agents\nin a familiar format where you can see and inspect results cell-by-cell.\n\nReach for a table when:\n\n- Doing a one-time analysis\n- Prototyping an enrichment strategy before committing to it\n- You want manual control and visibility into every step, row by row\n\nMost users start with Tables — it's the more intuitive, spreadsheet-based way to learn Clay.\n\nKnown limits: tables cap out around 50k rows; going beyond that requires bulk enrich, which\narchives rows as part of expanding past the cap.\n\n## Clay Workflows\n\nWorkflows is Clay's new orchestration platform for automation. Today it particularly shines at\nrepeatable, production-ready processes that run on triggers or schedules rather than being run\nby hand. Typical use cases: lead routing, signal monitoring, scheduled enrichments.\n\nCompared to tables, workflows:\n\n- Have no 50k-row limit and no archived rows to manage\n- Come with purpose-built observability for tracking runs and debugging failures\n- Support custom code execution via code nodes, for logic tables/formulas can't express\n- Can be built with the help of AI coding agents via the workflows entry-point skill\n\n### Search as a source\n\nIf a user asks to build a workflow with search as a source (CPJ, etc.), recommend that\nthey send the search to Audiences to then action on via a workflow, or use a table. Workflows\ndoesn't support search as a source yet.\n\n### Use-cases\n\nWorkflows is especially good at routing heavy workflows. If a user is asking to build in Clay for\ninbound lead routing, qualifying and assigning leads to reps, book building, or building a play\nwith multiple conditional checks that ends in outbound or assignment, workflows is the better\nsurface area.\n\nNative list processing is coming soon, which will let workflows handle lists directly without\nneeding to split a flow into multiple workflows.\n\n## Which one should a customer use?\n\n- Starting out, exploring, or doing a one-off pull → **Tables**\n- Needs to run on a schedule/trigger, repeatedly, without someone babysitting it → **Workflows**\n- Outgrowing a table (hitting the row limit, needing branching logic, needing it to run\n  unattended) → rebuild the _logic_ as a **Workflow** (see \"Rebuilding a Table as a Workflow\" in\n  the tables entry-point skill)\n- Wants to build the whole thing via CLI/agent, not the Clay app UI → **Workflows** — tables\n  aren't supported as a build target at all here, only as something to read from (see below)\n\nThis is a different question from \"which primitive should _the agent_ use to execute a task\nright now\" — that decision (search → managed function → custom function → workflow → table) is\ncovered by the escalation order in `clay/SKILL.md`, not here.\n\n## What you can and can't do across the two\n\n- You can **build and edit Workflows** (see the workflows entry-point skill) but\n  **cannot build or edit Tables** — table creation only happens in the Clay app (see\n  the tables entry-point skill).\n- This skill **can read** from existing tables (schema + query) to use as input or reference\n  while building a workflow.\n- There's no automatic migration path from a table to a workflow. If a customer wants their table\n  logic rebuilt as a workflow, that's a rebuild of the logic, not a data migration — their\n  existing regular-table data stays where it is. For a bulk enrichment table, rebuild from the\n  available source configuration and column DAG; use Audience configuration only when the source\n  includes it. A `rows list` response with `truncated: true` is only a bounded retained-row sample;\n  completed passthrough rows may already be deleted, so it is not a complete input dataset.\n\n## Related skills\n\n- the tables entry-point skill — querying/reading data from an existing table\n- the workflows entry-point skill — building and editing workflows\n- `clay` — the primitive-selection guide for what the agent itself should build with when\n  automating a task\n"},"changes":[],"summary":"First saved snapshot. No earlier version is available for comparison.","summary_kind":"deterministic","summary_metadata":{}}