All Updates
TL;DR
ICP profiles now include disqualifiers, hooks, and funding signals with automatic consistency checks and industry taxonomy matching. Onboarding lets you review and edit your ICP and search config before your first scrape. Deep Research shows an expansion saturation curve and recommends when to stop expanding. Hot lists flag companies outside your ICP size range and contacts needing enrichment. Connect Claude, Cursor, or any MCP client with one click from Integrations. New brand fonts and landing page atmosphere. Scrape jobs auto-retry on timeout with watchdog healing. OAuth hardened with redirect validation and refresh-token reuse detection.
ICP Intelligence Engine
- Disqualifiers and hooks — ICP generation now produces disqualifiers (reasons to reject a company) and hooks (conversation openers that resonate). Both are editable and feed into Deep Research and pipeline matching for tighter targeting.
- Consistency checks — AI-generated ICPs are automatically checked for contradictions — mismatched size ranges, location gaps, overlapping criteria — before saving. Warnings surface so you can fix issues upfront.
- Industry taxonomy — ICP matching now uses a canonical industry list with token-level matching, replacing loose string comparisons. Fewer false positives on companies in unrelated industries.
- Employee size enforcement — Numeric employee bounds from your ICP are now hard filters. Companies above your stated maximum or below your minimum are excluded from research results and pipeline matching.
- Inline editing — ICP detail pages now support click-to-edit on every field. No more switching between view and edit modes. Textareas auto-size to content.
- Fuzzy duplicate detection — Creating an ICP with a name similar to an existing one triggers a nudge, helping you avoid accidental duplicates.
Onboarding Improvements
- Editable ICP review — After AI generates your ICP during onboarding, you can now review and edit disqualifiers, hooks, and target titles before saving. No more accepting defaults blindly.
- Editable Search Config — Onboarding now shows your auto-generated search config with editable exclude keywords and target titles. Tweak before your first scrape starts.
- Auto-sizing textareas — All multi-line fields during onboarding expand to fit their content, eliminating inner scrollbars.
Deep Research
- Expansion saturation curve — After each expansion, Deep Research now shows how many new companies were discovered versus previous rounds. A visual curve helps you see when you’re hitting diminishing returns.
- Soft stop recommendation — When expansions start returning mostly duplicates, a recommendation appears suggesting you stop expanding and focus on the companies already found.
- Disqualifier steering — Your ICP disqualifiers are now injected into research queries, so AI actively avoids companies that match your rejection criteria.
- Broad-ICP warning — If your ICP criteria are too wide (e.g. all industries, all sizes), a warning explains why results may be unfocused and suggests tightening.
Hot List Improvements
- ICP size exclusion — Companies outside your ICP’s employee size range are now flagged with “Out of ICP: size” in their notes and excluded from top-tier rankings.
- Enrichment flags — Members without enriched contacts are flagged with “Needs enrichment” so you can prioritise Find Contacts on the right companies.
- Smarter tiering — Disqualified companies no longer receive S or A tier rankings. Tiering now respects both size fit and enrichment status.
AI Connector
- One-click connection — Connect Claude, Cursor, or any MCP-compatible AI assistant from the Integrations page. One click generates your config snippet — paste it into your client and you’re connected.
- Connected AI Apps — New section on the Integrations page shows all AI assistants currently connected to your account, with the ability to disconnect any of them.
- Connector guide — Step-by-step walkthrough with screenshots for connecting Claude Desktop, Claude Code, and other MCP clients.
Visual Refresh
- New brand typography — Headings now use Bricolage Grotesque and body text uses Inter Tight, replacing the previous font stack. Sharper, more distinctive across all pages.
- Landing page atmosphere — Hero section updated with a two-tone radial glow, engineered dot grid, and entrance animations for a more polished first impression.
- Dark-mode card elevation — Cards in dark mode now use layered shadows and subtle border tints for better depth perception.
Scrape Reliability
- Auto-retry on timeout — Scrape jobs that hit a soft timeout now automatically retry once instead of failing permanently.
- Watchdog healing — Stale scrape jobs that stop making progress are detected and healed automatically, preventing silent stalls in your pipeline.
- Fan-out timeout limits — Concurrent scrape operations now have per-task timeouts, so one slow source can’t block your entire scrape run.
Company Data Quality
- Smarter name casing — Company names are now only reformatted when they arrive in ALL CAPS. Mixed-case and lowercase names are trusted as-is, preventing over-correction.
- Fuzzy duplicate nudge — Creating a company or business context with a name similar to an existing one shows a nudge before saving.
Security & Infrastructure
- OAuth security hardening — Redirect URI validation now enforces strict matching. Refresh tokens include reuse detection to prevent replay attacks.
- Secret scanning — Pre-commit hook automatically scans for leaked secrets before any code is committed.
- IndexNow fix — Search engine index submissions were silently failing due to an authentication issue on the key file. Fixed — Bing and Yandex now receive page updates correctly.
- Password change fix — Resolved an issue that could cause an error when changing your password in certain conditions.
- Sitemap cleanup — Thin tag pages with minimal content removed from the sitemap to improve search engine crawl quality.