What is the difference between ID-based and agent-based remote access?

ID-based (attended, on-demand) access needs someone present at the far end to run a small application and share a code, and the session ends when the work does. Agent-based (unattended, managed) access installs a persistent service that holds an outbound connection open, so an operator can connect at any time with nobody present. Owned estates need the second; helping people who are present needs the first.

Do I need unattended access if my users are always at their desks?

Usually yes, for the machines nobody sits at: servers, tills, kiosks, signage, meeting-room PCs and anything in a plant room. It also removes the dependency on a user being available and willing to read out a code before you can start work, which matters most out of hours.

Why does a remote session feel slow on a fast connection?

Latency rather than bandwidth. A link with plenty of capacity but a long round-trip time feels sluggish because every keystroke and mouse movement waits for a return trip. Test candidates on a genuinely poor link and watch what each one degrades to keep the session usable.

How many operators should I license?

Base it on peak simultaneous sessions rather than headcount, then check how each vendor meters. Teams with several occasional operators are usually better served by concurrency-based or operator-unlimited pricing than by named seats.

Alternatives · 9 min read

Choosing an AnyDesk alternative for managed estates

Name: 247connect
Brand: 247connect

Written for: Support teams and MSPs who started with ad-hoc ID-based remote access and now manage an estate they own.

Session ID versus installed agent

An ID-and-code session needs someone at the far end to read out a number. An installed agent is already waiting, which is what estate ownership actually needs.

In short

Teams looking for an AnyDesk alternative usually want one of two things: a licensing model that suits several occasional operators rather than a fixed number of seats, or stronger unattended management of an estate they own. That second point is the one that reshapes the requirement. Ad-hoc connection by session ID is excellent for helping someone who is present; looking after two hundred company machines needs agents that are already installed, grouped, permission-scoped and logged, so nobody has to be at the far end to read out a number.

Key takeaways

Ad-hoc ID sessions and installed unattended agents solve different problems; most teams need both, weighted toward whichever the estate demands.
Once you own the devices, unattended access, device grouping and permission scoping matter more than raw session quality.
Check whether unattended access is included or sits in a higher tier, because that single line often decides total cost.
Exportable, operator-attributable session logs are a requirement in regulated sectors, not a nice extra.
Silent, scriptable deployment across the estate saves more time in year one than any in-session feature.
Model twelve months under each vendor's own current meter; per-seat and per-device pricing diverge sharply as a team grows.

The distinction that drives everything

Remote access tools split along one line: is software already on the far machine, or does someone put it there for the duration of a call? The ID-and-code model asks a present human to run a small application and read out a number. It is fast, needs no prior deployment, and is the right shape for helping a customer, a member of the public or a device outside your control.

The unattended model installs a persistent agent that holds an outbound connection open, so an operator can connect whenever needed with nobody present. It requires deployment, and it is the only workable shape for machines you are responsible for: servers, kiosks, tills, signage, plant-room PCs and any laptop whose user is not available to help you fix it.

The moment your estate is something you own rather than something you visit, your evaluation criteria change. Session smoothness stops being the differentiator and administration takes over: how devices are grouped, who can reach which group, what is logged, how agents are deployed and updated, and what happens when a machine is asleep or has been rebuilt.

What to score in an alternative

Use the same sheet for every candidate, and keep it to the rows that change the answer.

Unattended access included, not tiered: confirm managed agents are in the plan you intend to buy.
Device organisation: groups, tags, search and a device list that stays usable past a few hundred machines.
Permission scoping: whether a first-line operator can be restricted to the groups they support.
Authentication: enforced multi-factor on operator accounts, and sensible session timeout behaviour.
Audit: what is recorded per session, whether it names the operator, how long it is retained and whether it exports.
Deployment: silent installation via script, MDM or group policy, and how agents update themselves.
Network behaviour: outbound-only connections that work without inbound firewall rules or a VPN, including on home broadband and mobile tethering.
Attended sessions too, for the devices you do not own, with explicit user consent.
Hosting region and data residency, where sector policy or GDPR practice constrains it.
Twelve-month cost at your real operator, concurrency and device numbers.

Session quality: test it, do not read about it

Every vendor in this category claims a fast, smooth session, and on a good network every one of them delivers it. The differences appear on the connections your users actually have: a rural broadband line, a hotel network, a phone hotspot, a site behind a saturated uplink.

Test three things on a deliberately poor link. How long from clicking the device to a usable desktop. Whether typing and mouse movement stay tolerable when bandwidth drops, and what the tool degrades (colour depth, frame rate, resolution) to keep going. Whether file transfer in both directions completes without babysitting.

