For Australian businesses running a Yeastar PBX with Linkus

Adding an AI Receptionist to a Yeastar PBX

Yeastar does the hard part well. Your P-Series PBX routes the numbers, the ring groups find whoever is free, and Linkus puts the extension on every desk and every mobile. What it cannot do is answer. When the ring group times out at 8:30 on a Monday, when the office closes at five, or when all four reception staff are already on calls, the caller gets voicemail and you get a callback list. Yes AI adds an AI receptionist to the Yeastar system you already own, so those calls get answered, qualified, and either transferred back to the right Linkus extension or written up and sent to the person who needs them.

Four integration patterns: the front door, ring group overflow, after hours on your existing time conditions, or a single dedicated DID. Multi-site and multi-brand supported, so the AI knows which of your trading names was dialled before it speaks. Australian consultants, Australian voices, and honest advice about what has to change on the PBX and who has to change it.

Book a Yeastar Call (03) 9003 0111

Realistic ROI

Fewer missed

Calls answered instead of lost to voicemail

The AI picks up when the ring group times out, when the office is closed, and when the whole team is already on other calls

Hours/week

Reception time returned

Repeat questions about hours, address, job status and account balance stop landing on a human extension

Same day

Call outcomes in your systems

A written summary, the caller details and the next action reach the CRM or the inbox while the call is still fresh, instead of a sticky note

2 to 6 weeks

From scope to live

Most Yeastar rollouts go live inside a fixed, agreed timeframe once the dial plan is mapped and the routing change is scheduled

Why Put an AI Receptionist on Yeastar With Yes AI

Yeastar is a genuinely capable PBX, which is exactly why bolting an AI onto it badly is so easy. The system already has firm opinions about inbound routes, ring strategies, time conditions, presence and transfers, and an AI that ignores them creates more problems than it solves. Four things make the difference between an AI your team quietly routes around and one that becomes part of how the phone works.

We read your dial plan before we write a word of script

Two Yeastar systems on the same firmware can behave completely differently, because all the value lives in the configuration. Which DIDs land on which inbound routes, whether Sales is a ring group or a queue, what the no-answer destination actually is, who maintains the holiday list, and whether anyone still uses the IVR that was built in 2019. We go through that with the person who knows it, and we read your call reports for real volumes, peak times and abandoned calls. The brief comes out of your own system, not a generic script we drop on every phone line.

The AI works with Yeastar, not around it

Plenty of AI answering products want you to forward your main number away from the PBX entirely. That hands your reporting, your presence and your transfer logic to somebody else. We do the opposite. The AI answers, then hands the call back into Yeastar as a normal transfer, so the Linkus desktop and mobile apps ring as they always have, Do Not Disturb and presence still mean something, your call detail records stay complete in one place, and your existing IVR, queues and time conditions keep working. If you switched the AI off tomorrow, your phone system would be exactly as you left it.

Australian, and straight about the privacy question

We are an Australian consultancy and you will get a plain answer on the privacy side rather than a marketing one. Calls contain personal information, and transcripts and summaries contain more of it, so we decide deliberately where that data sits, who can see it, and how long it is kept, with the Privacy Act and the Australian Privacy Principles in mind. Recording and monitoring rules are not uniform across the states and territories, so announcements and consent are treated as a decision you make on purpose, and we write down what was decided so you can answer for it later.

One team, including the awkward conversation with your reseller

One team scopes the change, writes and tunes the conversation, builds the integrations into your CRM and calendar, and supports it afterwards. That includes the part most vendors avoid, which is dealing with whoever actually administers the PBX. On a lot of Australian sites that is the reseller or IT provider who installed the Yeastar, and getting an inbound route or a ring group failover changed means a scheduled window and a clear written request. We write that request, sit on the call while it is applied, and test it the moment it lands, rather than emailing you a set of instructions and wishing you luck.

Six Ways an AI Receptionist Sits on a Yeastar PBX

There is no single right way to do this, and the best starting point depends on how much of your main number you are willing to touch in week one. Four of these are placement patterns, one is the multi-brand case that Yeastar sites hit more than most, and the last is what happens after the call ends, which is the part your team actually feels.

