Marketing automation services that route leads and build reports

Every flow HorizonLux builds sends an alert to a named owner in Slack or email, with the failed step, the record and a link to the run. A sample reads: lead-intake stopped at step 3, Match in CRM, record 48213, after 3 retries. Calls to outside APIs retry, where the tool allows, before a run fails. During the build, our engineers watch these alerts. After handover, your team fixes flows with the runbooks, or we keep doing it on a support plan, responding within CONFIRM: response time.

Yes. HorizonLux lists every Zap you run, what it does and who owns it, then, with your approval, repairs, moves or retires each one. Monitoring matters here. Zapier's Help Center (checked 30 Sep 2026) says a Zap is turned off automatically when 95% of its runs error over seven days, and only Team and Enterprise plans get an email and a grace period before that. Every Zap we keep gets its own alert, so an error reaches a person instead of piling up unseen.

Yes, once the fields each flow reads are clean, which is why HorizonLux does that clean-up inside the paid pilot rather than as a separate audit. For each flow, we fix what it depends on: duplicates, the fields it needs and which tool owns each one. You approve every merge rule before it runs. After launch, a record the flow can't place by your rules, such as a lead that matches no territory, goes to a named person instead of the wrong rep.

HorizonLux recommends the tool your team can keep running at your volume, flow by flow. We weigh volume, how complex the flow is, where the data must live and who will maintain it. If your leads live in HubSpot, its own workflows may cover routing, but they need a Professional or Enterprise tier. Zapier bills in tasks, one per successful action step, and Make in credits. n8n Cloud bills per workflow execution, however many steps, and n8n can also run on your own servers. We use n8n daily.

Yes, HorizonLux can add AI steps to a flow that draft a follow-up, classify a reply or summarise a lead for the rep, using your brand rules. No AI draft sends itself. A person on your team approves each draft before it goes out, and every AI step is logged. We keep that review step because results vary: in a Gartner survey published on 29 Oct 2025, 45% of martech leaders using AI agents said vendor-offered agents missed their expectations. We build with OpenAI and Claude models.

HorizonLux builds each flow in your own CRM and automation accounts, so customer records pass only through tools you approve. If records must stay in the UK or EU, n8n can run on servers you control there. Our engineers work with the access you grant, each connection gets the least access its tool allows, and you can revoke both without us. Each runbook lists the personal data its flow moves and where, for your GDPR records. You own the flows, code, docs and IP. NDA on request.

HorizonLux agrees the price with you on the free 30-minute call, once it understands the flow, the tools it touches and roughly how clean your data is. We don't publish prices, because a two-step lead alert and a full CRM clean-up are different jobs. Work starts with a paid pilot on one flow. Running costs are separate and paid to your vendors: the CRM tier, automation tool usage and model calls for any AI step. We estimate those before the pilot starts.

The length is agreed on the free 30-minute call with Seif Sgayer, the founder, because it depends mostly on how clean your data is and how many flows you need. A routing flow on clean data moves faster than one that needs duplicates merged. HorizonLux starts with a paid pilot that has one agreed deliverable, and you can stop after it. Then flows are added one at a time, with a demo and a written update every week.