Taste
Taste is a list of refusals, enforced in code.
A model writing one email cannot see the ninety-nine before it. So the decisions that need that view are made before it writes.
The dossier
Research in. Empty data out.
One record per person, built before a word is drafted.
research.person$0.050
the person record
live profile
when LinkedIn is connected
enrich.company$0.005
cached 30 days
their product
two of their pages, then research
- currentRole
- title · company · since
- organizations
- up to 8, current first
- company
- domain · industry · stage
- status
- complete | partial
4,000 characters, never trimming the current role. Cached 90 days, shared across workspaces.
location: "United States"
Refused: a place, not a person
{"status":"failed"}
Refused: a failed lookup
"… is a role based email address"
Refused: a shared inbox
"Nothing Here, Yet"
Refused: an empty profile page
The prompt
Built in a fixed order.
Dashed blocks appear only when there is something true to put in them.
An observation about them earns nothing: they know their own company.
The angle
Picked before the model writes.
Left to choose, the model reached for the most writable angle.
Before, on a live ledger
Now, a separate call at temperature zero
- angle A
- angle B
- angle Cfits
- angle D
- Judge fit from the PROSPECT block only.
- When none clearly fits, choose the most general angle, never a specific one that would be false for this reader.
- Ignore how well-written an angle is. Writability is not fit.
“A fabricated citation is worse than no evidence at all.”angle-synthesis.md, checked in code
- dossierkept
- replies:2kept
- github:livekept
- https://… not in the evidencedropped
The linter
Sounds like a template? Held.
The product’s own linter, run on two letters to the demo’s fictional cast. Each is judged against the offer above it. The held one is from the demo. The clear one was written here, to the offer-first rules.
OFFER find the job behind each queue backpressure spike while your team grows after the round
Northwind + background job traces
Hi Priya, I came across Northwind's engineering blog and wanted to reach out — the post on queue backpressure was a great read. Tracepoint helps teams leverage per-job tracing, so incidents get faster, cheaper, and clearer. I'd love to connect if you are open to it. Best, Mira
- opens-on-observation
- em-dash
- banned-opener:I-came-across
- banned-cta:love-to-chat
- ai-vocab
- rule-of-three
OFFER turn tracing back on across your worker fleet without the per-host bill
tracing back on
Hey Grace, I can get tracing back on across your worker fleet without the per-host bill that turned it off. Tracepoint takes one import per worker and bills per job traced, so the size of the fleet stops setting the price. Worth doing before the next incident? Mira
CLEAR291
sends of "worth a 10-min back and forth"
120 of 312
second touches shipped a meeting ask dressed as a small one
55%
of 411 follow-ups opened "still curious"
Each count is quoted from the comment beside the rule it caused.
What this does not do
Where the rules stop.
Nothing checks the finished draft line by line
The angle's citations and the offer's numbers are checked against the research. The rest of the email is not.
Nothing it learns applies without your approval
What it learns from your edits and sends waits on the queue as a proposal, on email and LinkedIn.
Nothing opens on their posts
Posts feed the research behind the offer. The message opens on the offer, not on what they wrote.
No reply-rate lift is measured
The tool counts replies per angle and labels the counts a hypothesis.
Every rule is a file in the repo. Fork them and argue.
