Skip to main content
Feature ReleaseAnnouncements

What's New at Autobound Signal API — August 2026

August was our biggest data month of the year: podcast intelligence at full scale, subtype filtering on every entity endpoint (it cuts your credit spend), SEO signals rebuilt around AI-search visibility, and a portal sandbox that researches the open web. Here's what shipped.

·4 min read

Key Topics Covered

  • Podcast Intelligence at Scale
  • Subtype Filtering, Now on Every Entity Endpoint
  • SEO Traffic Signals, Rebuilt Around AI Search
  • The Sandbox Became a Research Agent

Article Content

August was our biggest data month of the year. We took podcast intelligence to full scale, made subtype filtering work everywhere (and cheaper), rebuilt SEO traffic signals around AI-search visibility, and turned the portal sandbox into a real research agent. Here's what shipped.

Podcast Intelligence at Scale

The rebuilt podcast signal types are live: 58,927 signals across ~27,700 companies, extracted from ~39K transcribed B2B podcast episodes, refreshed weekly.

  • Appearances - who joined which episode, with topics, verbatim talking points, and a snapshot of what they run
  • Claims - one row per distinct on-air business claim (65+ subtypes shared with our SEC/earnings vocabulary: aiInvestment, techAdoption, painPointDisclosed, ...), each backed by verbatim quotes
  • Sponsorships - actual ad reads, mapped to the sponsoring company

Transcript keyword search now works on both search endpoints via content_filters - find every company whose exec said "digital transformation" on a podcast this month. Full field reference in the podcast docs.

Subtype Filtering, Now on Every Entity Endpoint

signal_subtypes filtering was search-only. As of this week it works on companies/enrich and contacts/enrich too - ask for an account's full signal profile and keep only the subtypes you care about, server-side.

The part your finance team will like: enrich bills per returned signal, so the filter directly cuts your credit spend. A filter that matches nothing is free. Discovery is free too: GET /v1/signals/types returns the full per-type subtype vocabulary. Unknown values never error (you get a warning, not a 400), so one bad value can't break a batch job. The MCP server's enrich tools accept the same parameters.

New docs: the Subtype Filtering guide covers discovery, semantics, and billing impact with copy-paste examples.

SEO Traffic Signals, Rebuilt Around AI Search

The seo-traffic signal type relaunched on a far richer payload - 26.6M signals (3x prior volume), current through August. Each signal now carries 46 fields, including:

  • AI traffic - visits arriving from AI platforms (ChatGPT, Perplexity, ...), with the top platforms and even the top prompts driving them. Know which of your prospects are winning - or invisible - in AI search. That's a conversation opener nobody else has.
  • Digital health score - one number for how a company's web presence is trending, with a human-readable explanation
  • Brand strength, engagement quality, referral strength, geo concentration - each scored and explained
  • Global / country / category rank plus month-over-month and 3-month momentum

Signal Data API

Turn these insights into pipeline

700+ real-time buying signals from 35+ sources. Know exactly when prospects are ready to buy.

The Sandbox Became a Research Agent

The portal's AI sandbox can now research beyond our own data:

  • Web search + live browsing - it searches Google, reads the winners as clean markdown, and escalates automatically for bot-protected pages. Ask "how would signals help [prospect]?" and it reads their site before answering.
  • Run controls - a real Stop button, per-tool latency chips, and a stall detector. You can always bail; silence is always explained.
  • Readable results - searches render as dense tables (name / title / company / signals), not 80KB JSON dumps.
  • 48 starter prompts that rotate based on what users actually click.

Data Quality, Across the Board

  • Consistent industries: company.industry is normalized across all signal types
  • ProductHunt signals now ship with firmographics attached - company LinkedIn, industries, employee count, and revenue range, no second enrichment call needed
  • Freshness numbers you can trust: last_detected_at now excludes future-dated vendor rows
  • Faster content search: keyword queries return the newest matches quickly instead of scanning to exhaustion

What's Next

More transcript sources, deeper AI-traffic coverage, and subtype vocabularies for more signal types. If there's a signal you wish existed, tell us - the fastest-shipped features on this list started as customer asks.

Signal Data API

Turn these insights into pipeline

700+ real-time buying signals from 35+ sources. Know exactly when prospects are ready to buy.

Get Started

Everything above is live today. Grab a key at signalapi.autobound.ai, read the docs, follow the changelog, or talk to us. And if you missed it: what shipped in June.