Coldreach vs Outstack (2026)
The decision is whether anyone is answerable for the fact that a machine wrote it. Coldreach holds the whole chain. It monitors more than 113 million companies against a signal the operator describes in plain English, researches each lead across the company website, the regulatory filing, the hiring page, recent news and social activity, writes from what it finds and sends it, with an auto pilot mode running around the clock and no approval step, review queue or confidence threshold published. The research genuinely happened and no person did any of it, and the recipient has no signal available that would tell the two apart. Outstack splits the chain. It writes about a scraped individual and hands the copy to whatever tool the customer already sends from, so the moment of contact falls outside its walls while the authorship question originates inside them, and it offers no disclosure line, sender transparency option or downstream guidance. One vendor could answer for this and does not. At the other the question falls between two companies, and only the buyer stands in both halves.
- The deliverability stack is named by protocol and included at the entry price. Authentication is enumerated individually, covering sender policy framework, domain keys and domain based message authentication, alongside mailbox warmup, send pacing and reputation monitoring, and the whole is presented as included on every tier rather than as a feature to buy separately. Naming the three authentication standards rather than gesturing at deliverability is what separates this from most of the field. What is still missing is a threshold: no bounce or complaint limit, no blocklist monitoring and no inbox placement testing is published.
- The mechanism is described even though the model is not. The vendor names the document classes read for each lead, covering the company website, the regulatory filing, the hiring page, recent news and social activity, names what is extracted from them as a pain point, a status quo and an urgency, and explains the matching architecture, that a plain English description of a buying signal is evaluated against five or more data sources rather than resolved into filter boxes. That is real mechanism disclosure in a category that usually offers a logo and an adjective.
- The outcome numbers carry denominators. A customer result is published as 1,768 leads reached, 103 replies and 34 interested leads in a first month, two testimonials carry a full name, job title and named employer, and the vendor publishes its own internal figures with the attempts attached, stating that its agent books around 70 percent of its inbound pipeline and produces ten to fifteen qualified meetings a week from 700 to 800 outreach attempts. A 3.8 percent reply rate is stated across more than 500,000 messages. Discount the framing around that last figure, which is presented against an industry baseline the vendor supplies itself.
- What it declines to do is worth reading. The marketing sells conversion rather than undetectability, with nothing offered about evading spam filters or passing as a human, and a manual review and approval step sits between message generation and export, so a person reads the machine written copy before it leaves the platform for a sending tool. In a category where the recurring failure is manufactured signal, a vendor that neither manufactures identity nor markets evasion has made two deliberate choices.
- The research and the writing are metered separately, which is a budget control rather than a packaging quirk. Prospect profile enrichments and copy generation credits are sold as distinct units on the entry plan, so a buyer can scale the research without scaling the generation, or run enrichment alone and write the messages themselves. Every gating quantity sits beside the three published prices, and all features are stated as available on all plans with only volume limits differing.
- Bulk retrieval is designed in rather than negotiated at the end. Unlimited export of prospect data ships on every plan including the entry tier, the vendor states separately that all prospect data and generated messages can be exported, and an outbound endpoint will post the whole set to any system the customer nominates. That matters more here than at most vendors, because no terms of service, subscription agreement or privacy policy exists anywhere on the site, so the product is the only place a portability commitment lives.
Plain facts
| Coldreach | Outstack | |
|---|---|---|
| Primary category | AI SDR & Outbound Agents | Data & Enrichment |
| Founded | Not published | Not published |
| Headquarters | Not published | Exmouth, United Kingdom |
| Website | coldreach.ai | www.outstack.co |
Side by Side
| Axis | C Coldreach |
O Outstack |
|---|---|---|
|
AI Capability
|
||
| AI Centrality | ||
| Autonomy and Oversight Model | ||
| AI Disclosure and Model Transparency | ||
| Operational and Outcome Evidence | ||
|
Compliance and Risk
|
||
| Outreach Compliance Posture | ||
| Data Privacy Posture | ||
| Data Licensing and Provenance | ||
| Platform Terms Exposure | ||
| AI Safety and Data Stewardship | ||
| Recipient Disclosure and Authenticity | ||
|
Integration and Deployment
|
||
| Ecosystem and Integration Depth | ||
| Deployment Model and Data Residency | ||
| Security Certifications and Trust Center | ||
|
Commercial and Operational
|
||
| Commercial Transparency | ||
| Exit and Data Portability | ||
| Deliverability and Sending Discipline | ||
| Segment and Market Coverage | ||
The short version of each
Coldreach
Coldreach is a Y Combinator backed sales development agent built on the premise that the reason to contact an account matters more than the contact details, monitoring more than 113 million companies against a buying signal the operator describes in plain English. The GTM Tech Index grades it C on recipient disclosure and authenticity and records it as the clearest case of manufactured effort in the index, because the effort is not simulated at all: the system genuinely read the company's regulatory filing, its hiring page, its news and its social activity, found something specific and true, and opened the email with it. What did not happen is a person doing any of it, and no signal available to the recipient would distinguish the two.
