Illingworth claims 250 contacts enriched in 30 minutes

Richard Illingworth posted a Claude Code workflow that he says enriches 250 cold email contacts in about 30 minutes, against what he calls a 12-hour job in Clay.

The loop he described starts before the list exists. It takes a short brief on what he sells and who buys, asks follow-up questions to turn that into an ICP, then builds Apollo filters off a handful of companies he already knows are a good fit. From there it splits the list into two tiers, tier 1 being anyone already running a sequencer, tier 2 the wider pool. It pulls each contact's website, job listings and recent news into new columns, writes a three-step sequence off that enrichment data, and checks every merge field has a matching column before anything launches. After send it reads results by segment against the same enrichment columns and flags which emails to rewrite.

Illingworth's framing is that the columns are the asset. "every job after enrichment can then run off those columns," he wrote, which is the part that makes the 30-minute number reusable rather than a one-off.

Richard Illingworth

@RCIllingworth

X

every job after enrichment can then run off those columns.

The baseline is what got challenged

Eric Nowoslawski replied asking for receipts, not on the build but on the comparison. He offered to put an engineer on the Clay table that supposedly took 12 hours, on the grounds that the same work should run in under 15 minutes using the Clay CLI (a command line interface for driving Clay tables rather than clicking through the UI).

That is the whole disagreement. Nobody in the exchange is arguing that a Claude Code agent cannot do ICP definition, filter building, enrichment and sequence drafting in one pass. The argument is whether 12 hours is a fair description of what Clay costs for 250 contacts, or whether it describes a table built by hand by someone not using the CLI. If Nowoslawski is right, the comparison shrinks from 24x to roughly 2x, and a 2x gap is not a reason to rebuild your stack.

For people sending, neither number has a published table behind it yet. Both are single-operator claims about their own setup, with no shared list, no shared enrichment spec and no third party timing either run. The practical move is to time your own Clay table on a batch you already know, including the parts that actually eat the hours, which is usually waterfall enrichment and cleanup rather than the column config. Then you have a baseline that applies to your lists instead of someone else's.

Eric Nowoslawski

@ENowoslawski

X

hey send me the clay table link that took 12 hours to do this and I'll get an engineer to look at this because you should be able to do this in under 15 minutes with the clay CLI