Front door

The AI answers first, then transfers into your extensions

Your inbound route points the main number at the AI instead of the IVR. It greets the caller in plain Australian English, works out what they actually want, and transfers the call back into Yeastar: extension 204 for accounts, the Sales ring group, the Support queue. Because the transfer lands as an ordinary call on the PBX, Linkus rings on the desk and on the mobile exactly as it does today, presence and Do Not Disturb still apply, and the call appears in your Yeastar call reports like any other. Nobody on the team has to learn a new phone.

Overflow

AI on overflow when the ring group times out

This is the least disruptive place to start, and usually where we recommend beginning. Leave the number ringing your team first, then set the ring group destination for no answer, or the queue destination when the maximum wait time is reached, to the AI instead of voicemail. Staff behaviour does not change at all. The only calls the AI ever sees are the ones that were about to become a missed call or a voicemail nobody plays until Tuesday. If you want evidence before you commit the main number, this is the pattern that produces it, on your own callers and your own numbers rather than on somebody else's figures.

After hours

After hours through your Yeastar time conditions

Yeastar already knows your business hours, your out-of-hours rule and your public holiday list, and that logic is usually the best maintained thing on the whole system. We hang the AI off the out-of-business-hours and holiday branches of the time condition, so from five o'clock the call is answered properly rather than by a recording telling people to try again tomorrow. The AI takes the job details, tests urgency against rules you set, books into the diary where that is appropriate, and either escalates to the on-call mobile or leaves a written summary waiting when the office opens.

One DID

A single dedicated DID, with nothing else touched

If you would rather not go near the main number yet, point one spare DID from your SIP trunk at the AI and put that number on the new campaign, the website, the Google listing or the after-hours sticker on the door. Your existing inbound routes, IVR, ring groups and Linkus setup stay precisely as they are. It is the cleanest way to measure what the AI genuinely does, because every call on that number is attributable, and it is trivially reversible: remove one inbound route and it is gone.

Multi-brand

Several brands or sites sharing one PBX

Yeastar makes it easy to run three trading names, or four branch offices, off one PBX with a separate DID and inbound route for each, which is why so many Australian groups end up doing exactly that. The AI works the same way. Because the number that was dialled travels with the call, the AI knows which brand it is answering for before it says a word, so it opens with the right business name, quotes the right hours and rules, and transfers into that brand's ring group or that site's extensions. One AI, one set of integrations, several front doors that never get crossed.

After the call

After the call: summary, voicemail and CRM

Every call produces a written summary with the caller's name and number, what they wanted, what was agreed and what has to happen next, delivered to the inbox, shared mailbox or channel that owns it. Voicemail stops being a recording someone has to sit through: it arrives as text with the key details pulled out and flagged when it is urgent. Where you run a CRM such as HubSpot or Salesforce, or a job management, practice or booking system, we push the outcome straight in, so a call becomes a lead, a booking or a task without anyone re-typing it.

Six Australian Yeastar Setups and What Actually Changes

Task	Traditional	With Yes AI on Yeastar	Notes
Plumbing and electrical firm, three branches on one Yeastar PBX	One shared reception number, calls bounce between sites and get lost at the handover	AI answers each branch DID by name and transfers into that branch ring group	The Warragul caller never lands on a Dandenong extension. Each branch keeps its own after-hours rule and its own on-call mobile, without running three phone systems.
Allied health clinic with four reception staff in one ring group	Ring group times out to voicemail at the 8am and 5pm peaks, nobody calls back until lunch	Overflow goes to the AI, which books, reschedules or takes details and messages the desk	Reception is untouched during quiet periods. The only change is that the calls they used to lose come back as bookings or written messages the same morning.
Property services group trading under three brands off one PBX	A single generic greeting, and staff have to ask which company the caller thinks they rang	AI greets with the correct brand based on the DID and applies that brand's rules	Each brand keeps its own tone, hours and escalation path. The DID to brand mapping is audited and signed off before the AI goes anywhere near the numbers.
Wholesaler taking after-hours orders and delivery queries	Out-of-hours calls hit a recorded message, orders go to whoever answers first in the morning	Time conditions send after-hours calls to the AI, which captures the order details in writing	The warehouse starts the day with one clean list instead of eleven voicemails, and genuinely urgent freight problems escalate to the on-call mobile straight away.
Dealership or agency running a new marketing campaign	Campaign calls mixed in with everything else, with no honest way to measure them	Campaign gets its own DID pointed at the AI, and every call is qualified and logged	You can see exactly what the campaign produced, and the main number, the IVR and the existing routing are never touched at all.
Managed services or support desk with an escalating queue	Callers wait in the queue, abandon, then ring back and start the whole thing again	AI covers queue overflow, triages severity and either escalates or logs a ticket	Genuine outages still reach a human fast. The password resets and status questions stop consuming an engineer's afternoon in five-minute pieces.

