Lindy has added a way to book meetings that needs no new app, no booking link and no login on the other person’s side. You put the Lindy address on the CC line of a thread where two people are trying to find a time, and the coordination stops being your problem. It reads what has already been said, pulls the scheduling details out of the conversation, and fills whatever is missing with sensible defaults.
From there the behaviour splits two ways. If the thread has already settled on a time, Lindy books it. If no time has been agreed, it replies once with three open slots as clickable links, then puts the meeting on the calendar and sends the invites. Everything the recipient has to do stays inside the email they were already reading. That is a smaller claim than most agent launches make, and a more testable one.
Two conditions have to be true before anything happens at all: the address has to sit on the CC line, and the thread has to show that a time is genuinely being arranged. Where either is missing, Lindy stays out of the conversation. This article covers what shipped, how it differs from the template that already existed, why the last company to try the CC-a-bot model is now a footnote, and what a business should check before handing an inbox agent a calendar. The underlying large language model is the least interesting part of the story; the distribution model is the interesting part.
Table of contents
- What Lindy Actually Shipped on the CC Line
- Why CC-Based Scheduling Is Harder Than a Booking Link
- From Template to Behaviour: What Lindy Removed
- The Scheduling Problem Lindy Is Aiming At
- The Ghost of Amy Ingram: Why CC-a-Bot Failed Before
- Where Lindy Sits Among Calendly, Motion and Inbox Agents
- What Lindy Runs On: Surfaces, Memory and Approvals
- What Lindy Costs and What Your Security Review Will Ask
- The Risks to Weigh Before You Add Lindy to a Thread
- How to Pilot Lindy CC Scheduling Without Betting the Calendar
- Lindy CC Scheduling: Questions People Are Asking
- References and Further Reading
What Lindy Actually Shipped on the CC Line
The release is narrow by design, and the narrowness is the point. Nothing here asks the other party to change how they work.
The Lindy address is the whole interface
There is no new surface to learn. Adding the Lindy address to an existing thread is the entire instruction, and Lindy infers the rest from the messages already sitting above it. That is a different interaction model from a booking page, which asks the recipient to leave the conversation and pick from a grid.
Two Lindy behaviours, one trigger
The same trigger produces two outcomes depending on the state of the thread. Where a time has already been agreed between the parties, Lindy simply books it and sends the invite. Where nothing has been agreed, it replies once with three open slots rendered as clickable links, and the first click wins.
The reply-once discipline
That word “once” is doing a lot of work. An assistant that keeps chiming in on a thread is worse than no assistant, because every message it sends is another interruption for two people instead of one. Constraining Lindy to a single reply is a product decision about restraint rather than capability.
Plain-language constraints are honoured
Conditions written in ordinary prose in the thread are read and respected. A line asking for mornings only, or a thirty-minute call rather than an hour, carries through without anyone opening a settings screen. This is the part that a rules-based scheduler could never do, and it is where the model earns its place.
Nothing happens without both conditions
The guardrail is deliberately dull. The Lindy address must be on CC, and the thread must show a time is actually being arranged. A thread that merely mentions a date in passing does not trigger a booking, which matters more than it sounds when Lindy has write access to your calendar.
| Detail | What shipped |
|---|---|
| Trigger | Assistant address added to the CC line of an email thread |
| Second condition | Thread must show a meeting time is being arranged |
| If a time is agreed | Books it, sends invites, no reply needed |
| If no time is agreed | One reply offering three open slots as clickable links |
| Recipient requirement | None — no account, no scheduling page, no app |
| Constraints | Read from plain language in the thread (duration, time of day) |
| Setup required | None — replaces a template that had to be assembled |
| Announced | 3 September 2026 |
Why CC-Based Scheduling Is Harder Than a Booking Link
A scheduling link is a solved problem and has been for a decade. Reading a half-finished human negotiation and finishing it correctly is not, which is why this feature is worth more attention than a changelog line suggests.
Intent has to be inferred, not declared
Nobody writes “SCHEDULE: 30 minutes, Tuesday or Wednesday” in an email. They write “any chance you’re free later in the week for a quick catch-up?” Lindy has to decide whether that sentence is a scheduling request at all, and it has to be willing to decide that it is not.
