Build an investor list from this week's funding news
Funding announcements name the investor, the stage and the cheque, every weekday, in free RSS feeds. Filter to stories that mention both a raise and an amount so you never pay a model to read a conference ad. Score each investor against your thesis with a stated reason, and make the draft open on the deal they just did or it does not get saved.
Fundraising tools charge four figures a month to tell you who is writing cheques. The announcements are public and they are dated.
Recency is the signal a database cannot give you
An investor database answers a question you mostly already know the answer to: which funds invest in your category. What it does not tell you clearly is which of them deployed capital this month, at your stage, into something adjacent to what you are building.
Funding news answers exactly that, and it does it with a date attached. It also gives you the thing that makes a cold email land, which is a specific, checkable, recent fact about the person you are writing to.
These four feeds carry real funding coverage and returned items when we last checked them:
https://techcrunch.com/category/venture/feed/
https://news.crunchbase.com/feed/
https://tech.eu/feed/
https://www.eu-startups.com/feed/Check a feed before you build on it rather than trusting a list. Two well known startup publications returned HTTP errors on their obvious feed URLs when we tested, which is the sort of thing that turns into a workflow silently producing nothing on Monday morning. Whatever you use, let a dead feed fail without taking the run down with it.
Filter before you pay
A funding feed is not only funding. It carries opinion pieces, conference promotions, hiring posts and product news. Sending all of it to a model is the fastest way to make this expensive and pointless at once.
Structure filters it almost perfectly, because a real funding story has a recognisable shape. Require both halves:
raise word raises | raising | secures | closes |
lands | nets | bags | series A-E |
pre-seed | seed round | funding round
amount $ EUR GBP followed by a figure and a
magnitude: 12M, 8 million, 1.2bn
require BOTH, then drop obvious noise
webinar | tickets | discount | hiring | opinionRequiring the amount is what does the work. Conference ads mention Series A founders constantly, and opinion pieces mention funding without ever reporting one. A headline carrying both a raise verb and a figure is nearly always a real round.
Add a recency window and a deduplicate on the story link. The same round gets re-reported across outlets, so without the link deduplicate you will score one raise three times and draft three intros to the same fund.
Score with a reason, and never default to a pass
Ask for the facts and the judgement in one call, and require the judgement to explain itself:
company who raised
round seed, Series A, as reported
amount as reported, do not convert
sector three words or fewer
investors array of named investors, never invented
leadInvestor if one is named
fitScore 1 to 10 against your thesis
fitReason one sentence, referencing THIS roundTwo conditions must both hold before anything gets drafted, and the second one catches a bug that is easy to ship:
- Somebody must be named. A round reported without its investors scores high on paper and has nobody to email. An empty investor list is an automatic miss.
- An unparseable score is a miss, not a pass. Writing
Number(score) || 7turns a real zero into a seven and quietly passes everything the model fumbled. Parse properly, recover a digit from formats like8/10, and if nothing numeric comes back, record that and fail it.
Keep the reason sentence in the tracker alongside the score. It is how you notice the scoring has drifted after a prompt change, which you will not see from numbers alone.
The intro has to open on their deal
The entire reason this works is that the mail references the round the investor just did. Take that away and it is a cold email about you, which is what every other cold email in that inbox is.
Give the model the round, the company and your own traction, and forbid everything else. Then verify the output before saving:
if the body never names the funded company
-> flag it: this is a blast, not an intro
also reject
no subject line
under 50 words not credible
over 250 words will not be read
[PLACEHOLDER] left in the bodyCap the traction claims to numbers you supplied in the config. A model asked to be persuasive about a company it knows nothing about will produce a plausible growth figure, and sending an invented metric to an investor is a mistake you cannot walk back.
Everything lands in drafts. You get one first impression per fund, and the value here is the research and the blank page, not the clicking of send.
Who this is for, and what to change
Founders raising now
The problem. Building the investor list is a week of work that has to happen before any conversation starts, and it is stale by the time it is finished.
What to change. Run it daily during the raise and turn it off afterwards. Set the threshold high, because ten good drafts a week is a real pipeline and thirty is a spam problem.
Founders not raising yet
The problem. The right time to meet an investor is months before you need them, which is exactly when it feels pointless.
What to change. Drop the ask entirely. Keep the scoring and the tracker, and let the drafts be short notes with no request in them. The tracker becomes the list you actually raise from later.
Operators and angels
The problem. You want to see what is being funded in your space, not to email anyone.