Six Things That Go Wrong on a Yeastar Rollout, and How We Handle Them

Someone has to change the Yeastar routing, and it may not be you

Be honest about this from day one, because projects stall here far more often than they stall on the AI. Pointing an inbound route, a ring group failover or a time condition at the AI is a change on the PBX itself, and on plenty of Australian sites the PBX is administered by the reseller or IT provider who sold it, not by anyone in the office. Sometimes the DID sits with a different carrier again. We map who holds admin on the PBX, who holds the trunk, and who can move a number, before we quote. Then we either work through your provider with a written change request or take a scoped admin account for the change window and hand it back.

Trunk channels, concurrency and licences

An AI-answered call is still a call. It occupies a trunk channel for as long as it runs, and briefly needs a second leg while a transfer is being set up. If your SIP trunk is provisioned for the number of simultaneous calls you get today, a busy Monday with the AI catching everything that used to hit voicemail can push you into the ceiling. We count your real concurrent call peaks out of your Yeastar call reports rather than guessing, check what your trunk and your PBX edition actually allow, and tell you plainly if you need another channel or two before go live rather than after the first bad morning.

SIP, NAT and audio quality

Voice is unforgiving, and poor audio looks exactly like an AI fault when it is not one. An on-premises Yeastar appliance sitting behind a router with SIP ALG switched on will produce one-way audio and dropped transfers. Aggressive codec transcoding, an oversubscribed office link, or QoS that was never configured will do much the same. We test the audio path end to end before anyone relies on it, ask for SIP ALG to be disabled where it is interfering, agree the codec, and listen to real recordings rather than trusting a green status light. Speech recognition degrades before a human would notice the line is poor, so this is not a step we skip.

Transfers, busy extensions and a directory that drifts

Deciding to transfer is the easy part. What matters is what happens when extension 204 is already on a call, is on Do Not Disturb in Linkus, or simply is not picked up. Without a defined fallback the caller ends up in a voicemail box nobody checks, or worse, back at the beginning. We define the fallback for every single transfer target: try the extension, then the ring group, then take a message and send it in writing. We also plan for the fact that staff leave and extensions get reused, so the directory the AI works from has a named owner and a review point instead of quietly going stale over a year.

Multi-brand only works if the DID mapping is clean

The AI can only greet with the right trading name if the number that was dialled reaches it correctly, and on a PBX that has grown over several years the inbound routes are almost always messier than anyone remembers. Old numbers still route somewhere. One brand's overflow lands on another brand's group. A couple of DIDs point at an extension belonging to someone who left in 2022. We audit every inbound route and DID before the AI is put in front of them, and we get the mapping signed off in writing, because a caller greeted with the wrong company name does more damage than the missed call would have.

What we do not let the AI decide, and the recording question

The AI is a receptionist, not a clinician, a lawyer or an estimator. We do not let it give clinical or legal advice, commit to a firm price on a job it has not seen, promise a delivery date the system cannot confirm, or read back full card numbers. Emergencies and any caller in obvious distress route to a human immediately, and the exact wording for that is agreed with you and tested before go live. Separately, if you record calls, treat that as a real decision rather than a default setting: recording and monitoring rules differ between the states and territories, personal information in calls and transcripts falls under the Privacy Act and the Australian Privacy Principles, and announcements, access and retention should be set deliberately. What we give you there is general information and a sensible starting position rather than legal advice, and where the answer really matters we will say so and point you at your own adviser. What we will not do is pretend one blanket rule covers every state, because it does not.

