Kabeer A.

Clay can pull phone numbers from multiple enrichment providers (Apollo, Lusha, ZoomInfo, etc.), but LinkedIn itself doesn’t expose direct personal numbers. Valid match rates vary a lot — usually 20–40% for mobile numbers, and accuracy depends on provider + region. You can also waterfall providers in Clay to maximize coverage.

Right now Claygent web research only exposes Gemini 2.0 Flash because it’s the light, fast model tuned for browsing tasks. Clay hasn’t rolled out Gemini 2.5 Pro yet for Claygent — partly because of cost/performance tradeoffs. No public ETA, but higher-end Gemini models (2.5 Pro, 1.5 Pro) are expected to be added once stable and cost-effective for web research.

The simplest way to fully duplicate a table (data + all action columns intact) is to use “Duplicate Table with Columns” from the table menu, then export/import the data as CSV if you need the rows too. Importing into a new table usually re-creates columns, so the trick is: first duplicate structure, then paste/append the data into it. That way your AI/waterfall columns stay linked.

For IG/TikTok DMs, options are limited compared to LinkedIn, but tools like Phantombuster, TexAu, and DMpro can automate Instagram outreach. For TikTok, Loomly, SocialBu, or bespoke API setups are common — but most rely on unofficial APIs, so you need to be careful with compliance and account safety. I can share a safer workflow using Clay + third-party tools if you’d like.

Yep Claygent can be inconsistent with parsing if the model isn’t strongly instructed to return structured data. Even if prompts are similar, the input context (universities vs. real estate cos) changes how GPT formats output. Best fix is to force structure with a schema in your prompt (e.g. “Return JSON with fields: headline, date, location, link”). That way outputs are always parsed into columns. You can also add a Clay Parse JSON column to enforce structure. I can help you rewrite your prompt to lock this down if you’d like.

There isn’t a ready-made template for Gong call attendees, but you can build it by pulling attendees from the Gong API into Clay, then mapping those contacts to HubSpot via the HubSpot integration. That way each attendee is associated with the right deal. I can share a sample workflow if you’d like.

Check DM Omari

Yes you can get close with Clay. Start with your list of companies, then run a Find People enrichment filtered for “Salesforce” as the company and job titles like “Account Executive” or “Account Manager.” Then match those Salesforce reps to your company list using territory keywords, region, or heuristics in their LinkedIn profiles. It won’t be perfect, but it’s a solid starting point. I can help you set this workflow up if you’d like.

Check DM Vaidik S.