Source: GTM Tech Index, 2026
Outstack
Outstack is a prospect research and message generation layer that sits upstream of whatever sending tool a team already runs, researching each imported person individually and producing a lead score plus personalised copy written against the customer's own product. The GTM Tech Index grades it C on recipient disclosure and authenticity, where the shape is a split rather than a deception: machine written copy about a scraped individual is produced here and sent elsewhere, so the moment of contact happens outside the vendor's walls while the authorship question originates inside them. No disclosure line, identifier or sender transparency option appears in the generated output, and no guidance is given to the customer about disclosure downstream.
Source: GTM Tech Index, 2026
Common questions
Is Coldreach better than Outstack?
They sit at the same grade on recipient disclosure and arrive there from opposite architectures, which is the decision. Coldreach owns research, writing and sending, so one company could answer for the fact that a machine produced the message, and no approval step, review queue or confidence threshold is published anywhere. Outstack writes and hands off, so the moment of contact happens inside a third product it never sees. The GTM Tech Index grades Coldreach A on AI centrality and B on AI disclosure, deliverability, ecosystem and operational evidence, against Outstack's four D grades on privacy, platform terms, security and evidence.
Does the recipient know a machine wrote it?
Nothing at either vendor tells them, and Coldreach is the sharper case because the research is not faked. Its system reads the company's regulatory filing, hiring page, news and social activity, finds something specific and true, and opens the email with it, so the recipient receives exactly the individual attention that would once have taken a representative an hour, produced in seconds against a monitored universe of more than 113 million companies. The research is real and the researcher is not. Outstack generates the same kind of copy and offers no disclosure line, identifier or sender transparency option in the output, and no guidance to the customer about disclosure downstream.
Who does the sending?
Coldreach does and Outstack deliberately does not. Coldreach runs the full outreach layer with an auto pilot mode described as working around the clock, and publishes a named deliverability stack including sender policy framework, domain keys and domain based message authentication alongside warmup, pacing and reputation monitoring, included on every tier. Outstack connects no mailboxes, runs no campaigns and dispatches nothing, handing finished records and drafted copy to whatever tool the team already sends from, so the reputation consequences and the statutory obligations both land at that downstream step rather than with this vendor.
How much do Coldreach and Outstack cost?
Outstack publishes a usable price and Coldreach mostly does not, which is why the GTM Tech Index grades them B and C respectively on commercial transparency. Coldreach does publish five real figures, at 899, 1,099, 1,499 and 1,999 dollars a month for self service tiers and 3,000 for a fully managed service, and there is no pricing page holding them, no volume attached to any tier, no lead, message, seat or signal quota anywhere, and every call to action leads to the same demo form. Outstack publishes sixty seven, one hundred and ninety seven and six hundred and ninety seven dollars monthly with every gating quantity beside them.
How much of the research is real?
Genuinely real at both, and only one of them tells you what it read. The index grades Coldreach A on AI centrality, the removal test leaving nothing behind, because signals are described in plain English and matched by comprehension rather than filter boxes across five or more named source types covering job postings, company news, website content, social posts, employee changes and securities filings. Outstack researches each imported person individually as well, and names one source for how they arrived, a search scraper for a professional network's paid prospecting product, with the enrichment layer behind it naming no partner and no verification provider.
How does the GTM Tech Index grade Coldreach and Outstack?
Both are graded on the same seventeen capability and compliance axes from public sources, at documentation depths of seventeen and thirteen respectively. The GTM Tech Index grades Coldreach A on AI centrality, and B on AI disclosure and model transparency, deliverability discipline, ecosystem depth and operational evidence, with every other axis at C and no D anywhere. Outstack takes B on AI centrality and commercial transparency, with D grades on data privacy posture, platform terms exposure, security certifications and operational evidence. The index publishes no composite score and declares no winner.
Related comparisons
Other published head to head assessments involving these vendors or their closest peers. The full set for this category is on the Data & Enrichment page.
The shared silence is training, tenancy and retention, and the interesting part is that the two absences are not the same kind of thing. Neither vendor states whether customer material trains or improves any model, whether learning is scoped to a single tenant, how long prompts, research output and generated messages are held, or which subprocessors see any of it.
At Coldreach that gap is unverified rather than established, because two legal documents exist and are published, terms of service and a privacy policy, and neither was read on this pass, so either could carry the answer and both should be checked before a buyer relies on the absence. At Outstack the same gap is established, because no terms of service, subscription agreement or privacy policy exists at any route on the site, and the footer is complete and short enough to settle it.
The material at stake is substantial on both sides: at Coldreach a research corpus assembled on more than 113 million companies alongside customer records synchronised from a system of record and prospect replies flowing back, and at Outstack scraped personal data on named individuals held beside each customer's own commercial positioning, which the platform is explicitly told to learn in order to write in their voice. The exposure worth naming follows from the split this pair turns on.
Article 50 of the EU AI Act requires that a person be told when they are interacting with generated content, and the obligation is generally understood to attach to whoever deploys the system into contact with that person, which at Coldreach is its own customer and at Outstack is a customer using a third sending tool the vendor never sees. Neither publishes a position.
The benign reading is available and is probably correct, since obligations of this kind are ordinarily handled in terms rather than on marketing pages, and nothing here asserts that either vendor is out of compliance with anything.