What to change. Delete the drafting half. The scored sheet with a reason per round is the product, and it costs one model call per story.
Agencies and service businesses selling to funded startups
The problem. A company that just raised has budget and a mandate to spend it, and that window is short.
What to change. Point the thesis at the company rather than the investor, and score on whether the raise implies a need you serve. The rest of the workflow is unchanged.
The shape of it
Schedule: every weekday morning
-> Set: thesis, stage, sector, traction, threshold
-> Sheets: read who you already contacted
-> RSS x3 -> Merge -> Remove Duplicates on the link
-> Code: keep only real rounds, drop anything seen before
-> Filter -> Loop over stories
-> AI: extract the round, name the investors, score the fit
-> Code: parse safely, an unparseable score is a miss
-> IF it passes
no -> back to the loop, costs nothing more
yes -> AI: draft the intro on that specific deal
-> Code: reject a draft that never names them
-> Gmail: save as draft
-> Sheets: append to the tracker
-> Gmail: tell you what was draftedTwenty nodes, one paid credential, and a miss costs one call rather than two because the drafting step sits behind the gate. This is a free n8n template you can import.
Investor outreach questions, straight answers.
Why use funding news instead of an investor database?
Because a database tells you a fund exists and funding news tells you it is actively writing cheques this month, at what stage, into what sector. Recency is the part that matters for outreach. A fund that led a seed round three weeks ago is a different prospect from one that last deployed eighteen months ago, and most databases will show you both the same way.
Which funding feeds are worth reading?
TechCrunch's venture category, Crunchbase News, Tech.eu and EU-Startups all publish full RSS feeds with real funding coverage. Crunchbase News in particular is dense with round detail. Check any feed before you build on it: two well known startup publications returned HTTP errors on their obvious feed URLs when we last checked, so a feed that used to work is not the same as one that works now.
How do you avoid paying a model to read conference ads?
Filter on structure first. A real funding story mentions both a raise word and an amount: raises, closes, secures, lands, plus a figure with a currency and a magnitude. Require both, drop anything with obvious noise words like webinar, tickets or hiring, and drop stories older than your window. Everything that survives is worth a model call and almost nothing else is.
What should the model actually extract?
The company, the round, the amount, the sector and the named investors, plus a fit score against your thesis and one sentence explaining the score. The reason sentence is the important part. A number on its own tells you nothing about whether to trust it, and a stated reason lets you spot a scoring drift in the sheet rather than after fifty bad drafts.
What if the model returns a score you cannot parse?
Treat it as a miss, never as a pass. This is a specific bug worth guarding against: writing Number(x) || 7 turns a legitimate zero into a passing seven. Parse the number, recover a digit from formats like 8/10, and if nothing numeric comes back record that it failed to parse and let it fail the threshold. A scoring step that silently defaults to a pass will fill your drafts folder with nonsense.
Should a story with no named investor still go through?
No. A round can be reported without naming who led it, and there is nobody to email. A high fit score with an empty investor list has to fail, or you generate a personalised intro addressed to nobody. Make it an explicit condition rather than assuming the score covers it.
What makes the intro different from a generic cold email?
It opens on the specific deal that put the investor on your list. That is the whole premise, and it is also the easiest thing to lose, because a model with a thin brief will fall back on flattery about the fund. Check the draft actually names the funded company before you save it. If it does not, it is a blast with a personalised sender, and it will be read as one.
Should these emails send automatically?
No. Investor outreach is reputationally expensive and unrecoverable, and you get one first impression per fund. Create drafts and read them. The workflow removes the research and the blank page, which is most of the work, and leaves the send as a decision you make.
How do you avoid contacting the same investor twice?
Keep a Contacted tab and read it at the start of every run. Deduplicate before scoring rather than after drafting, so a repeat never costs a model call. This also matters because funding feeds re-report the same round across several outlets, so the same fund can surface three times in one morning from three different stories.
Is your investor list current, or six months old?
A community of builders shipping marketing and AI automations together.
Keep reading.
Stop retyping your call notes
Your notetaker already wrote the summary. Fire on a Gmail label so you never pay for transcription, and produce records instead of more prose.
Never walk into a call cold
One page on who you are meeting, thirty minutes before. Built from their own site and free news, with the rules that stop it inventing what it could not find.
The host's email is in the feed
Apple's catalogue needs no API key and the contact address sits in the show's RSS. Filter out shows that stopped publishing, then draft a pitch that quotes a real episode.
Build the next one with us.
A community of builders shipping marketing and AI automations together.