Key takeaways
- Every form on optechsol.llc writes straight into GoHighLevel: contact, tags, and a New Leads opportunity, no manual entry on my end.
- New leads get tagged "ai research pending" the moment they come in.
- A scheduled task drains that queue, researches the business behind the lead, and writes a pre-call dossier note directly on the contact.
- It never contacts the lead itself. It only prepares me, so by the time I call, I'm not starting cold.
- It runs every two hours through the working day, not instantly, because a queue that drains steadily is easier to trust than one racing to be first.
I used to pick up the phone for a new lead knowing almost nothing beyond a name and whatever they typed into a form. That's a rough way to start a call. You spend the first five minutes just orienting yourself while the other person is already wondering if you know anything about their business at all. The lead itself wasn't the problem. The blank space between "someone filled out a form" and "I actually know who they are" was.
Where the lead comes from
Every form on the site, the main apply gate and the individual landing pages, posts to a Cloudflare Function at /api/lead. That function writes the contact into GoHighLevel directly, applies the right tags, and creates a New Leads opportunity, all without me touching anything. The moment that happens, the contact also gets tagged "ai research pending." That tag is the whole trigger for what happens next. Nothing fires off automatically before that tag exists, and nothing gets left behind once it does.
What the research pass actually does
A scheduled task checks for that tag every two hours during the working day. For each lead it finds, it researches the actual business behind the name: their own website, what their reviews say, whether they show up on the socials you'd expect a business like theirs to run. Then it writes all of that up as a structured note directly on the contact record in GoHighLevel, titled clearly enough that I can open the contact right before a call and read the whole picture in under a minute. Once the note is written, the tag flips from "ai research pending" to "ai research done," so the queue always shows exactly what's left to process and nothing gets researched twice.
What it deliberately does not do
It never reaches out to the lead. Not an email, not a text, nothing. The entire job is to prepare me, not to represent me to someone who hasn't heard from a real person yet. That boundary was intentional from the start. A lead who filled out a form on my site is expecting to hear from a person, and the first message they get should actually be from one. Research happening quietly in the background before that call is a completely different thing than an automated message pretending to be me.
The lesson worth being honest about
Two hours felt slow when I first set this up. I kept wanting to make it check every fifteen minutes so nothing sat around stale. But most leads don't get called back in the fifteen minutes after they fill out a form anyway, and a tighter check interval on a task that's also running alongside a dozen other scheduled jobs just means more contention for the same limited number of automation slots at any given moment. Once I actually watched how the fleet behaved when too many tasks tried to run too often, the two hour window stopped looking slow and started looking like the right amount of patience for a job that isn't actually time-critical. Not everything needs to be instant. It needs to be done before the moment it's actually used, and a call rarely happens inside the first two hours after someone submits a form.
Why this matters for Bay Area and San Jose businesses
If you're fielding leads through a website form, you already have most of what this needs: a CRM that can hold a tag and a note, and a form that already knows the lead's name and business. The part that's usually missing isn't data, it's someone or something taking the ten minutes to turn that name into an actual picture of the business before the call happens. That ten minutes, done consistently on every single lead instead of only the ones you happen to have time for, is the difference between sounding like you did your homework and sounding like everyone else who cold-called them that week.
Practical steps
- Make sure every lead source writes into one place. Scattered forms with no common destination can't be researched consistently.
- Use a tag as the trigger, not a guess about timing. A tag is checkable. A guess about when someone might call isn't.
- Keep the automation on the research side of the line, never on the outreach side. Preparing you is safe. Speaking for you to someone who hasn't heard from you yet isn't.
- Pick a check interval that matches how the lead actually gets used, not the fastest interval you can technically run.
We build systems like this at On Point Tech Solutions in Los Gatos, serving San Jose and the whole Bay Area. Every one, you own outright, done for you or taught with a full SOP. If your leads are landing in a form and going cold before anyone researches them, that's a fixable gap. Start at optechsol.llc.