Missing details need defaults, not questions
Most threads never state a duration, a timezone or a preferred window. An agent that asks about each one has simply moved the back-and-forth from two humans to three participants. Filling the gaps with defaults is what makes the feature useful, and it is also what makes it risky when the defaults are wrong.
The other side is a stranger to the system
The recipient of that three-slot reply has no relationship with Lindy and no reason to acquire one. Anything that asks them to sign up, install something or trust an unfamiliar domain will fail at exactly the moment the product needs to work. Keeping the whole interaction inside the email is the only viable design.
Getting it wrong is expensive in a way search is not
A bad search result costs a click. A wrongly booked meeting costs two calendars, an apology and a small amount of professional credibility. That asymmetry is the reason this category has stalled repeatedly, and it explains why Lindy holds anything with outside consequences behind an approval step.
Why this is an agent problem, not a calendar problem
Calendar availability is arithmetic. Deciding that a sentence is a booking request, extracting a constraint from a subordinate clause, and choosing three slots that a specific stranger will plausibly accept is judgement. That is the gap between workflow automation as it has existed for years and what agentic products are now attempting.
From Template to Behaviour: What Lindy Removed
Scheduling by CC was already possible on this platform. What changed is that it no longer has to be built, and the distance between those two states is the actual news.
The build that used to be required
The published template walked the user through copying a dedicated mail address out of an email trigger, setting a condition that decides whether a time was proposed or still needs searching, configuring a calendar action with working hours and daily meeting caps, and writing the reply text Lindy sends back. Four configuration steps, each with its own way of being wrong.
Configuration is a tax most buyers never pay
Every one of those steps is a place a pilot dies. Teams buy an automation platform, get through the demo, then discover the value is gated behind an afternoon of building. The templates exist precisely because the building is the barrier, and templates only shift that barrier rather than removing it.
What “no build” actually changes
Moving scheduling from something configured in advance to behaviour that runs on the thread itself changes who can adopt it. A configured workflow belongs to whoever built it. A behaviour belongs to anyone who knows an email address, which is a materially larger group inside a company.
The trade you make in exchange
Defaults you did not choose are defaults you cannot inspect. The template version let you set working hours and daily meeting caps explicitly; the zero-build version decides for you until you tell it otherwise. For most users that is a good trade. For anyone whose calendar has hard rules, it is a reason to read the settings before adding the address to a live client thread.
| Aspect | Template workflow (before) | CC behaviour (now) |
|---|---|---|
| Setup | Four configuration steps | None |
| Address | Copied out of an email trigger | Already available |
| Branching logic | Condition built by the user | Built in |
| Working hours and caps | Set explicitly during build | Defaulted unless changed |
| Reply wording | Written by the user | Generated |
| Who can use it | Whoever built the workflow | Anyone with the address |
The Scheduling Problem Lindy Is Aiming At
The case for Lindy on the CC line does not rest on how clever the agent is. It rests on how much unstructured coordination is currently sitting in inboxes, and Microsoft has measured that in some detail.
The measured shape of the working day
Microsoft’s 2025 Work Trend Index special report, “Breaking down the infinite workday”, drew on Microsoft 365 signals plus a survey of 31,000 knowledge workers across 31 markets, run by Edelman Data x Intelligence between 6 February and 24 March 2025. The average employee receives 117 emails and 153 chat messages in a workday.
Interrupted every two minutes
The same research found workers are interrupted roughly every two minutes during core hours — around 275 times a day — by meetings, emails and messages. Coordination overhead is not a rounding error in that number; scheduling threads are among the most reliably fragmenting messages anyone receives.
Most meetings are not planned
Sixty per cent of meetings are ad hoc, and one in ten is booked at the last minute. Meetings after 8 p.m. rose 16% year on year. Those are the meetings that get arranged over email rather than through a recurring invite, which is exactly the population an inbox-resident assistant can reach.
What the numbers do and do not prove
They do not prove that an assistant fixes any of it. What they establish is that the volume is real, that most of it is unstructured, and that a tool which only works when both parties already use the same platform can reach very little of it. That is the argument for meeting people on the CC line rather than on a booking page.
