B2B software on LinkedIn: turn inbound messages into booked demos
For a small B2B software company, LinkedIn messages are where a lot of the best leads quietly land. A founder posts about a feature, and a head of operations writes: does this work with our system, how much is it, can we see a demo. Those messages arrive while the founder is on calls, and a lead who waits two days has often booked a demo with someone else. An AI agent can reply to people who message you, answer from your product and pricing notes, and book a demo with the right person. It does not cold-message anyone. This playbook covers the note, the qualifying questions and where the line sits on security, contracts and discounts.
Why inbound LinkedIn messages matter so much
People who message a founder or a salesperson on LinkedIn have usually done some homework. They read a post, looked at the website and want a quick, human answer before they fill in a form. They are often senior and busy, and a reply that answers their actual question and offers a time to talk is exactly what they want.
The trouble is that these messages are scattered across several people's inboxes, mixed in with recruiters and sales pitches, and easy to miss. An agent on the accounts that receive them makes sure every real enquiry gets a clear reply the same day.
Inbound only: why the agent never cold-messages
This playbook is about replying, not prospecting. The agent answers people who message you first and people who are already in a conversation with you. It does not send connection requests or pitches to strangers. Automated cold outreach on LinkedIn is what the platform restricts most firmly, it can cost a founder their profile, and it rarely suits a buyer who expects a considered conversation.
What customers ask on LinkedIn
- Does it integrate with Salesforce or HubSpot?
- How much is it for a team of 25?
- Can we see a demo this week?
- Is there a trial, and what does it include?
- Where is our data stored, and do you have a security pack?
- How long does it take to get set up?
- Can you do a discount if we sign for two years?
An example conversation
This illustration shows an operations lead who messaged a founder after a post. Names, prices and details are examples.
| Who | Message |
|---|---|
| Priya | Hi Sam, saw your post about automated rota planning. We have about 60 staff across three sites. Does it work with BrightHR? |
| AI agent | Hi Priya, thanks for getting in touch. Yes, Rotaly connects to BrightHR, so staff and leave come across automatically. Teams of your size are a typical fit. |
| Priya | Good. Roughly how much would it be for 60 people? |
| AI agent | Our Team plan is £4 per staff member a month, so about £240 a month for 60 people, with set-up included. A demo would show how it would work across your three sites. Could I ask what you use for rotas today? |
| Priya | Spreadsheets, honestly. It's a mess. |
| AI agent | That's very common, and moving from spreadsheets is a big part of what Rotaly helps with. Alex on our team has demo slots on Wednesday at 10am or Thursday at 3pm. Which would suit you? |
| Priya | Thursday. |
| AI agent | Booked for Thursday at 3pm with Alex. He will send a calendar invite with the video link, and I have passed on that you are moving from spreadsheets across three sites. |
What the agent answers
- What the product does, who it is for and which tools it connects to, from your product note.
- Published prices and what each plan includes.
- How a trial or pilot works, if you offer one.
- Two or three qualifying questions, such as team size, current tools and timing.
- Booking demos into the right person's calendar and passing on the context.
- Links to your docs, case studies and public security page.
What it always hands to you
- Security questionnaires, data protection agreements and detailed data questions.
- Discounts, custom pricing, multi-year deals and anything about a contract.
- Existing customers with support questions, bugs or billing problems.
- Feature requests and roadmap questions beyond what you have published.
- Partners, investors, press and anyone asking about the company rather than the product.
- Complaints, and anyone who is unhappy with a previous conversation.
Pass the context on with every demo
The best thing an agent can do for a demo is make it shorter to prepare. Tell it to pass on what it learned, such as team size, current tools, the problem they mentioned and their timing, to the person running the demo, in one short note. Your salesperson then opens the call already knowing the buyer's situation, which is the difference between a generic tour and a useful conversation.
The LinkedIn limits that matter here
| Limit | Spaces limit per LinkedIn account |
|---|---|
| Messages | Up to 100 a day |
| Gap between messages | At least 20 seconds |
| New conversations you start | 8 an hour, 30 a day |
A normal message goes to a 1st-degree connection. Anyone else needs a connection request or an InMail.
Set it up in an afternoon
- Write a product note: who the product is for, what it does, integrations, published prices, how trials work, your qualifying questions and the topics that always go to a person. There is an example below.
- Connect the LinkedIn accounts that receive inbound messages, usually the founders' and sales team's, in Spaces. You sign in to LinkedIn through a secure link and approve the connection.
- Connect your AI agent using the steps on our agents page. Chat apps such as Claude or ChatGPT reply when you ask them in a chat, so for replies around the clock use an always-on agent such as Claude Code, Codex, OpenClaw or Hermes. Give it the note, and the calendars of the people who run demos if your agent can reach them through its own tools.
