Blog

Almost every product in interview scheduling now describes itself as AI-powered. To a large extent, that is quite accurate. 

The question of whether the software is autonomous though is very difficult to answer. While this can’t be evident in a sales demo, asking the right questions can surely resolve it. 

The five questions below are all requests to see something, and they are ordered so that each one is harder to satisfy than the last. 

1. Ask to see a booking that nobody approved 

Request a specific instance in which the software selected a time, sent it, and no person touched the decision. Ask to see the audit trail alongside it. 

This establishes whether autonomy exists at all, or whether human approval has simply moved to an earlier point in the process. A good number of products in this category act independently only after somebody has confirmed a shortlist of times, which is a genuine convenience and a different thing from what the marketing describes. 

If the response is an offer of flexibility, along the lines of being able to configure how much oversight you want, the useful follow-up is to ask what the system does on its default setting and how many customers change that setting. Configurability tells you what the software can be told to do. It does not tell you what it does. 

2. Ask to see a thread with an executive assistant 

Request a real exchange between the software and an EA who manages a senior’s calendar. Ask for the whole thread, from first contact through to a confirmed time. 

This is where senior hiring is decided. An executive assistant will not create an account in a vendor’s portal in order to release their person’s time, and they tend not to answer with availability. They answer with terms and with questions of their own, such as which colleagues will be in the room and how long the meeting needs to run. A system that cannot conduct that exchange in ordinary email will be worked around within a month, whatever the product roadmap says. 

Examine the thread for two things. Did the software answer the assistant’s questions accurately, and did it hold the assistant’s conditions when it went on to arrange everybody else? 

3. Ask to see the exception log 

Request a summary of everything the system escalated to a human being over the last thirty days, along with the reason for each escalation. 

This is the single most revealing document in an evaluation, and it is the one least likely to be volunteered. Every autonomous system has a boundary, and the escalation record is where that boundary is written down. A vendor who cannot produce such a record is telling you either that the escalations are not measured or that they are not something the vendor wishes to discuss, and both answers are informative. 

4. Ask to see a loop that broke in the middle 

Request a case in which a multi-round sequence had to be re-planned, and ask specifically what happened to the rounds that came after the one that moved. 

Interview loops carry structure that is invisible in a demonstration and expensive in production. There may be a panel that cannot convene until a technical assessment is complete, an interviewer who must not meet the candidate before the debrief, or buffers between rounds that stop meaning anything once a round crosses a timezone boundary. When the fourth round of nine slips by a week, the question is whether the system understood the remaining five as a structure that needed re-planning or as five independent bookings that happened to concern the same person. 

5. Ask to see what the software wrote in the ATS 

Request the system-of-record entries generated for one candidate through a complete multi-round process, unedited. 

Autonomous action that nobody can audit creates a new problem in place of the old one. In regulated hiring, where the process itself may be reviewed, an action without a record is worse than an action taken slowly. What you are looking for is whether the bookings, the changes, the reasons behind those changes, and any feedback gathered afterwards arrived in the system without a person transcribing them. 

An answer that describes an integration is not an answer to this question. An integration describes a pipe, so examine what travels through it, in which direction, and which side initiates. 

When a vendor cannot share any of this 

Some of these requests touch customer data, and a refusal on those grounds is legitimate rather than evasive. There are three reasonable ways around it, and a vendor who wants to be evaluated properly will offer one before you have to ask. 

They can redact and anonymise, which works well for the assistant thread and for the ATS entries. They can run the exercise in a sandbox seeded with real complexity. Or they can introduce you to a reference customer who is willing to open their own account and walk you through the last month of it. 

What should alarm you is a vendor who offers none of these and returns to the tailored demo instead. 

Our own exception log 

Evie handles the first, second, fourth, and fifth requests without difficulty, and those are the questions we often get asked. 

On the third, our logistics escalations are limited by constraints that nobody has recorded anywhere. If a director informally avoids Friday afternoons, or two colleagues should not sit on the same panel, the system does not know until somebody tells it, and until then it will keep proposing arrangements that are technically valid yet socially wrong. 

We believe that is the right boundary to have, because it is a gap in knowledge rather than a gap in authority. A knowledge boundary moves as the system is told more. A boundary built into the design does not move at all. It remains a real limitation, though, and a vendor who claims to have no equivalent is worth a longer conversation rather than a shorter one. 

Why this is a better use of an hour than a demo 

Evidence of past behaviour predicts if the software is a right fit. Questions like the above will tell you more about the next year of your coordinators’ working lives than any feature comparison, and they take less time to make than the demonstration they interrupt. 

We built Evie as an autonomous coordinator rather than a scheduling assistant, which is why we have often faced questions like these on our sales calls. Companies using Evie report scheduling roughly 10× faster, around 50% fewer reschedules, and close to three months of recruiter time returned each year.

Here is a message that arrives in a recruiting coordinator’s inbox several times a week. 

