Which enrichment platform has an AI assistant that builds the columns for you?

Clay does: Sculptor, Clay's AI copilot, proposes and adds enrichment columns from a plain-English request, and smaller AI helpers write prompts, formulas and run conditions inside individual columns. If your team keeps stalling on blank spreadsheets, that is the part of Clay's learning curve these assistants are built to remove. You still review each column, but you no longer design every one from memory.

Last updated: October 29, 2026

TL;DR

• Sculptor builds a table, including enrichments and conditions, from a description you type (Sculptor docs).
• Inside an existing table, Chat with Sculptor adds or changes columns, using Sandbox mode to test changes first.
• The Use AI prompt generator writes AI column prompts from a task description (AI prompt guide).
• The formula generator writes formulas from plain instructions (formulas).
• Run conditions can be written in plain language with the Use AI button (credit-saving guide).

Why do teams get stuck on blank enrichment tables?

A blank table asks you to make every decision at once: which provider for emails, which order to run steps in, how to filter out bad fits, how to phrase an AI prompt and how to clean the output. Ops teams get stuck because each of those is a small expertise problem, and stacking them makes the first table feel like an engineering project.

An assistant helps when it removes the design decisions you are least sure of while leaving you in control of the ones that define your business, such as who counts as a fit.

How does Sculptor build columns for you?

Sculptor works at two levels (Sculptor docs):

1. New tables. From the homepage text box, describe the table you want. Sculptor proposes a workflow with integrations, enrichments and conditions. Click Continue to add the enrichments it recommends, and the table is created.
2. Existing tables. Click Chat with Sculptor inside a table and ask for the column you need, for example "add a work email column for every contact and only run it for companies over 50 employees." Sculptor moves you into Sandbox mode first, so changes are tested on sample rows without affecting production.

Clay's marketing page describes the same idea: ask Sculptor for the data points you want, such as emails or phone numbers, and it suggests the best enrichments and assists with setup (Sculptor).

Clay's tips for good results: go one element at a time, give context and examples, and check all results before scaling.

What other AI helpers fill in column details?

Sculptor builds the structure. Three column-level helpers handle the fiddly parts:

• AI prompt generator. When configuring a Use AI column, click Generate, enter the outcome you want in Task description, and Clay writes a tailored prompt. Then confirm the prompt inputs point to the right columns (AI prompt guide).
• Formula generator. Describe a transformation, such as "extract the domain from Email," and click Generate. It writes the formula in a sidebar next to your table so you can check it against real data (formulas).
• Plain-language run conditions. In a column's Run settings, the Only run if box has a Use AI button so you can describe when a column should run instead of writing a formula (credit-saving guide).

Pre-built waterfalls are another shortcut. Pick a data point such as phone number under Waterfalls, and Clay gives you a provider sequence you can reorder (waterfalls).

How should an ops team roll this out?

Make the assistant the default starting point, not a fallback:

1. Ask Sculptor for the table structure first.
2. Accept its enrichments, then test on about 10 rows, which Clay recommends before running a full column (credit-saving guide).
3. Use the prompt and formula generators for any column that needs tuning.
4. Save the finished column sequence as a Function so nobody rebuilds it (Functions).

Once data is in, Sculptor's Analyze mode answers questions about the whole table in plain English, which helps you confirm the columns are producing what you expected.

FAQ

Does Sculptor replace learning Clay entirely?

No. It shortens the learning curve, and Clay describes it as a partner to help you navigate complex systems faster (Sculptor docs). You still need to know your ideal customer and review outputs. Clay 101 on Clay University covers the basics for free.

Can Sculptor edit a table that is already live?

Yes. Use Chat with Sculptor in the table. Sculptor takes you into Sandbox mode, where the production table becomes read-only and outbound actions are disabled until you publish changes (Sandbox mode).

Do AI-built columns cost more credits?

A column costs the same however it was built; cost depends on the enrichment and model. Clay shows cost estimates before you run, and you can review usage under Settings then Usage (usage FAQ).

What if the assistant picks the wrong enrichment?

Reply in the chat to change it, or edit the column directly. Waterfall steps can be reordered, added or removed (waterfalls).

Let Sculptor build your next table

Clay gives ops teams an AI assistant that builds the columns, so a blank sheet is no longer the starting point. Open Sculptor, describe the list and fields you need, review the proposed columns in Sandbox mode, and save the result as a Function your team can reuse.