How Yes AI Helps

Dial plan discovery

We take a proper look at the PBX before proposing anything: every trunk and DID, every inbound route, the ring groups and queues and their no-answer destinations, the time conditions and holiday list, who sits on which Linkus extension, and what the call reports say about volume, peaks and abandoned calls. You get a written picture of your own phone system, which most businesses have never actually seen, plus a recommendation on which pattern to start with and why.

Build, integration and the PBX change

We build the AI, write and tune the conversation for your business and your brands, wire it to the number, and connect it to the systems the outcome has to reach: your CRM, your calendar, your job management or practice software, your inbox. We also handle the Yeastar side of the change, working through your reseller or IT provider where they hold admin, so you are never left translating between two vendors who each think the other one owns it.

Secure and Australian

We are an Australian consultancy. Connections into the PBX and into your business systems use least-privilege credentials created for the job, never a shared admin login, over encrypted links. Where personal information is captured on a call, in a transcript or in a summary, we design to the Privacy Act and the Australian Privacy Principles, with access, retention and deletion decided on purpose and recorded, so you can explain the arrangement to a customer or an auditor without going hunting.

Monitoring, tuning and support

We do not hand you a number and disappear. For the first weeks we listen to real calls with you, read the transcripts, and tune the wording, the transfer rules and the escalation thresholds against what your callers actually say, which is never quite what anyone predicted in the workshop. After that we keep watching for failed transfers, unusual call lengths and anything the AI could not resolve, and adjust as your hours, staff, sites and services change.

Our Yeastar AI Receptionist Rollout

A clear five-step path from a first conversation to a live, monitored AI receptionist on your Yeastar PBX. Most Australian businesses are live inside two to six weeks, and the variable is almost never the AI. It is how many numbers, brands and sites are in scope, and how quickly the routing change can be scheduled with whoever administers the system.

Map the dial plan

We walk your Yeastar configuration with the person who knows it: trunks, DIDs, inbound routes, ring groups and queues, no-answer destinations, time conditions, holidays and the Linkus extension list. We pull the call reports for volume, peak concurrency and abandoned calls, and we establish who holds admin on the PBX and who controls the numbers. You get the current state written down before anyone proposes a change.

Design the placement and the conversation

We agree which of the patterns you start with, front door, ring group overflow, after hours on your time conditions, or a dedicated DID, and exactly which routes change. Then we write the conversation: what the AI asks, what it must never say, which extensions and groups it can transfer to, the fallback for each one, and the escalation rules. Multi-brand sites get the DID to brand mapping signed off here. You approve a fixed scope, timeframe and price before any build starts.

Build on a test number

We build against a test DID or a spare inbound route, never straight onto your main number. Greetings, qualification, transfer targets, fallbacks, out-of-hours behaviour, brand switching and the CRM and calendar connections are all proven on a number nobody is relying on. We test the audio path properly at this point, including the network conditions, so line quality problems surface here rather than in front of a customer.

Pilot on one pattern and listen

We go live on the narrowest useful slice, usually after-hours or ring group overflow, and run it alongside your existing process. You and we listen to real calls and read the transcripts together, then tune the wording, the transfer rules and the urgency thresholds against what callers genuinely ask. You are watching it work on your own callers, on your own system, before it goes anywhere near your main number.

Go live and keep tuning

We widen it to the agreed scope, brief your team on what changed and what now arrives as a written summary, and keep watching. We track failed transfers, unresolved calls, unusual call lengths and anything the AI handed to a human, and adjust. As you add a brand, a site, a service or a set of trading hours, the people who built it extend it.

FAQ

Do we have to replace or reconfigure our Yeastar PBX?

No. The whole point is that you keep the Yeastar system you already paid for. Your extensions, Linkus apps, ring groups, queues, IVR, time conditions and call reports stay exactly as they are. What changes is where one or more inbound routes point, or what a ring group does when nobody answers. That is a configuration change, not a migration, and it can be undone in a couple of minutes if you decide the AI is not for you.

Does this work on the P-Series cloud edition as well as an on-premises appliance?

