a lead research bot that explains why each company fits

the routine

send the company pages you are allowed to research and describe a good customer in observable terms. this version starts with companies, without collecting personal addresses.

each tuesday at 2pm, review the company urls i gave you in your own browser. i’m researching independent design studios that publicly offer client workshops. exclude recruitment agencies, directories and companies outside my stated region.

make a shortlist in this chat. for each company include its name, website, the passage showing it fits, the source url and date checked. add a reason to include it, a reason it might not fit and the next question i should investigate. mark team size, budget and buying intent unknown unless a source directly supports them.

remove duplicate companies. do not guess email addresses, collect private contact details, add anyone to a crm or contact anyone. skip pages with access restrictions and list them at the end.

when

tuesdays at 14:00, using your approved company list

needs

a browser-equipped bot, company urls, a customer profile and exclusion rules; no crm connector

what comes back looks like this (an example, not a real inbox)

research shortlist for sam, fictional companies

cedar studio
fit: supplied services page says “client workshops for small teams”.
source: cedar studio services page supplied in chat, checked 12 november 2026.
possible mismatch: the page describes brand workshops; your offer concerns product testing.
unknown: budget and whether anyone is looking for help.
next question: does the studio run product testing workshops too?

excluded: jo’s talent desk, a recruitment agency.
not reviewed: priya design’s services page could not be opened.
no contacts collected; no outreach sent.

define fit using things a page can show

“likely to buy” asks the bot to infer too much. “publishes a workshop service page and works in this region” gives it evidence to look for. give exclusions equal weight: a directory listing can use all your keywords while representing no potential customer at all.

a source should support the specific reason for inclusion. a company website proves a company has a website, not that it has a budget or a project planned. use market research first when you have not yet decided which kind of company to approach.

keep research separate from outreach

the routine produces a table in chat. it neither adds records to a sales system nor writes to the people listed. that lets you discard a weak match before it becomes someone else’s unwanted message.

if you later connect a supported email app, sends ask first unless you give the bot a daily sending allowance. one exception: a bot on Full access with no sending rule of its own may send up to 25 a day without asking, so give full-access bots Ask every time under sending.¹ email triage is a separate routine for handling replies once conversations actually exist.

check the rules before contacting anyone

before cold outreach, check the rules that apply where you and the recipient are, including consent, identification and opt-out requirements. keep an exclusion list and respect requests not to be contacted. public contact information does not settle those questions. this is not legal advice.

also respect source sites’ terms and personal data rules when building the list. the web scraping guide keeps collection scoped to named fields. do not treat a bot’s fit score or a polished opening line as evidence that contact is welcome.

the guard rails

• make a source passage a condition of inclusion in the shortlist
• exclude existing customers and companies on your do-not-contact list before research
• avoid inferred personal emails and unsupported claims about intent to buy
• review applicable outreach rules before turning any research entry into a message