The honest counter-argument
Some of those 275 interruptions exist because coordination is genuinely hard, and some exist because arranging a meeting is a socially acceptable way to avoid deciding something. Automating the first kind saves time. Automating the second kind just makes it cheaper to hold meetings that should not happen, and no vendor’s demo shows you that failure mode.
The Ghost of Amy Ingram: Why CC-a-Bot Failed Before
This idea has a history, and it is not a happy one. Anyone evaluating Lindy should know what happened to the company that got here first.
x.ai and the Ingrams
x.ai — the 2014 company of that name, no relation to the later xAI that SpaceX now owns — was founded by Dennis Mortensen and built “Amy Ingram” and “Andrew Ingram”, assistants you copied on an email to hand off scheduling. The pitch was almost word for word the pitch being made today, and for a while it was one of the most visible autonomous agent products in the market.
What it could not do
It ran before natural language processing was good enough to carry the load, so it leaned on narrow parsing plus a human review layer behind the curtain. Anything phrased unusually broke it, and the cost of the humans quietly patching the gaps scaled with usage rather than shrinking. The economics never resolved.
How it ended
Bizzabo announced the acquisition on 3 June 2021. Eight weeks later, on 28 July, Mortensen emailed customers that the standalone scheduling tool was being sunset, with 31 October 2021 given as the final day of operation. The founding team went to events software; the CC-a-bot pattern went dormant.
What is actually different about Lindy now
Three things. Language understanding is no longer the bottleneck, so unusual phrasing is a normal case rather than a failure. Lindy is not a single-purpose product, so scheduling does not have to carry the entire subscription on its own. And Lindy already sits in the inbox for other reasons, so the CC address is a marginal addition rather than a new vendor relationship.
What is not different
The recipient still has to trust an unfamiliar sender, a wrong booking still costs more than a wrong search result, and the feature still has to survive contact with people who reply “any of those work” to a list of three links. History says the hard part was never the parsing.
Where Lindy Sits Among Calendly, Motion and Inbox Agents
There are three established ways to make scheduling less painful, and the CC model is a fourth that behaves quite differently from all of them.
Link-first schedulers
Calendly and its imitators solved the problem by moving it out of email entirely. They are reliable and widely understood, and the cost is that one party does the work of choosing while the other publishes availability. They work best when there is a clear host and guest.
Optimising schedulers
Tools in the Motion and Reclaim family take an existing calendar and rearrange it around priorities and focus time. They are strongest on the internal calendar of someone already overloaded, and they have very little to say about a negotiation happening in a thread with an external party.
Inbox and admin agents
This is the crowded category, and it is where Lindy competes most directly. We covered a near neighbour recently when Catch launched an AI admin assistant for calendar, email and phone calls — same problem space, different entry point. The distinguishing question across the category is what the counterparty has to do.
The distribution argument
CC is a distribution channel, not just an interface. Every thread the address is added to introduces Lindy to people who have never heard of the vendor, in a context where it either works visibly or fails visibly. That is a sharper feedback loop than a dashboard, and a considerably cheaper one than paid acquisition.
| Approach | Where it lives | What the other party must do | Best fit |
|---|---|---|---|
| Booking link | A web page you send | Leave the thread and pick a slot | Clear host and guest |
| Calendar optimiser | Your own calendar | Nothing — it is internal | Protecting focus time |
| Shared-platform assistant | Inside one vendor’s suite | Use the same platform | Internal meetings |
| CC-line agent | The thread you already have | Read one email, click one link | External, unstructured threads |
What Lindy Runs On: Surfaces, Memory and Approvals
Scheduling is one behaviour inside a broader assistant, and the surrounding architecture is what a buyer is actually evaluating.
The surfaces it already occupies
It works in Gmail through a Chrome extension, answers in Slack channels and direct messages, runs from iMessage, and joins calls on Google Meet, Zoom and Microsoft Teams to record them and file notes into folders described in plain language. Up to five inboxes can be connected on a seat.
Integrations and MCP
Lindy connects to Gmail, Drive, Calendar and Notion, plus more than a thousand other integrations, and it supports MCP servers so an internal tool can be attached without waiting for a native connector. Model choice is left to the user on a per-task basis rather than fixed by the vendor.
