Skip to main content

For GTM and RevOps teams

Score, route, and trigger on signals, not guesses

You own the scoring model, the routing rules, and the sequences. Autobound gives you 700+ signals across 35 feeds as clean fields you can weight, sync, and fire on. No new UI for reps to learn.

What you build

From signal to action

Four ways RevOps activates the data. Each demo runs on real records; the rules, the joins, and the rails are yours to configure.

01

Signal fires, sequence starts

Define the trigger once. When a matching record lands, the owner gets a task, the right contacts get enrolled, and the rep gets the quote and the source. The rule reads in plain words because the fields do.

Delivered via

GCS pushREST API
Workflow · signal to sequence
  1. 1
  2. 2
  3. 3
  4. 4
01Signal lands
Rockwell AutomationROK
earnings-transcript / revenueAcceleration

Q3 organic sales up 10%, EPS up over 20%

relevance 0.9sentiment positiveconfidence high
02Rule matches

When a record arrives where

  • event is revenueAcceleration
  • relevance is 0.9 or higher
  • account has an owner in the CRM
03Actions run
  • Task for the account owner, with the headline
  • Enroll finance contacts at rockwellautomation.com in the post-earnings sequence
  • Post to #signals with the source link
04Evidence attached
“We delivered a strong quarter with double-digit year-over-year growth in sales and earnings exceeding our expectations.”
Blake Moret, Chairman and CEO · earnings call, Aug 4, 2026

The record is real (content/samples/earnings-transcripts.csv). The rule, the sequence, and the Slack channel are yours to configure in your CRM or workflow tool.

02

Rules you can trust, because every record shows its work

A rule is only as good as the field it reads. Each signal carries the verbatim quote it was extracted from, the event we typed from it, and an evidence stage from speculative to confirmed. Set the floor at reported and predictions never reach a sequence.

Delivered via

GCS pushREST API
How we extract a signal
1 · sourcer/selfhosted
“Business Edition free version (for up to three nodes) will be the only free option going forward for 3.x.”
verbatim, stored as data.evidence
2 · event
Portainer
pricingChange

Portainer 3 drops the free Community Edition.

company.domain portainer.io
3 · evidence stage
confirmed
  1. speculativeA prediction. Nothing has happened yet.
  2. rumoredSomeone heard something. Unverified.
  3. reportedA secondhand claim that it happened.
  4. confirmedA first-hand account or an official notice.

Set the floor at reported and predictions never reach a sequence. Prominence: core.

Real Reddit signals from the delivered weekly batches, as shown in the Reddit launch post. Quotes are verbatim from data.evidence. Every feed with text evidence carries the same three parts.

03

Enrichment into CRM and sequencer

Push signals onto the account and contact objects your team already works in. Companies match on domain or LinkedIn URL, contacts on email or LinkedIn URL, so the match is a join, not a project.

Delivered via

GCS pushREST API
Match keys
your CRM row
domaincytiva.com
email•••••@cytiva.com
LinkedIn URLlinkedin.com/in/par-hermansson-ba600911
matched on domainmatched on emailmatched on LinkedIn URL
Autobound record · workMilestone
company.domaincytiva.com
contact.email•••••@cytiva.com
contact.linkedin_urllinkedin.com/in/par-hermansson-ba600911
Cytivacompany · matched on domain or LinkedIn URL
Par Hermansson · Sr Director - IT Business Enablement, Commercialcontact · matched on email or LinkedIn URL

Real record from content/samples/work-milestones.csv; the email is masked. Every company record carries the domain, and contact records add the LinkedIn URL and email where available. Keep signal_id for deduplication.

04

Into the warehouse for scoring, into the CRM for routing

One signal is a hint. Three in the same quarter is a buying window. Land every feed in your warehouse by GCS push, weight funding, hiring, and leadership changes per account, and push the ranked list to the CRM before reps log in.

Delivered via

GCS pushFlat file
Ingestion · 35 feeds to your stack

Autobound feeds

  • News
  • Hiring velocity
  • Earnings transcripts
  • Work milestones
  • Reddit mentions

Delivery

  • GCS pushdaily JSONL, Parquet, CSV
  • REST API< 200 ms
  • Generate Insights APIper contact
  • MCP serveragent-native

Your stack

  • Your warehouseBigQuery, Snowflake
  • Your productfilters, scores, cards
  • Your CRMtimeline, tasks
  • Your agentsClaude, Cursor

Same records on every rail. Pick the one that fits your architecture; most platforms start with GCS push for the backfill and the REST API for record-level refresh.

How it arrives

Built for how your stack already works

GCS push for the daily scoring run. REST API for record-level refresh. Flat file for the backfill. Generate Insights API for the last line.

GCS push

Daily to weekly

Scheduled exports to your Google Cloud Storage bucket in JSONL, Parquet, or CSV.

gs://autobound-<signal>/<date>/output.jsonl

REST API

JSON over HTTPS

Real-time search, enrich, and filter across every signal type.

$ curl https://signals.autobound.ai/v1/companies/search

Flat file

Snapshot

Bulk exports for evaluation, backfills, and warehouse loads.

output.csv · output.jsonl · output.parquet

Generate Insights API

Per contact

Signals distilled into ready-to-use insights for personalized outreach.

email or LinkedIn URL → ranked insights

FAQ

Common questions

Which object does a signal attach to?

Every record has an association field: company or contact. Company signals carry the domain. Contact signals add the LinkedIn URL and email where available, plus the resolved employer.

How fresh is the data?

File pushes run daily to weekly, set per feed. The REST API returns current records in under 200 ms, so a scheduled search can run as often as your sequencer needs.

Do we need engineering to set this up?

GCS push needs someone to load files into your warehouse or CRM. The REST API works from any tool that makes an HTTP request, including Clay columns and workflow builders.

Can we start with one feed?

Yes. Start with the feed behind your first trigger, then add feeds as you add rules. Pricing follows the feeds you license, the delivery method, the refresh frequency, and the volume.

How do we avoid double-firing on the same event?

Every record carries a signal_id and a detected_at timestamp. Store the id with the task or enrollment it created, and skip records you have already acted on.

Bring your scoring model. We bring the inputs.

Send us your ICP and your first three triggers. We map them to feeds and send sample files for your accounts.