Latency, not bandwidth, is usually what makes a session feel bad. A link with plenty of capacity but a long round trip feels sluggish however fast the pipe is, which is worth knowing before you blame a product.

Cost modelling for occasional operators

The common failure is buying seats for people who connect rarely. A team of eight where two are ever connected simultaneously is badly served by per-named-operator pricing and well served by concurrency-based or operator-unlimited models. Conversely, a small team looking after a very large estate should watch per-device charges closely.

Build the model on your own numbers under each vendor's current published meter: technicians who need console access, peak simultaneous sessions, devices needing agents, and any add-on you would truly enable. Then re-run it at the headcount you expect in two years. The cheapest option today is often not the cheapest option at the point you grow, and switching again costs time.

Security expectations for an owned estate

An always-installed agent is a privileged component by design. That is manageable, and the controls are well established in public guidance: encryption in transit, enforced multi-factor authentication for operators, least-privilege scoping so operators reach only their own device groups, explicit consent for attended sessions, and complete session logs attributable to a named individual and retained long enough to be useful.

Add two operational habits that no product provides for you. Review the operator list on a schedule and remove leavers the day they leave. Read the session log occasionally, so it is a control rather than a file nobody has ever opened.

Evaluating an alternative for an estate you own

1. Count the estate honestly
Devices needing permanent agents, devices you support ad hoc, platforms and versions in use, and sites with awkward networks.
2. Confirm unattended access is in your intended plan
Not on the feature list somewhere: in the tier you are budgeting for, along with grouping and permission scoping.
3. Deploy to a representative sample
Include one old machine, one remote site and one device on a mobile connection. Use the silent install route you would use for real.
4. Test the unhappy paths
Sleeping device, saturated link, session handover, reboot with reconnect, two-way file transfer, and a device that has been rebuilt.
5. Check the audit output
Run several sessions, then export the log and confirm it names operator, device and time in a form your auditor would accept.

Common mistakes

Choosing on session smoothness when the real requirement is unattended administration of an owned estate.
Buying named seats for operators who connect once a week.
Assuming unattended access is included in the entry tier.
Testing only on the office network, where every product looks excellent.
Leaving agents installed on decommissioned machines, so the device count and the attack surface both drift upward.
Never reading the session log, which turns an audit control into a formality.

Frequently asked questions

What is the difference between ID-based and agent-based remote access?: ID-based (attended, on-demand) access needs someone present at the far end to run a small application and share a code, and the session ends when the work does. Agent-based (unattended, managed) access installs a persistent service that holds an outbound connection open, so an operator can connect at any time with nobody present. Owned estates need the second; helping people who are present needs the first.
Do I need unattended access if my users are always at their desks?: Usually yes, for the machines nobody sits at: servers, tills, kiosks, signage, meeting-room PCs and anything in a plant room. It also removes the dependency on a user being available and willing to read out a code before you can start work, which matters most out of hours.
Why does a remote session feel slow on a fast connection?: Latency rather than bandwidth. A link with plenty of capacity but a long round-trip time feels sluggish because every keystroke and mouse movement waits for a return trip. Test candidates on a genuinely poor link and watch what each one degrades to keep the session usable.
How many operators should I license?: Base it on peak simultaneous sessions rather than headcount, then check how each vendor meters. Teams with several occasional operators are usually better served by concurrency-based or operator-unlimited pricing than by named seats.

Why teams choose 247connect

Managed agents as standard
Unattended access to devices you own is part of the product, alongside on-demand sessions for devices you do not.
No per-operator penalty
Unlimited operators means occasional technicians, out-of-hours cover and apprentices cost nothing extra to add.
About eight seconds to a session
Fast enough that technicians reach for it during a phone call instead of asking the user to describe the screen.
Outbound-only, zero-trust connections
AES-256 in transit, no inbound firewall rules and no VPN, so remote sites, home broadband and tethered devices all work the same way.
Named-operator session logging
Every session attributable to a person and a device, which is what an audit or an incident review actually needs.

Where 247connect fits: it is built for exactly this shape of problem, an owned estate that needs quick, logged, permission-scoped access without a large platform to administer. It does not attempt to be a full RMM suite with deep patch automation and integrated billing. If your requirement is fast managed and on-demand access at a fixed price with unlimited operators, the 14-day trial covers enough devices to run the sample deployment described above.

Start a 14-day free trial See the feature list