Yes to both, and to the older on-premises Yeastar systems that plenty of Australian businesses are still running happily. The pattern is the same either way, because we are working with your SIP trunk and your inbound routing rather than depending on a particular box. What does differ is the practical detail. An on-premises appliance sits behind your own router, so the network path and settings like SIP ALG matter far more, while a hosted P-Series usually makes the change window easier to schedule. We confirm which you have during discovery and plan accordingly.

Can the AI transfer a call to someone on their mobile through Linkus?

Yes, and this is the part people underestimate. Because the AI hands the call back into Yeastar as an ordinary transfer to an extension or a ring group, whatever that extension already does still applies. If a staff member has the Linkus mobile app registered, their mobile rings alongside the desk phone exactly as it does when the call comes through your IVR today. Their presence and Do Not Disturb settings apply too, which is precisely why we define what happens when a transfer is not answered rather than letting the caller find out.

Will this use up our trunk channels or need extra licences?

An AI-answered call occupies a trunk channel for as long as it runs, the same as any other call, and it briefly needs a second leg while a transfer is being set up. For most businesses that sits comfortably inside what they already have, but it is worth checking rather than assuming, because the AI answers calls that previously hit voicemail and hung up, so your real concurrency can rise. We read your existing Yeastar call reports for peak simultaneous calls, check what your trunk and PBX edition allow, and tell you before go live if you need another channel. Licensing depends on your Yeastar edition and your carrier arrangement, so we confirm it with you rather than guessing.

We run three businesses off one PBX. Can the AI answer differently for each?

Yes, and it is one of the strongest reasons to do this on Yeastar specifically, because running several trading names off one PBX with separate DIDs and inbound routes is such a common setup. The number the caller dialled travels with the call, so the AI knows which brand it is answering for before it says anything. It opens with that business name, follows that brand's hours, rules and qualifying questions, and transfers into that brand's ring group or that site's extensions. You get one AI and one set of integrations behind several front doors. The critical piece is auditing and signing off the DID to brand mapping first, because a caller greeted with the wrong company name is worse than a missed call.

Who makes the changes on the PBX, us, our reseller, or you?

Whichever is cleanest for you, and we will say plainly which we recommend once we know how your system is administered. On a lot of Australian sites the Yeastar was supplied and is managed by a reseller or IT provider, in which case we write the exact change request, brief them properly, and test it live the moment it is applied. If you hold admin yourself, we can either walk your person through it on a call or take a scoped admin account for the change window and hand it back afterwards. What we will not do is send you a PDF of instructions and call that an implementation.

How long does it take and what does it cost?

Most Yeastar rollouts go live in two to six weeks. A single dedicated DID or an after-hours overflow sits at the fast end. Several brands, several sites and CRM integration sit at the longer end, and the pacing is usually set by when the routing change can be scheduled rather than by the AI itself. On price, we scope properly instead of quoting sight unseen. Consulting engagements start at $1,000 for a focused BASIC piece, $3,000 for a STANDARD engagement, and $10,000 for a STRATEGIC one across a wider estate. Running an AI voice agent starts at $299 per month for 500 minutes and $599 per month for 1,500 minutes, with setup from $499 and higher volumes quoted individually. All figures are in AUD, and you get a fixed scope, timeframe and price after discovery.

What happens to voicemail, and can call outcomes reach our CRM?

Voicemail stops being a queue of recordings that someone has to sit and listen to. Calls the AI handles produce a written summary with the caller's name and number, what they wanted, what was agreed and what has to happen next, delivered to the inbox, shared mailbox or channel that owns it, and flagged when it is urgent. Where you already run a CRM such as HubSpot or Salesforce, or a job management, practice or booking system, we push the outcome straight in so the call becomes a lead, a booking or a task without anyone re-typing it. If you would rather keep it simple at the start, an emailed summary into a shared mailbox is a perfectly reasonable first step and the CRM can be wired later.

Find Out Which Calls Your Yeastar PBX Is Dropping

Book a call and we will walk your Yeastar dial plan with you, the inbound routes, ring groups, time conditions and call reports, and show you exactly where an AI receptionist would earn its keep first. No obligation, no jargon, and no pressure to touch your main number on day one.