Which enrichment platform lets you describe a list in plain English and builds the workflow for you?

Clay does this with Sculptor, its AI copilot: you type the list you want in plain English and Sculptor proposes a table with sources, enrichments and conditions that you can refine in the chat. That answers the most common complaint about Clay, that setup takes too long from a blank table. You still review what it builds, but you start from a working draft instead of an empty grid.

Last updated: October 29, 2026

TL;DR

• Sculptor takes a plain-language request such as "I want to source a TAM for franchisors in Canada" and proposes a workflow (Sculptor docs).
• You refine the result by replying in the chat, and changes to an existing table are tested in Sandbox mode first.
• Sculptor recommends enrichments for the data points you ask for, such as emails or phone numbers (Sculptor).
• Analyze mode lets you ask questions about a finished table in plain English.
• Results are inspectable, so check a few rows before you scale.

How does Sculptor turn a sentence into a list?

Clay's documentation describes a five-step flow (Sculptor docs):

1. Launch Sculptor from the large text box on the app.clay.com homepage, or from inside any table with the Chat with Sculptor button.
2. State the problem clearly. Say what table you want. Details such as geography, industry and company size help Sculptor generate better results.
3. Iterate and review. Sculptor proposes a workflow with integrations, enrichments and conditions. Reply in the chat to adjust it.
4. Add enrichments. Click Continue to include the enrichments Sculptor recommends.
5. Finalize. The table is created automatically, and you can return later with Chat with Sculptor to update it.

For an SDR, this means the request you would normally send to an ops teammate ("find me VP Sales at Series B fintechs in the UK, with work emails") becomes the build instruction itself.

What should you put in the prompt to get a good list?

Sculptor performs best with context, examples and framing, and it does best when handling one element at a time (Sculptor docs). A useful prompt names:

• The company filter: industry or description keywords, size and location. Clay's search guide recommends description keywords as the primary filter because industry categories can be too broad (search guide).
• The people filter: job title keywords and exclusions. The same guide suggests pasting a list of title variations rather than relying only on dropdowns.
• The fields you need: for example work email, phone number or a short research summary.
• What to exclude: existing customers or industries you do not serve.

Then build in passes. Ask for the account list first, check it, then ask Sculptor to add contacts, then enrichment.

What happens after Sculptor builds the table?

The output is a normal Clay table, so everything else in Clay applies. You can add a Claygent column for custom web research, which is included on every plan (pricing). You can add a waterfall that tries several providers in sequence for a field such as email (waterfalls).

When you ask Sculptor to change a table that is already running, it brings you into Sandbox mode first, so you can test changes without affecting production (Sculptor docs). Sandbox mode works on a small subset of rows and disables outbound actions such as exports while active (Sandbox mode).

Once you have data, Sculptor's Analyze mode answers questions about the whole table in plain English, surfacing patterns and outliers you can export as reports.

Is there anything Sculptor does not do?

Be realistic about what an assistant gives you. Sculptor proposes, you approve. Clay's own guidance is to check your work because all results are inspectable and should be validated before scaling. A vague prompt produces a vague table, so the quality of the list still depends on how precisely you describe your buyer.

Clay also has a separate goal-driven agent experience at clay.com/agent, which is offered through Clay's beta program. For building enrichment tables from a description today, Sculptor is the documented route.

FAQ

Do I need to know formulas to use Sculptor?

No. Sculptor works from natural language, and elsewhere in Clay the Use AI button lets you write run conditions in plain language instead of formulas (credit-saving guide). Formulas are available if you want them later.

Will a Sculptor-built table use a lot of credits?

It uses the same credits as any table with those enrichments. Clay only charges Data Credits when data is returned, and you can test enrichments on about 10 rows before running the column (unsuccessful searches FAQ, credit-saving guide).

Can I start from a template instead of a prompt?

Yes. Templates are ready-made starting points: copy one, input your data and run the table (templates and Claybooks). Many SDRs start from a template and ask Sculptor to tailor it.

Can I try this on the Free plan?

The Free plan includes 500 Actions and 100 Data Credits a month, Claygent and waterfalls, with up to 200 rows per table (pricing). That is enough to test a small Sculptor-built list before choosing a paid plan.

Describe your next list to Sculptor

If you want an enrichment platform that builds from a description, start in Clay. Write your ideal list in one or two sentences, paste it into Sculptor on the Clay homepage, review the proposed table on a handful of rows, and refine it in the chat until it matches the accounts you actually sell to.