Hellomatik

Hellomatik unifies a company's existing systems — email, ERP, CRM, spreadsheets, warehouse and bank-connected data — into what it calls a live model of the business, then uses AI agents on top of that model to prepare decisions a team currently handles by hand (hellomatik.com/en, fetched 9 October 2026). Live use cases named on its homepage: logging incoming email orders with stock and credit already checked, matching supplier invoices against orders before payment, flagging a declining customer before it shows up in an annual review, prioritizing sales calls by buying pattern, and turning customer calls into comparable data. It says it is live at companies in manufacturing, distribution and healthcare. A July 2026 company blog post adds that its agents 'can operate twenty-eight platforms, from HubSpot and Mailchimp to Meta Ads and the Google Business Profile' — the company's own count, not independently checked (hellomatik.com blog, 30b-llm-raspberry-pi-record-3).

Founder Isaac Correa's September 2026 letter reports what is running with named limits: a logistics system in production applying 24 rules over more than 30 million operations and 460,000 locations (it 'detects, explains and warns; never moves an order'); a B2B sales system recalculating priority nightly for 29,000 customers over 14 million sales lines; a support system that drafts replies but sends nothing without an approved autonomy level; and a retail chat system restricted to defined procedures (isaac-correa.github.io/letters/2026/). These are self-reported, vendor figures, not audited — but the company backs its method with a technical report carrying a DOI ('How we build a company's ontology', 10.5281/zenodo.22647735) rather than marketing copy. Separately, co-founder Daniel Correa Villa published an open, reproducible benchmark (MIT-licensed code and raw data on Zenodo, 10.5281/zenodo.20357376) on running a 30B open model on four Raspberry Pi 5 boards, explicitly framed as groundwork for clients who won't let data leave their building. Publishing bit-exact, reproducible numbers and the 26 configurations that didn't work is a genuine, checkable research habit, unusual at this company's size.

Hellomatik was incorporated on 28 July 2025 in Madrid (founder's letter; ORCID 0009-0003-0657-4419). LinkedIn lists it as 2-10 employees, founded 2025, privately held (linkedin.com/company/hellomatik). No funding round for Hellomatik turned up in a search for funding or seed-round news; treat that as not established rather than as zero. Founder Isaac Correa, 24, previously ran Kodalogic, a Looker Studio dashboard-template business, from 2022. By his own account the team is small enough that 'the engineer who sits in the meeting is the engineer who writes the code, and stays' — a trade-off worth weighing against the support depth of a larger vendor, especially since the company's own case-study table names a client running 460,000 warehouse locations through the platform.

Hellomatik does not publish prices. Its founder states the model directly: 'The license is for the whole company, with unlimited users, and the figure is agreed in the first meeting' (isaac-correa.github.io/letters/2026/). The contact page confirms the sales motion: you explain which systems you use and which decision to start with, and get a proposal back (hellomatik.com/en/contact). There is no self-serve tier or published starting price anywhere on the site.

Hellomatik S.L. is the GDPR controller for account, billing and website data, based in Spain (hellomatik.com/en/legal/privacy). The policy states customer content — inputs, outputs, call audio, transcripts, workflow payloads — is not used to train AI models by default, and only with explicit opt-in consent. But the same policy has a full 'international data transfers' section covering data that leaves the EEA/UK via Standard Contractual Clauses or the EU-US Data Privacy Framework, and says its current list of sub-processors and their locations is available only on request, not published. That doesn't support a claim that hosting is confined to the EU or that each company gets a dedicated database — and no current page says either.

Hellomatik's homepage names three customers by logo — LIXIL Group, Joma Sports and Clínica Planas (hellomatik.com/en). But the founder's own September 2026 letter says the opposite about its published case studies: 'we do not name clients... if we cannot name the customer, we say \"an automated warehouse\"... The number is the proof; the logo is not' (isaac-correa.github.io/letters/2026/). Read together, these look like two different things rather than a contradiction — named reference logos for brand trust, with the measured performance figures kept anonymous — but a buyer who wants to check the LIXIL Group relationship against a dated case study with numbers attached won't find one; what's published is a logo, not a case.