For outbound founders, SDRs, and agencies
Your last cold email opened with
“I noticed you’re using HubSpot.”
That’s why your reply rate is 1.2%. Clawback shows what they actually shipped, said, or argued about in the last 30 days — with source URLs. Not their LinkedIn title.
Free in beta · No card · 90-second briefs · Built for Claude Code first
Clawback Brief · live sample
OpenAIlast 7 days
API pricing will be $5 per 1M input, $30 per 1M output, with a 1M context window.
Remember, you'll need fewer tokens per task than 5.4 thanks to the new context handling.
x.com/sama/status/2047379036419014928
GPT-5.5 supports a 1M token context window, image input, structured outputs
For gpt-4o-mini and gpt-4.1-mini with the non-preview web search tool, search content tokens are billed as a fixed block of 8,000 input tokens per call.
developers.openai.com/api/docs/changelog
OpenAI releases GPT-5.5 and GPT-5.5 Pro in the API
GPT-5.2 Pro is $168 per million output tokens. GPT-5.2 output tokens cost $14/million, delivering more capability per dollar.
news.ycombinator.com/item?id=42183...
Altman apologizes after OpenAI failed to alert police before Tumbler Ridge incident
"We're deeply sorry that you found out about our gross negligence." OpenAI responding to safety pressure as the GPT-5.5 launch lands.
reddit.com/r/news/comments/1suz24j/...
Imagine running this on your worst-stuck account — you'd open the email with the @sama tweet from 3 days ago.
Run a brief on my account →The problem
Most "personalized" outreach is fake personalization.
A LinkedIn title, company size, and an old funding announcement are not context. They’re demographics dressed up as research.
Your reps, agents, and content workflows need to know: what changed recently, who’s talking about it, whether the signal is real or weak, what claims are stale, and what angle is safe to use.
Generic AI tools are good at writing. They’re bad at knowing whether the premise is current.
A specific thing they posted, said, or shipped in the last 30 days — with a source URL.
The moat
Google aggregates editors. Clawback searches people.
Reddit upvotes. X likes. YouTube transcripts. TikTok engagement. Polymarket odds backed by real money. That’s millions of people voting with their attention and their wallets every day. Clawback searches all of it in parallel and scores it by what real people actually engage with.
What Clawback reads, ranked by engagement
Upvotes + top comments
Likes + reposts + replies
Transcripts + views
Captions + engagement
Points + comments
PRs + stars + releases
Odds backed by real money
Post-Twitter conversations
AT Protocol posts
Editorial + blogs (cited)
Why no one else has this
Each platform is a walled garden.
Every platform has its own API, its own tokens, its own auth. No single AI gets to read all of them — except Clawback, which bridges them.
- ✗Google doesn't touch Reddit comments or X posts.
- ◐ChatGPT has a Reddit deal but can’t search X or TikTok.
- ◐Gemini has YouTube but not Reddit.
- ✗Claude has none of them natively.
- ◐Perplexity web search only — no TikTok, no Polymarket.
The unlock isn’t a better search engine. It’s a dozen disconnected platforms, bridged by an agent — so the last 30 days of upvotes, likes, transcripts, and prediction-market odds end up in one brief, on your screen, before you send the email.
How it works
Three steps from query to brief.
1. Enter a query.
A company, topic, URL, person, market, or prospect note. Pick a window: 7, 14, 30, or 90 days.
2. Clawback searches for recent signal.
It separates verified recent signals, stale information, weak signals, unsupported claims, and unavailable sources. Every claim cites a source URL.
3. You get an actionable brief.
Use it to write a better sales email, prep for a meeting, build a content angle, or decide what not to say. JSON for workflows, Markdown for humans.
Get a brief on your worst account
Two fields. One real brief in 24 hours.
Pick the company, person, topic, or URL you actually want to research right now — the prospect you’re stuck on, the competitor you’re tracking, the trend you keep hearing about.
We’ll run it through Clawback and email you back the brief — source-backed, freshness-bucketed, no hallucinations. If it’s useful, we’ll get you set up with the local Claude Code skill. If not, you got a free brief.
- · No credit card.
- · No sales call.
- · Built for outbound founders, SDRs, agencies, and content teams.
Who it's for
Five workflows where stale context kills the email.
Sales prospecting
Find the recent trigger before you send the email. Stop pasting LinkedIn job titles into prompts and calling it personalization.
Founder-led sales
Walk into calls knowing what changed at the company in the last 30 days, not just what their website says.
Content / SEO / GEO
Turn fresh market changes into posts, briefs, outlines, and claims worth verifying. Built for the era of Google Discover and AI-cited answers.
Agency workflows
Run repeatable research across clients, prospects, competitors, and markets. Same brief format every time, source-backed and dated.
Competitive monitoring
Track recent moves without drowning in alerts and tabs. Run a topic on a schedule, get a brief instead of a firehose.
FAQ
Questions buyers ask before they sign up.
What do I actually get?+
A source-backed brief about a person, company, topic, or URL. It separates verified recent signals from stale info, weak signals, and unsupported claims. Output as Markdown for humans and JSON for workflows.
Where does the data come from?+
Public engagement signals across Reddit, X, YouTube, Hacker News, GitHub, Polymarket, Bluesky, Threads, and the web — scored by upvotes, likes, views, and prediction-market dollars. Every claim cites a URL.
How fresh is "fresh"?+
Default window is the last 30 days. You can also pull 7, 14, or 90 days. Every signal is bucketed by recency: <7d, 7–14d, 14–30d, >30d.
Will it just make stuff up?+
No. Every claim in the brief traces to a source URL. If a source returned nothing, the brief says so explicitly under "Source Status." If signal is thin, you get a "verify before acting" banner.
How is this different from Apollo, Clay, or LinkedIn Sales Nav?+
Those are great for contact data and firmographics. Clawback is about behavioral and narrative signal — what changed recently, what people are saying, what angle is most likely to land — not who works where.
Will you scrape my LinkedIn / Gmail / CRM?+
No. Clawback never touches private data, gated platforms, or accounts that require auth. It works only on public signal.
How do I use it today?+
Right now Clawback runs locally as a Claude Code skill. Sign up for the beta to get the install steps and a sample brief on the topic of your choice.
Give Clawback one query.
Send a company, topic, or prospect. We’ll show you the brief it returns and whether it belongs in your workflow.