Hey Diego, yeah, I agree.
I usually provide the company name and some firmographi data. I’m now building a Clay workflow that also includes an additional_information field.
Depending on the company and data source, this might include the address for a local business, the country for a SaaS company, or the industry if the data came from Apollo or a similar source.
I then pass all of that context into the Claygent through the additional_information field.
One reason I’m not using newer models is that they’re significantly more expensive.
Even a model like 5.6 Luna has fairly high input/output-token costs. That alone is not a major issue, but reasoning tokens are counted as output tokens, which add to the cost much more.
When running large campaigns, using bigger models therefore becomes difficult to sustain.
Ideally, Clay would allow us to use models through OpenRouter, since that would provide more flexibility around model selection and costs.