“Thanks for sending these over. Tuesday at three does not work and Wednesday is difficult. He could probably manage Thursday morning if you can keep it before eleven, although I would need to know who else is joining and roughly how long you will need him for. Is this the final stage, or is there another round after this one?” 

That message came from an executive assistant and is often a common scenario at which most interview scheduling software stops being useful, because the message contains nothing a calendar can act upon. There is no slot to book. There is a condition, a partial refusal, two questions, and an unstated requirement that somebody explain the shape of the process before any time is released. 

We think such moments explain most of what has frustrated talent teams about scheduling automation over the past decade, and it explains why buying better calendar software cannot fix it. 

People do not send availability, they send positions 

Products in this category are built on a reasonable assumption, which is that the job involves collecting availability from several parties and finding the overlap. Availability is data. It can be requested, stored, compared, and resolved, and software is already very good at all four of those operations. 

The difficulty is that people rarely send availability. They send positions, meaning a statement of what they are willing to do and under what circumstances, offered with the expectation that somebody will respond to it. A position is not a data point waiting to be filled. It is a move in a conversation, and what it requires in return is a reply rather than a lookup. 

Once you see coordination this way, the pattern of failure becomes fairly easy to predict. Scheduling software handles the cases where somebody happens to answer with clean availability, which tends to mean junior candidates, single-round screens, and interviewers with unprotected calendars. It struggles everywhere else, and everywhere else is where senior hiring lives. 

The four queries that calendar software cannot answer 

• The condition: Thursday morning works, but only before eleven, and only if the meeting is kept to forty-five minutes. A condition is not a slot. It is a slot with terms attached, and honouring it may require moving somebody else who has terms of their own. 

• The counter-offer: None of your proposed times work, but here is a different one, which happens to fall outside the window you were working within. Accepting a counter-offer means recalculating everything downstream of it, and declining one means opening a further round of correspondence with somebody senior enough that the correspondence needs care. 

• The question: Who else is on the panel, how long will this take, and is there another stage after this? These are not obstacles to scheduling. They are the assistant doing their job, which is to protect their principal’s time by understanding what it is being spent on. The time will not be released until somebody answers, and the answer has to be accurate. 

• The silence. No reply at all after two days, from somebody whose participation the whole arrangement depends upon. Silence is the hardest of the four, because it carries no information about what to do next. Somebody has to decide whether to follow up, work around the person, or reopen the entire sequence. 

None of these requires more computing power. They do require a decision about what to do next, and that has to be made by something with the authority to make it. 

Every unanswered reply becomes a wait state 

The commercial consequence of all this is easier to see if you follow a single loop. 

Suppose an interview process has five rounds and each round involves one exchange of the kind described above. A coordinator who can attend to those exchanges immediately resolves each in a matter of hours. A coordinator who is handling forty other requisitions gets to them the following morning, and if the reply arrives on a Friday afternoon, they resolve them on Monday. 

These waits accumulate. Two days lost per round across five rounds is a fortnight added to a hiring process, which is long enough for a candidate in a competitive market to accept something else. 

When teams review a slow hire afterwards, the delay is usually attributed to unresponsiveness, either the candidate’s or the panelist’s. In our experience a good deal of it is simply the sum of replies that sat unanswered because the person who could answer them was busy answering others. 

Of the 300k+ conversations Evie has coordinated, the ones worth discussing are the assistant-managed executive calendars, the multi-round sequences, and the cross-timezone loops, because those are the cases that generate replies of all four kinds and generate them repeatedly. 

Self-service does not remove the negotiation, it relocates it 

The common way to respond to such problems is to send the candidate a link with several times on it and let them choose. We understand the appeal, and for straightforward hiring, it works perfectly well. 

In complex hiring though, it just moves the negotiation onto the people least equipped to conduct it. The to-and-fro emails become so extensive that eventually the scheduling tool is being used for simple bookings and worked around for the difficult ones. 

The work did not disappear. It moved into inboxes, where it is neither visible nor measurable. 

The question worth asking instead 

If coordination is a negotiation, then the useful thing to establish about any system is not how quickly it books a meeting. Rather, ask what the system does in difficult scenarios and handle email queries. 

• Will it read the condition, honour the constraint, answer the assistant’s two questions accurately, and commit to a time without waiting for somebody on your team to approve it? 

• Will it recognise the counter-offer as a counter-offer and work out what has to move? 

• Will it notice the silence on day two and decide what to do about it? 

These are answerable questions, and the answers predict what a team will experience in the third month far better than any feature comparison does. What autonomy changes is who handles the reply at those critical moments, rather than whether awkward replies arrive at all. 

Coordination has never really been about finding time in a calendar. It has been about conducting a conversation with several people who each have a legitimate claim on their own hours, and doing it quickly enough that a good candidate is still available at the end of it. Software that cannot hold that conversation will keep handing it back, and it will keep handing it back at the moments when the process can least afford to stop. 

We built Evie over 10 years ago to work the negotiation, not the calendar. Since day 1, Evie has been an autonomous coordinator. Teams running it report scheduling roughly 10× faster, around 50% fewer reschedules, and close to three months of recruiter time returned each year. 

