Every request goes to the right person. Automatically.
Fivexer assigns each ticket, case or field job by skill and current load — and records who was considered, who was blocked, and why.
A working API key in about two minutes. No card, no sales call. Free to 10,000 tasks a month.
Matched in your browser — not a recording. Held; resume it above.
Testreel OÜ, registered in Estonia (reg. 12592678).
What routing by hand actually costs
Someone re-balances the queue by hand, every day
A team lead reads who is free, who is drowning and who is qualified, then moves work around. It is real work, it is never finished, and it stops the moment they take leave.
The best-fit person sits idle while the queue ages
Work lands where it was routed, not where it should have gone. The waiting is invisible until a customer counts it.
A case reaches someone who was never certified for it
One misroute, handled by the wrong person — and no record of who else was considered or why they were passed over.
The screen your ops lead runs the day from
Queues, waiting times and who is carrying the load — one view, and one click to pause, bump or reassign. Pause someone below. This one is not a picture of the console; it is the engine, running here.
Take someone off the board and watch where their work goes.
A re-creation of the Fivexer console, driven by a support-desk simulation held in this tab. Every number above is computed, not drawn — which is also why the waits read in seconds rather than minutes. Open the full simulator
You set the rules. The engine does the assigning.
Written down once, enforced on every request — then the engine learns who resolves what fastest inside those rules. It can never learn its way around one.
Skills and languages
Everyone's real capabilities, written in the same words your requests already use.
Strength of fit
How strongly each kind of work leans toward each person — all the way down to never.
Urgency and distance
What jumps the queue, and who is actually closest to the job.
Hard limits
Certifications and capacity are checked before anything is assigned. A zero-weight veto is absolute.
Every decision keeps its own record: who was considered, who was blocked, and the reason. That is what turns "why did they get it?" into a lookup instead of an argument.
Your first wins, in week one
No quarter-long rollout. Both land with the first routed request.
Misroutes stop on day one
People who aren't certified for a case are never eligible — not deprioritized. Blocked.
On the decision showing below, 5 of 7 candidates never reached scoring.
Every decision shows its work
Ask why did this go to Sam? and get the actual answer: who was considered, who was blocked, and what decided it. This one is live — a real record from the console view above.
+3 without the Technical skill — never considered
Whatever it is you are handing out
The ticket reaches someone who has actually solved this before.
Route by product area, language and tier — and keep the newest agent off the case that needs ten years.
Skip the infrastructure. Keep the engine.
Everything on this page, as a hosted API — your first task routed minutes after signup.
Runs in the EU. Tasks, workers and decision traces stay in Frankfurt. What doesn't stay in the EU.
import { Fivexer } from '@fivexer/sdk';
const fx = new Fivexer({
// your API origin + the sk_test_… key
baseUrl: process.env.FIVEXER_BASE_URL,
apiKey: process.env.FIVEXER_API_KEY,
});
await fx.workers.upsert({
id: 'agent_007',
tags: ['english', 'technical_support'],
});
await fx.tasks.create({
tags: ['english', 'technical_support'],
priority: 5,
});
// matched in ms — webhook fired, trace recordedOr hand it to an agent — 14 tools over the same API:
claude mcp add fivexer -e FIVEXER_API_KEY=sk_test_... \ -e FIVEXER_BASE_URL=<api origin> -- npx -y @fivexer/mcp
Works with the tools you already build with
Anything that can call an API can route. Shown for compatibility; no affiliation implied.
For the engineer you'll hand this to: match decisions compute in under a millisecond, measured live by the simulation on this page, over Redis sorted sets — no message broker alongside.
Show the veto the learning layer can't route around
await assignmentMatcher.addUser({
id: 'agent_1',
tags: [],
routingWeights: {
'support:*': 100, // eligible
'lang:*': 0, // hard veto for all language-tagged assignments
default: 0, // hard veto for default matching fallback
},
});What we can sign today — and where the gaps are
You are going to ask procurement to approve a young EU vendor. Here is the whole picture — what is already in place, and where the gaps are with what covers them today.
In place
Not yet — and what covers it today
The full account, including the subprocessor table and the backup policy, is at /security. Procurement questionnaires go to support@fivexer.com and get a straight answer about what we can and cannot sign.
“I wrote the routing engine, I run the infrastructure, and I answer this mailbox. If you are weighing up whether a small EU vendor is a safe bet for something as load-bearing as who gets your work — write to me directly and ask the hard version of the question.”
Viljar Võidula · viljar@fivexer.com · the engine on GitHub
Why teams pick Fivexer — and what it costs
The four differences that decide it, then published list prices against your own numbers — Fivexer Cloud is free up to 10,000/mo, paid tiers from €99/mo.
Your data, your infrastructure
Managed EU cloud, self-host the open-source (MIT) engine, or bring your own Redis — the locked-in platforms are hosted-only, at any price.
Rules that never bend
A zero-weight veto is absolute — the learning layer cannot override it. Elsewhere, that's config you maintain yourself.
Decisions you can explain
Every match records who was considered, who was blocked, and why — not just a log of what happened.
Any kind of work, fairly spread
Tickets, cases, field jobs, AI agents as workers — fairness modes keep anyone from drowning while others idle.
Pro plan — €99/mo ($107 converted).
100 free + 49,900 × $0.06 per month — usage-priced routing, the closest like-for-like.
25 agents × $115/agent/mo on annual billing — buys the whole helpdesk, not just routing.
back in your budget at these numbers — the cheapest alternative shown is $2,875/mo against Fivexer Cloud's $107/mo.
Need voice, SMS, or a full helpdesk built in? That's Twilio's and Zendesk's home turf — Fivexer is the routing engine only, by design. Feature-by-feature tables: Twilio TaskRouter, Zendesk, Genesys / NICE / Five9.
List prices as of August 2026: Twilio TaskRouter published rate ($0.06/task after the first 100 free); Zendesk Suite Professional, the entry plan with skills-based routing. Fivexer Cloud prices are published EUR list prices converted to USD at 1 EUR = 1.08 for comparison. Excludes negotiated discounts, HA replicas, and egress. The products overlap, not match — TaskRouter is the closest comparison; Zendesk bundles a full helpdesk. Free tier up to 10,000 assignments/mo. Paid billing starts when checkout launches; launch price locked for 12 months from first invoice. See /pricing →
What happens when launch pricing ends
- 60 days' notice before anything changes.
- 12-month launch price lock — early paid customers keep published prices from first invoice.
- 6 months grandfathered — launch workspaces keep their allowance.
- Export is always free, on every plan.
- The engine is MIT — run it on your own Redis, anytime.
In writing: launch terms · where your data runs
Right people. Right work. Right time.
A working API key about two minutes from now — no card, no sales call. If you ever want out, the engine is open source and your rules and traces leave with you.
Enterprise, bring-your-own Redis, or just questions? Talk to us.