- Tell the agent how to recognise existing customers and send them to support, rather than into a sales conversation.
- Decide how it treats recruiters and sales pitches: a short polite reply, or none at all.
- Agree who handles security questionnaires and contract questions, and tell the agent to hand those chats straight to them.
- Read every conversation for the first two weeks and tighten the note where the agent was vague or went too far.
A sample product note
Keep it factual and current, especially prices and integrations. Here is the shape of one, with example details.
PRODUCT: Rotaly. Rota planning for businesses with 20 to 500 shift workers.
FOR: hospitality, care homes, retail and clinics with several sites.
CONNECTS TO: BrightHR, Xero, Sage HR, Google Calendar.
NOT YET: Workday (on the roadmap, no date to share).
PRICES (examples, published on our pricing page):
- Team: £4 per staff member a month, set-up included.
- Enterprise (over 300 staff): custom, Alex will quote.
TRIAL: 14 days on real data, with a set-up call.
QUESTIONS (one or two at a time):
- How many staff and sites?
- What do you use for rotas today?
- Is there a date you are working towards?
DEMOS: Alex (UK), Marta (EU). 30 minutes. Pass on what you learned.
ALWAYS TO A PERSON: security questionnaires, DPAs, discounts,
custom pricing, contracts, existing customers (send to support@),
roadmap dates, partners, investors, press.The LinkedIn limits in practice
On LinkedIn a normal message goes to a first-degree connection, and anyone else needs a connection request or an InMail. Inbound replies avoid both, because the buyer has already started the conversation. The agent simply answers in that chat.
Spaces paces every message and action on each connected account and keeps it inside safe limits, with lower limits while a newly connected account warms up. The limits are set by Spaces, and you cannot set lower ones of your own, so if you want the agent to reply more sparingly, tell it so in its notes.
A second example: a security question
Larger buyers often ask about security early. The agent should share what is public and pass the rest on. Names and details are examples.
| Who | Message |
|---|---|
| David | Before we go further, where do you host data, and can you complete our security questionnaire? |
| AI agent | Of course, David. Rotaly is hosted in the UK, and our public security page covers encryption, access controls and certifications: [link]. I have passed your questionnaire request to Marta, who handles security reviews, and she will reply today. |
| Marta | Hi David, Marta here. Happy to help. Could you send the questionnaire over by email to security@rotaly.example? |
The agent shared only what the company publishes and passed the detailed questions to the person responsible. It never fills in a security questionnaire or makes promises about certifications that are not written in its note.
Common mistakes software companies make
- Using the agent for cold outreach. Keep it to inbound replies and existing conversations.
- Letting it agree discounts or "special pricing" to close a demo. Those always go to a person.
- Answering roadmap questions with guesses. Share only what you have published.
- Pulling existing customers with a bug into a sales conversation instead of sending them to support.
- Forgetting to pass on the context, so the demo starts from scratch.
Who you message
In short
For a B2B software company, an AI agent on LinkedIn makes sure every inbound message gets a clear, useful reply the same day and that good leads become booked demos, with the context passed on. Security, contracts, discounts and existing customers go to the right person.
Write the note, connect the accounts that receive enquiries and read the conversations for the first two weeks. Spaces paces every action and keeps each account inside safe limits, but LinkedIn restricts automation on personal profiles and no tool can promise an account will never be limited, so keep the agent to inbound replies only.
Pricing
FAQ
Can the agent send cold outreach on LinkedIn for us?
This playbook keeps it to inbound replies only. Automated cold outreach is what LinkedIn restricts most firmly and it can cost a founder their profile. Let the agent answer people who message you, and keep outreach personal.
Which LinkedIn accounts should we connect?
The ones that receive inbound messages, usually the founders' and sales team's personal profiles. Use real, established accounts with a history of connections.
Can it share our prices?
Yes, the prices you publish and what each plan includes. Custom pricing, discounts and multi-year deals go to a person.
Can it answer security questions?
It can share what is on your public security page and pass anything detailed, such as questionnaires and data protection agreements, to the person responsible.
Can it book demos into our calendars?
Yes, if your agent can reach them through its own tools. Tell it who runs demos for which region or team size, and to pass on what it learned so the demo starts with context.
What about existing customers who message?
Tell the agent to recognise them and send them to support, or hand the chat to their account manager. They should not end up in a sales conversation.
Can I still reply from LinkedIn myself?
Yes. When you reply or take over a conversation, the agent stays out of it until you hand it back.
Should the agent say it is an AI?
Many companies do, in the first reply. Software buyers are usually comfortable with it as long as the answers are accurate and a person joins for anything important.