A decade of relationships your AI can finally read.
The proprietary part is the history: who you backed early, which LPs return the call, who introduced whom. Models turn over every quarter. That record should not live inside one of them.
The job to be done
A page for the pressure behind the query.
Start with the work that falls through the cracks, then decide whether an external relationship memory earns a place in your day.
Wend, measured
92
people from one WhatsApp read, before you grant a single permission.
0.63s
to read 69,000 messages and find the people in them. No model call.
2.4s
for one extraction on the Mac itself, at no cost.
Measured on a real Mac running macOS 26.4.1. On-device extraction needs Apple silicon with Apple Intelligence turned on.
How Wend helps investors
The first hour imports the decade
Upload your LinkedIn archive and years of connections land in one pass. WhatsApp opens with no permission prompt. Add your own research provider account and roles are re-checked on a rolling cadence, every change arriving as something you approve.
Questions only a real graph can answer
Who do I know in venture capital. How am I connected to her. Which founders have I not spoken to since spring. A typed graph answers by joining rows, so the list is complete rather than whatever a model recalled.
A day of pitches becomes memory
Ambient Mode reads the window you work in: the names in the Zoom call, the founder profile you open after, the thread where the intro happened. It reads the screen, never your microphone, and the raw text stays on your Mac.
Every answer traces back
Facts carry the email, page or message they came from, and you can open it. When two sources disagree about a title, both are shown. That is what makes a fact repeatable in a partner meeting.
Deal flow stays yours
Never sold, and no model provider in our stack may train on it. The raw material stays on your Mac; the graph syncs so your agent answers while the laptop is shut. Export the whole thing whenever you want.
With Wend, you can
Ask which founders you promised to revisit this quarter, and get the promise with its source
Trace the warm path into a portfolio company's next customer
Pull every conversation with an LP before the next raise
Keep the whole network readable from Claude and ChatGPT at the same time
Change your default assistant next year and lose nothing
FAQ
Is this a fund CRM?
No. Wend is single-user by design. If a partnership needs a shared system of record with deal stages, buy software built for that. Wend is what one investor uses so their own assistant can answer questions about their own network.
Other relationship tools connect to my AI as well. What is actually different here?
Agent access is common now. What varies is what a write may do. In most of them a tool call commits, so a misread name enters your record looking exactly like something you typed. In Wend it waits for your approval.
What can my assistant actually do with it?
Read the graph, answer questions about it, trace connection paths, propose new facts from anything you hand it, draft email, make scheduling links and prepare intros. Every write is a proposal you approve, and nothing sends itself.
How is my deal flow protected?
The raw material stays on your Mac: message text, screen captures, imported archives, and the bulk import sends none of them to a model. The graph syncs so your agents answer around the clock. Nothing is sold and nothing trains a model.
What does it cost?
The memory is free forever, with no card and no cap on people. Wendy is $29 a month or $290 a year, one flat price however large your network grows. Your graph is never deleted for non-payment.
Wend for other people like you
Relationship memory built for investors.
It installs on your Mac, builds the graph out of what is already there, and answers from the AI you already use. Every fact keeps its source. $29 a month, flat at any network size.
Access is invite only right now.