Havilo
The Matchmaking Engine

Ask the room. Get the receipts.

A question becomes a plan. A plan becomes a person. A person arrives with the paragraph that earned them.

Example
Havilo
Query

I need a mentor who can actually help Aurora's CTO. Someone who has scaled an AgTech engineering org from a handful of people to something close to five hundred, ideally with operating time in the Southern Hemisphere — and, honestly the hardest part, patient enough to coach a first-time technical founder without rolling their eyes.

Reading
conceptualAgTech engineeringstrict~5× order-of-magnitude scaleinferenceSouthern Hemisphere signalbehavioralPatience with first-time founders
I.Ask

Tell us in your own words.

A line. A paragraph. A rambling page with three asks buried in it. The system reads the whole thing the way a careful chief of staff would — it holds the background you forgot to restate and the constraint you meant to mention.

The range the system accepts
  • three words

    “find me customers”

    The asker is identified. Context, ICP, and calendar supply the rest.

  • a sentence

    “A mentor who's scaled an AgTech firm from single digits to five hundred engineers.”

    Strict numbers, conceptual domain, one implied geography.

  • a paragraph

    “For Aurora's Series B — post-510(k), targeting $18M — ten funds with a real reimbursement pattern. Cut anyone the partner team already emailed. Draft the intros in Sarah's voice.”

    Exclusion, inference, composition. All three handled in one plan.

  • a whole essay

    “Here's what the last six months have looked like. We shipped the clinical validation, lost a CMO, and the new fundraising timeline is …”

    Narrative context absorbed; explicit ask inferred from the closing lines.

How a phrase is readone word · three asymmetries
phrase
“raised their Series B”
implies a company
⇌
both ways
implies a date ± 2 weeks
Funding events carry temporal precision.
a company raised
→←
one way only
its founder is current
The founder may have left before the close.
the round was led by X
⇢
most of the time
X sits on the board
Two out of three leads take the seat.
The engine carries thousands of these. It knows where implication runs, where it stops, and where it only runs most of the time — which is where the best judgment lives.
II.Reason

Meaning before search.

Every word in your question is pointing somewhere before any search begins. Which phrase is a hard filter and which is a soft signal. Which attribute crosses from a person to a company, and which does not travel back the other way. The reading is the work. Everything after it is retrieval.

III.Resolve

The connections nobody indexed.

Evidence lives in different corners of the network and rarely agrees. A paragraph from 2019 contradicts a deck from this quarter. Two founders appear in the acknowledgments of a paper neither of them posted. The system reads for the disagreements, the sideways distances, and the edges no schema anticipated.

Running every path at once is the only way to let the weak edges survive.

An edge nobody taggedsurfaced · 2016 paper
E.B.
A.V.
shared advisor
Eduardo Braga · ex-CTO AgroSmart · São Paulo
Aurora Bio's CTO · first-time founder · Milan → SF
the edge

Both studied under the same advisor at Politecnico di Milano. Neither of them lists it in public. The paper they're acknowledged in was published in 2016.

The graph did not have this edge. The system read for it, and found it, because every path gets to contribute — including the ones that start from writing rather than from metadata.
Briefing · for Sarah
Oct 16 · 08:41
MW
Top match · 94
Dr. Marcus Wei
CIO · Cedars-Sinai

Cedars-Sinai CIO since 2022; previously led the diagnostics platform rollout at Mayo. Published a POV last month on 510(k) device integration that reads like your product roadmap, footnoted with the exact reimbursement pathway you're walking.

James Whitfield sat next to him at the January JPM panel. James still owes you the intro; Marcus would take it.

you said

“Cedars is the white whale, and reimbursement is what'll kill us.” Marcus spent his first year at Mayo fighting exactly that — he'll know what you're walking into before you finish the first slide.

Diagnostics rollout · MayoWarm via J. WhitfieldWhite-whale accountPOV aligned · last month
three sentences · six sources · one paragraph that mattersdraft intro queued
IV.Explain

Composed for you.

A name with a paragraph. Or a full briefing. Or a six-page memo — whichever the question asked for. Every line is written knowing what the reader already heard, what they already pushed back on, and what would actually change their mind on Thursday.

The answer that cites your October 14 call is the answer that gets the meeting.

V.Compound

Every call sharpens the next match.

Matchmaking is better next week than it is today. The Interviewer sits on the first two minutes of every meeting and writes down what wasn't anywhere else — straight back into the next query's context. The more the network talks, the more precisely it introduces itself.

See EnrichmentSee the full flywheel
The loopfive stations · endless
  1. 01
    Match
    Sarah ↔ Marcus queued
  2. 02
    Meet
    First two minutes · Interviewer listens
  3. 03
    Extract
    "Cedars hates SaaS"
  4. 04
    Remember
    Signal written to Marcus dossier
  5. 05
    Next match
    Accounts like Cedars · ranked differently
The next plan already runs on what the last call said

A network you can ask.

Book a demo. We'll show you what the system thinks of your cohort — in the voice of someone who read every page.

Read the Alchemist case study →
Havilo

The relationship intelligence
platform
for modern institutions.

Platform

  • Matchmaking
  • Enrichment
  • Data Ingestion
  • Meeting Intelligence
  • Interviewer
  • Proactive Intelligence
  • Workflow Automation

Solutions

  • Accelerators
  • University Programs

Company

  • About
  • Contact
© 2026 Havilo·Privacy·Terms·