See how it handles a multi-round loop: evie.ai/how-it-works 

An AI recruiting coordinator is software that owns interview coordination end to end; proposing times, negotiating with candidates, interviewers and their assistants, booking the logistics, handling reschedules, and writing the outcome back to the ATS. All this, without a person approving each step. 

What a recruitment coordinator actually does 

It helps to start with the human job, because the calendar is a surprisingly small part of it. 

A coordinator gathers availability from people who do not reply promptly. They assemble panels, then reassemble them when someone declines. They handle rooms, video links, dial-in details, etc. They write to candidates, confirm, remind, and manage the awkward conversations when something moves. They chase interviewers for feedback after the fact, and they keep the ATS honest so that the next person looking at the record can trust it. 

Finding a mutually free slot is perhaps a fifth of that. The rest is negotiation, exception handling, and record-keeping. Any assessment of what AI can take on has to be measured against the whole job, not the easy fifth. 

Autonomous or assistive: the distinction 

The clearest way to locate any tool is to ask what it does at five specific moments: 

1. Who decides the time: An assistive tool surfaces options and waits for a person to choose. An autonomous one decides, proposes times, and commits. 

2. Who contacts the candidate: While an assistive sends a link or hands the task back to a human, an autonomous platform writes to the candidate and holds the conversation that follows. 

3. The interviewer’s executive assistant: For most tools, EAs will not log into a vendor’s portal to pick a slot. An autonomous coordinator writes to the EA directly and negotiates. 

4. No reply after forty-eight hours: Assistive tools nudge, then the item sits in somebody’s queue. Autonomous follows up and, if the follow-up fails, replans. 

5. The ATS record: Assistive expects a human to update it. Autonomous writes back, because an action nobody can audit is not finished. 

In other words, autonomy is not binary, and most products sit somewhere along this line rather than at one end. 

Why such autonomy is difficult 

If the distinction were only a matter of removing confirmation clicks, every vendor would have crossed it years ago. Five things make it genuinely hard. 

• Negotiation, not just querying: The calendar rarely contains a clean slot that suits five people. The work is trading between constraints. Moving one interviewer to free another, deciding which of two imperfect options costs less, such decisions require judgment rather than a query. 

• The parties are outside your control: Candidates, executive assistants and external panellists have no logins, no obligation to adopt a workflow, and no interest in learning one. Anything that depends on their cooperation with a tool will fail. Email is the only reliable common ground. 

• Exceptions are the normal case: Interviewers drop a couple of hours before the interview. Candidates go quiet. Someone else books over a held slot. A system that only handles the happy path and escalates everything else is barely solving anything. 

• Sequences have memory: A multiple-round process is one object, not multiple bookings. Round seven may depend on who took round four. Holding that structure through various reschedules is a different engineering problem from booking a single screen. 

• The work must be auditable: Autonomous action without a trail in the system of record creates a new problem. When and where did things fall off? What made an interview schedule extend past the deadline? Without records in place, the process itself becomes subject to review. 

What an autonomous recruitment coordinator does not do 

An autonomous recruitment coordinator: 

•  does not decide who to interview 

• does not score or rank candidates 

• does not exercise judgment about panel composition 

• does not handle the conversations that need a person (an accommodation request, a difficult withdrawal, a candidate who needs reassurance rather than a calendar invitation, etc.) 

Coordination and evaluation are different jobs, and a system that blurs them is harder to trust with either. 

How to tell whether a tool is truly autonomous 

Four questions, best asked during a demo while evaluating an autonomous AI recruitment coordinator rather than after: 

1. Does it schedule without a person approving each step, or does approval simply move earlier in the process? 

2. Will it email an executive assistant and negotiate, or does it require the interviewer to self-serve? 

3. What happens at 11pm when an interviewer cancels for the morning? 

4. Who updates the ATS, and can that update be audited? 

The answers to these tend to be more revealing than any feature list. 

Evie has worked this way since 2013 — built as an autonomous AI recruitment coordinator before the category had a name, and now coordinated 300k+ interviews, including multi-round sequences, EA-managed executive calendars, and back-to-back on-site schedules. Teams using it report scheduling roughly 10× faster, around 50% fewer reschedules, and close to three months of recruiter time returned each year. 

Know more about Evie AI: evie.ai/how-it-works 

FAQ 

Is an AI recruitment coordinator the same as an AI recruiter? 

No. Sourcing and screening decide who enters the process; coordination runs the process itself. Conflating them is the most common source of confusion in this category, and it matters because the two carry very different risk profiles. 

Can AI schedule interviews with no human involvement at all? 

For coordination, yes, including multi-round loops, EA-managed calendars and cross-time zone panels. For deciding who to interview, no, and it is not meant to. 

Does it replace recruitment coordinators? 

It replaces the logistics of coordination. The parts of the role that involve candidate advocacy, process design, and judgment do not go away. Rather, they finally get the time back from endless schedule negotiations.