Hellomatik

HM-2's core idea is a single 'company blueprint': one written definition of each data item, each procedure and each formula, built by reading a company's existing systems (ERP/CRM, warehouse, email) on a schedule the company sets. Agents work from that blueprint in four ways — answering read-only questions, raising alerts when a tracked figure changes, drafting multi-step 'automations' (procedures done the same way every time), and preparing decisions for a human 'approver' to accept, amend or reject. A worked logistics example on the site shows four concrete tasks: a daily list of unshipped orders with delay reasons, a carrier freight-invoice audit that flags mismatched lines, drafted replies to order-status enquiries checked against the warehouse and carrier tracking systems, and drafted replies to delivery-issue reports with proof-of-delivery attached (https://hellomatik.com/).

The vendor's stated learning cycle is three steps: the agent drafts a proposal from the blueprint and prior approved corrections; the approver accepts, amends or rejects it, and that ruling is recorded in an approval log; and after any amendment or rejection, the agent drafts a change to the procedure itself, which only takes effect once the approver accepts it — so a single correction is meant to apply to every similar decision going forward rather than being relearned each time (https://hellomatik.com/product). The company also states each data item and each figure has exactly one definition and one formula company-wide, so every department reasons from the same numbers rather than each team keeping its own version.

There is no pricing page, no plan tiers and no free trial anywhere on hellomatik.com; the only calls to action across every page ('Book a meeting', contact form) lead to a sales conversation. This is consistent with a bespoke, per-company 'blueprint' product — building a definition of one company's systems is not a commodity SKU — but it means a buyer cannot compare cost against competitors, or even get a ballpark, before talking to sales.

Hellomatik's privacy policy states it does not use customer content — inputs, outputs, call audio, transcripts, chats, workflow payloads — to train AI models by default, and any such use requires separate, revocable opt-in consent (https://hellomatik.com/privacy). Voice-module defaults: call recordings kept 14 days, transcripts kept 6 months, with 10 days of post-cancellation retention for reconciliation before deletion or anonymisation. It states processors are bound under GDPR Article 28, transfers outside the EEA/UK rely on Standard Contractual Clauses or the EU-US Data Privacy Framework, and it publishes a formal Vulnerability Disclosure Policy under RFC 9116 with a stated 3-business-day acknowledgement for security reports (https://hellomatik.com/security). None of this is independently audited — there's no SOC 2, ISO 27001 or similar certification claimed on the site.

The legal entity is Hellomatik, S.L. (CIF B22803126), registered in Spain (https://hellomatik.com/privacy). Its own Substack, active from July 2025, described an earlier product framed around answering calls and messages across voice, WhatsApp, chat and email (https://hellomatik.substack.com/about); the current hellomatik.com site is built around the broader 'company blueprint' and HM-2 platform positioning, and its most recent substantive Substack posts date to October 2025, so it's not established whether the company has since repositioned from a conversational-support tool to a decision-automation platform or is running both. We found no independent reviews (G2, Capterra listing pages returned bot-verification walls or 404s; no Trustpilot profile exists) and no funding or headcount information from a primary source, so claims about scale or customer base should be treated as unverified until the company publishes them or a reader can confirm directly.