Intent, without
the black box
Today's intent data is a topic score with no reason behind it. We're building the engine that reads 1B+ real buying signals and learns from actual closed-won deals - so it can tell you who's in market, and the exact signals that prove it.
Contribute your outcomes data. Get a map of who's about to buy in your market.
The signal graph
1B+ signals
Hiring, funding, tech adoption, leadership moves, launches, and 700+ more.
Your contribution
Closed-won deals
The one thing black-box vendors don't have: what actually converted.
The output
Who's buying now
In-market companies, ranked, with the exact signals that prove it.
1B+
buying signals
700+
signal types
50M+
companies
250M+
contacts
The problem
Intent data is broken
Your reps have learned to ignore it, and they're right to.
It's a black box
A topic "spiked." You have no idea why, or whether it means anything at all.
It's one signal
Web visits and bid-stream, stripped of everything else a buyer is actually doing.
It never learned
It has no memory of what closed. It was never trained on deals that actually happened.
What we're building
A signal-native intent engine
We already track 1B+ buying signals across 700+ signal types - hiring, tech adoption, funding, leadership moves, product launches, and more - on 50M+ companies and 250M+ contacts.
The missing piece is outcomes: which of those signal patterns actually preceded a purchase. Layer real closed-won and closed-lost data on top of a billion signals, and you can finally answer who is in market, at what stage, and which specific signals prove it.
No black box. Every prediction comes with the breadcrumbs.
The mechanism
From closed-won wins to who's buying now
Read the patterns across won deals, layer billions of signals on top, and find the companies exhibiting the same fingerprint today. Here's how that looks.
Auto-playing. Click any step to explore.
Closed-won cohort
The model looks back at what winners had in common in the weeks before they bought.
won deals analyzed
Aggregate as few as 10 co-op deals plus a handful of your own — the more outcomes, the sharper the pattern.
Won · $84k ACV
Mid-market · RevTech
Won · $120k ACV
Enterprise · Logistics
Won · $46k ACV
SMB · Fintech
Won · $210k ACV
Enterprise · Analytics
… and 0 more across the cohort
Illustrative example. Company names are fictional; signal types are real sources in the Autobound graph.
Already validated
We didn't just theorize this. We backtested it.
Before asking anyone to contribute, we ran the thesis end to end. We trained gradient-boosted models on 14 months of our signal history - 82.8M signals across 31 types - against real technology-adoption outcomes, then measured how well the signals predicted who would adopt within 90 days. They predicted it exceptionally well.
4–7.6×
more buyers in the top decile than random targeting
0.97
peak model accuracy (AUC)
82.8M
signals across 31 types, 14 months of history
90-day
forward prediction window
Predicting adoption, by category
Five category-specific models. Every one beat the generic baseline by a wide margin.
Top-decile lift: the top 10% of predictions contained 4–7.6× more actual adopters than you'd hit by targeting at random.
What the model keyed on
The strongest predictors, in order.
- 1
Tech-stack freshness
how recently they last adopted anything new
- 2
Adoption velocity
new tools added in the last 90–180 days
- 3
Total tech footprint
how much they already run
- 4
Hiring signals
roles opening in the relevant function
- 5
Signal recency
how fresh the underlying evidence is
Gradient-boosted trees - auditable, not a black box. Every score traces back to the signals behind it.
That was one label: technology adoption. The co-op brings the highest-value label of all - what actually closed - and extends the same method to every buying decision, not just tech.
The data asset
Everyone has signals.
We have the answer key.
Intent vendors sit on oceans of signals but no labels - they never learn which patterns actually ended in a purchase. The co-op is a privacy-safe stream of real closed-won and closed-lost outcomes: the training labels that turn signals into predictions. It compounds with every contributor, and it can't be back-filled or scraped.
Supervised learning needs labels. Outcomes are the labels - and only contributors have them.
↻The moat compounds
↺ which attracts more contributors - and a latecomer starts from zero.
Opt-in, not scrapable
Outcome data is contributed under NDA. It can't be crawled, or bought off the shelf.
Compounds over time
Every contributor sharpens every prediction. The lead compounds and can't be back-filled.
Privacy-safe by design
We profile buyer patterns, not customer lists. That's exactly why contributors say yes.
Layer this on an identity graph and every score resolves to the company - and the person - to reach. Signals + outcomes + identity = person-level intent that actually converts.
Why contribute
What you get for contributing
Your in-market map
We analyze your closed-won deals, learn the signal pattern behind your best wins, and hand you a list of ICP-fit companies showing that same pattern right now.
Early access to the engine
Design-partner access as we build it, free while it's in development. Shape it before anyone else can buy it.
A voice in the roadmap
Tell us which signals matter for your market and how this plugs into your stack. Design partners set the priorities.
Zero lift to start
Share an anonymized sample or connect your CRM. We run the analysis. Nothing moves without a mutual NDA.
Apply in two minutes
Tell us what data you have and what you sell. We'll review it and follow up with next steps, including a mutual NDA before any data changes hands.
Apply to contributeSOC 2 Type II Certified
Independently audited security controls. Enterprise-grade data protection.
No data moves without a mutual NDA. See our Privacy Policy.
Who can join
Three ways to get involved
If you have deal outcomes, you can help build this - whatever your size.
For sales teams
Bring your CRM
A B2B company or sales team with closed-won and closed-lost history. That deal outcome data is exactly what the engine learns from.
For platforms
Aggregate at scale
A platform or data partner who can supply anonymized closed-won / closed-lost records across many companies. Let's talk co-development.
For individuals
Just have deals to share
Even an individual seller with a book of closed deals moves the needle. If you have outcomes, you can contribute.
Getting started
How partnering works
Apply
Tell us what data you have and what you sell. Takes two minutes.
Share a sample
Anonymized closed-won / closed-lost, or a CRM connection. Mutual NDA first, always.
We analyze
We map the signal patterns behind your wins against 1B+ live buying signals.
You get your map
See which companies are in market now, plus early access to the engine.
FAQ
Questions, answered
No. Start with an anonymized sample. We profile buyer patterns, not your customer list, and we're not reselling your leads.
Nothing to contribute. Design partners get early access to the intent engine free while we build it.
We learn which signal patterns precede a purchase, so we can spot lookalike companies that are in market now. Your data trains the model. We don't hand your leads to anyone.
Still useful. Apply and we'll tell you whether it's enough to run an analysis on your market.
We formalize that before any data changes hands. For aggregators and platforms, we're open to co-development and commercial terms that protect your data and your business.
Help build the intent engine
that actually shows its work
Contribute your closed-deal data, get a map of who's about to buy in your market, and help define the next generation of intent. No one else is doing this yet.
Apply to contributeTwo minutes to apply. Mutual NDA before any data is shared.