Memory you can open and edit
Workspace context is kept in plain files that can be opened and edited. That is an unusual and welcome design choice: an assistant whose memory is a readable file is one whose mistakes can be traced to a line rather than attributed to a black box. It also means a bad fact can be deleted rather than argued with.
Approvals as the default guardrail
Anything carrying outside consequences — sending an email, updating a ticket, posting to a channel, publishing a document — waits on approval from a named person. Read-only lookups from approved sources do not. The same guardrail governs the scheduling flow, which is the honest answer to “what if it books the wrong thing”.
Forty-plus skills and shared context
Lindy ships with more than forty built-in skills covering research, analysis, decks and dashboards, and lets teams create their own. For an organisation already investing in intelligent automation, the scheduling behaviour is the cheapest possible entry point into that surface area.
What Lindy Costs and What Your Security Review Will Ask
Pricing is per seat with a shared credit pool, which is a meaningfully different model from per-task billing and worth understanding before a pilot.
The published Lindy plans
Plus is $29.99 per user per month with 3,000 credits. Pro is $99.99 with 15,000 credits, roughly five times Plus. Max is $199.99 with 35,000 credits, around twelve times Plus. Enterprise is quoted and adds the controls a procurement team asks about.
| Plan | Price per user / month | Credits per user / month | Adds |
|---|---|---|---|
| Plus | $29.99 | 3,000 | Standard usage |
| Pro | $99.99 | 15,000 | 5x the Plus allowance |
| Max | $199.99 | 35,000 | ~12x the Plus allowance |
| Enterprise | Quoted | Shared pool plus bonus | HIPAA and signed BAA, audit logs, SSO, dedicated support |
How credits are consumed
Everyday asks — an answer, a summary, a lookup, closing out a meeting, drafting a reply — cost between 2 and 250 credits. Deep work such as researching a competitor and writing the report runs 250 to 1,000. Big builds like a dashboard or an end-to-end internal tool run 1,000 to 2,500. A CC scheduling reply sits at the cheap end of that range.
Seats, pooling and the absence of surprise bills
Every seat contributes its allowance to one shared pool the whole workspace draws from, and credits refresh each cycle without rolling over. If the pool runs low, credit-using actions pause and admins are told, rather than an invoice arriving. Anyone who uses Lindy occupies a seat, including someone who only ever mentions it in Slack.
The Lindy compliance posture
SOC 2 Type II certification is reported, audited by the Johanson Group, alongside GDPR and PIPEDA compliance, with HIPAA and a signed business associate agreement available on the enterprise tier. Data is stated as encrypted, never sold, and not used to train models by Lindy or its providers. SSO and audit logs are enterprise-tier controls.
What to ask that the marketing page will not answer
Where the mail is processed and for how long it is retained. Whether a thread Lindy was CC’d on is stored in full or only the extracted fields. Whether the approval log records the content of the message that triggered an action. Those three questions determine whether this passes a data protection impact assessment, not the certification badges.
The Risks to Weigh Before You Add Lindy to a Thread
None of what follows is a reason to avoid the product. All of it is a reason to pilot it deliberately rather than rolling it out by word of mouth.
An agent on the CC line reads everything above it
To find the scheduling details, Lindy reads the thread — including whatever was forwarded into it three replies ago. People CC things without rereading them. That is a data handling question before it is a security question, and it applies every time somebody adds the address to a live client conversation.
Instructions can arrive from outside your organisation
An email thread is untrusted input written partly by people you do not employ. Any agent that acts on the contents of a message is exposed to text crafted to steer it, which is why we have covered how AI agents can pass authentication and still drift or leak data. Approval gates narrow the blast radius; they do not remove the exposure.
Your calendar is a scarce resource with no undo
A booking link fails safe: nobody picks a slot, and you notice. An agent that books fails loudly, on somebody else’s calendar, with an invite already sent. Working hours and daily meeting caps exist for this reason, and they are worth setting explicitly before the first external thread rather than after the first mistake.
Availability is a form of disclosure
Three open slots reveal more than most people intend — that you are free on a Friday afternoon, that your week is empty, that you are not where you said you would be. It is minor in most contexts and material in a few, particularly in sales and in any regulated advisory relationship.
