Roadmap

This roadmap outlines planned enhancements to the Signal Database. Timelines are estimates and subject to change based on customer feedback.

📬

Feedback welcome: Have a signal you'd like to see or a schema improvement that would help? Contact [email protected]

Quick Navigation

• Now - In progress
• Next - Coming soon
• Later - On the horizon
• Recently Shipped - Live today

Now (September–October 2026)

🔗 Unified Entity Resolution - rolling out across every pipeline

Every signal we deliver depends on one hard problem: resolving a messy real-world mention to the right company or contact. We've built a single shared resolution service - one engine, every signal source as an adapter - and are rolling it out pipeline by pipeline.

How it works: each request walks a waterfall - a persisted truth table of verified links, then cost-ordered vendor lookups, then live-page corroboration with LLM verification. Below the confidence floor the service abstains rather than guesses. Every resolution is written back, so the knowledge base compounds with every pass.

As pipelines migrate, expect measurably higher match precision - and versioned corrections to historical data where new resolution supersedes old matches (announced in the Changelog before they land).

🧭 Company Identity on Every Signal - coming soon

Every signal will say exactly who it is about - the whole company, a subsidiary, or one site such as an office, store or hotel - plus how we connected it to the company and how confident we are. Company objects also gain hq_country.

• identity.level - company, subsidiary or site, with the name and location as the source shows them
• identity.method - the company's own source, a verified ID, or a name match
• identity.confidence - high, medium or low, so you can filter before you load

Additive only: no existing field changes. See the fields and examples →

📈 Signal Volume Expansion

Major coverage increases in flight across the library:

⚡ Delivery Frequency Increases

Refresh and delivery cadence step up across the library. Each change is announced in the Changelog as it lands.

🔍 Automated Quality Auditing (Groundtruth)

Our independent audit engine asks one question of delivered data: is it true? It scores signals on fidelity (does the signal match its source?) and truth (does the claim match reality?) separately.

• Live today across eight signal families (news, podcasts, Hacker News, new-business formation, federal contracts, SEC filings, earnings transcripts)
• First full-library audit pass complete
• Next: periodic automated re-audits with alerting, expanding coverage to all 34 signal types

Next (Q4 2026)

🆕 New Data Sources

📦 Delivery APIs for Flat-File Customers

New Signal API endpoints so bulk-delivery customers can track their data programmatically:

• Upcoming delivery schedule - see what's landing and when
• Record counts - per delivery, per signal type
• Delivery health - extends the manifest files already live in beta

🆔 Canonical Entity IDs

Feeds move from embedded company objects to stable entity IDs - firmographics stay fresh via the linked record, and subsidiary/hierarchy rollups unlock at that stage.

{
  "company": {
    "autobound_id": "ab_company_12345",
    "domain": "stripe.com",
    "linkedin_url": "linkedin.com/company/stripe"
  }
}

🤖 ML-Accelerated Resolution

Bringing a trained model into the resolution path - candidate ranking and match scoring learned from our accumulated truth table - for faster, cheaper, and more accurate entity matching.

🎯 Predictive Intent - Outcome-Trained

Preparing our intent engine's next training generation on large-scale, real-world sales outcomes - billions of closed-won and lost-opportunity records - alongside the new Reddit corpus and the expanded patent, Twitter/X, and news corpora above.

📊 Monitoring-Pool Management API

Completing the monitoring API: list, read, and remove endpoints alongside the existing add endpoint, with automated pool→collection sync.

Later (2027)

🧬 Unified Schema Across All Signal Types

One schema spanning the entire library. The subtype standardization program below is the visible first phase; full unification is now formally on the roadmap.

Current state: 700+ subtypes across 35+ signal types, with naming that varies by source. Target: consistent category_specificEvent naming, with legacy subtypes supported via aliases for backward compatibility.

📱 New Social Sources (Exploring)

• Instagram (business accounts) - under evaluation
• TikTok - under evaluation

🔄 Cross-Signal Event Blending

The same business event, recognized across sources: executive-change signals from earnings calls, SEC filings, and Reddit joined with hiring-trend data; funding events deduplicated across news and filings, linked via a related_signals array.

Recently Shipped

• Reddit on a new engine - the Signal API's Reddit type now carries 15 typed subtypes (customer feedback, buying intent, outages, pricing and product changes, and more), refreshed weekly. Bucket delivery is unchanged
• Podcast signals GA - company + contact podcast signals, out of beta
• Patent coverage expansion - September delivery: 47,313 records, +47% month-over-month
• Weekly SEC cadence - 10-K/10-Q/8-K/20-F/6-K and earnings transcripts on weekly delivery; news on daily delivery
• Manifest files for S3/GCS drops (beta) - per-drop manifests enabling event-driven pipelines
• Delivery changelog - every delivery logged publicly with record counts

See the Changelog for the full history.

Stay Updated

Track all schema changes, new signals, and API updates in our Changelog.

Questions?

• Sales - [email protected] for roadmap priorities
• Support - [email protected] for technical questions