The failure mode nobody demos
The interesting failures are not “it booked the wrong day”. They are the thread where two colleagues were negotiating politically around who should attend, and an agent resolved it by booking. Coordination friction sometimes carries information, and an assistant optimising it away cannot tell the difference.
How to Pilot Lindy CC Scheduling Without Betting the Calendar
A pilot that produces a decision looks quite different from a pilot that produces enthusiasm. Four constraints are enough.
Give Lindy one meeting type and one person
Pick a recurring, low-stakes category — an internal catch-up, a supplier check-in — and give it to one person for two weeks. A single owner who can describe every failure in detail is worth more than ten users who report that it was “mostly fine”.
Write the constraints down before you start
Working hours, minimum notice, maximum meetings per day, which domains are in scope. If those are not written down, you will not be able to say afterwards whether Lindy broke a rule or simply revealed that you never had one. This is the same discipline that makes any business process automation project auditable.
Measure the right thing
Not hours saved, which nobody can substantiate. Count the number of round-trip emails required to book a meeting, before and after, and the number of bookings that had to be corrected. Two numbers, both countable, both arguable in a review meeting.
Decide who owns the mistakes
Somebody has to be accountable when a client gets a wrong invite, and it should be decided before it happens rather than during. In practice this belongs with whoever owns the rest of your workflow automation, not with the individual who happened to add the address to a thread.
Where this fits a wider automation programme
Scheduling is a good first agent precisely because it is small, visible and reversible. If it works, it is a low-risk proof that autonomous AI agents can hold a real responsibility inside the business, and a sensible on-ramp to the intelligent automation work that follows. If it fails, it fails cheaply and in public, which is exactly what you want from a first attempt.
Connect it to the processes you already run
The value only compounds when Lindy sits alongside the systems that do the rest of the work. Teams that already run workflow automation will get more out of a CC agent than teams introducing both at once, because the surrounding process is already documented and the exceptions are already known.
Lindy CC Scheduling: Questions People Are Asking
Does the other person need a Lindy account?
No. That is the central design claim. The recipient reads an email and clicks one of three links, with no account, app or scheduling page required on their side.
What happens if a Lindy thread is not about scheduling?
Nothing. Two conditions must both hold — the address on the CC line, and a thread that shows a time is genuinely being arranged. Where either is missing, Lindy stays out of the conversation.
Can Lindy handle “mornings only” or “thirty minutes”?
Yes. Constraints written in ordinary language in the thread are read and respected, without anyone opening a settings screen. That is the capability a rules-based scheduler cannot match.
Was this possible before?
Yes, as a workflow you had to assemble first — copying the mail address out of an email trigger, building a condition, configuring a calendar action with working hours and caps, and writing the reply text. The release reaches the same outcome with none of that build.
How much does it cost to run?
Plans start at $29.99 per user per month for 3,000 credits. A scheduling reply falls into the everyday-ask band of 2 to 250 credits, so the scheduling behaviour itself is not what will consume an allowance.
Is it safe to point at client email?
It depends on answers the marketing page does not give: mail processing location, retention, whether full threads or only extracted fields are stored, and what the approval log records. Ask those before pointing it at anything regulated.
How is this different from Calendly?
A booking link asks the other party to leave the conversation and choose from a grid. A CC agent finishes the conversation where it already is. The two are complementary more than competitive — one suits a clear host and guest, the other suits an unstructured negotiation.
Has anyone tried this before?
x.ai ran “Amy Ingram” from 2014 on the same CC-a-bot premise, was acquired by Bizzabo in June 2021, and shut the standalone scheduling tool down that October. The premise was right and the technology was not ready. That is the bet being re-run.
Who is behind Lindy?
Flo Crivello, previously a product manager at Uber and founder of the virtual office company Teamflow, founded Lindy in 2023. It has since raised roughly $50 million according to public funding trackers, and is based in San Francisco.
References and Further Reading
Lindy can now schedule meetings from email thread — TestingCatalog
Breaking down the infinite workday — Microsoft Work Trend Index
Bizzabo Acquires x.ai to Launch AI-powered Scheduling — GlobeNewswire
Announcing Our Biggest Step in Reimagining How People Meet and Connect — Bizzabo
x.ai’s Scheduling Tool is Shutting Down after Bizzabo Acquisition
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.