Use cases by function

AI agents for SME sales: quotes, CRM and follow-ups

Roles involved
sales representatives, sales managers, bid and quote managers, sales assistants, owners who do the selling themselves
Common software
Bexio, Abacus, Proffix, HubSpot, Pipedrive, Salesforce, Microsoft Dynamics 365, Microsoft 365 (Outlook, Teams, SharePoint), Microsoft Copilot Studio, Zefix, simap.ch
25

Briefing every customer meeting in the calendar

Today

Before a meeting, the salesperson goes through the prospect’s website, the commercial register, recent e-mails and the CRM history, often in the car for lack of time.

With AI

As soon as a customer meeting appears in the Outlook calendar, the agent checks the CRM, recent e-mails, the commercial register via Zefix and the prospect’s website, then posts a one-page brief in Teams the day before: what the company does, contacts, open quotes, questions to ask. The salesperson reads it, corrects anything wrong and decides how to approach the meeting.

What to measurePreparation time per meeting, and how many briefs the salesperson considers usable without major rework.

Tools Microsoft 365 Copilot agent, ChatGPT Business or Claude with connectors, Outlook, Teams, existing CRM, Zefix

26

Targeted prospecting with first-contact drafts

Today

Prospect lists are bought in or compiled by hand, poorly qualified, and the first message is the same for everyone.

With AI

Each week, the agent matches your criteria (region, sector, size) against public signals: new entries published in the Swiss Official Gazette of Commerce, job adverts, websites. It creates or completes the records in the CRM, citing the source of each piece of information, and prepares a tailored first message for each one as a draft. The salesperson chooses whom to contact, reviews each message and sends it personally, in line with the Unfair Competition Act.

What to measureThe reply rate to first contacts, compared with your current messages over the same period.

Tools HubSpot, Pipedrive, Zefix, Swiss Official Gazette of Commerce, Outlook, Microsoft Copilot Studio or n8n

27

From meeting to CRM, without retyping

Today

Visit notes stay in a notebook or a voice memo, the CRM is only half filled in and promises made in conversation get lost.

With AI

After a Teams meeting or a site visit, the salesperson drops in their notes or a voice memo; the agent transcribes them, fills in the CRM fields (need, budget mentioned, next step, deadline), creates the follow-up tasks and drafts the summary e-mail to the customer. The salesperson approves the record and sends the summary; the agent logs every change.

What to measureThe share of meetings documented in the CRM within 24 hours, and the number of key fields left empty.

Tools HubSpot, Pipedrive, Salesforce, Microsoft Dynamics 365, Teams, Outlook, dictation app, API integration

28

Following up quotes that got no reply

Today

Quotes that have been sent go unanswered for weeks, and nobody sees which ones deserve a follow-up before the Monday sales meeting.

With AI

When a quote has had no activity beyond the period you set, the agent built into the CRM reviews the history and recent e-mails, summarises where the deal stands, drafts a follow-up message and adds the opportunity to the sales meeting agenda. The sales manager sets priorities; the salesperson adapts and sends.

What to measureThe average time between sending a quote and the first follow-up, and the number of quotes closed without any reply from the customer.

Tools Agents built into HubSpot or Salesforce, Microsoft Dynamics 365 Sales with Copilot, Outlook, Bexio

29

Technical quotes prepared as soon as an enquiry arrives

Today

An enquiry arrives with a specification, drawings or a quantity table; the quote manager copies out the requirements, looks for comparable past quotes and rewrites the same technical paragraphs.

With AI

As soon as an enquiry lands in the quotes mailbox, the agent extracts the requirements from the specification, finds comparable past quotes in the archive, checks items, availability and reference prices in the ERP, then creates the quote there as a draft with your standard texts and a list of points to clarify. The quote manager checks feasibility, quantities and prices; management approves before it is sent.

What to measureThe time between receiving an enquiry and sending the quote, and the number of corrections made to the first version.

Tools Abacus, Proffix, SAP Business One, Outlook, SharePoint quote archive, custom-built agent·Sectors Manufacturing, Logistics

30

Construction quotes prepared from tender documents

Today

For each tender, the contractor reads a specification running to dozens of pages, picks out the relevant standard bill of quantities (NPK/CAN) items and enters its prices by hand in its estimating software.

With AI

When a tender file comes in, the agent reads the specification, carries the relevant NPK/CAN items into your estimating software with your reference unit prices, flags unusual items, conditions and deadlines, and drafts the questions to send to the client. The project manager or estimator checks every item; management sets the final price.

What to measureTime spent per tender and the number of pre-filled items the estimator has to correct.

Tools Messerli, Sorba, CRB standard item catalogue (NPK/CAN), Excel, custom-built agent·Sectors Construction

31

Monitoring and preparing public tenders

Today

Tenders published on simap.ch are checked whenever someone remembers to; the ones you pursue take hours to check against the suitability criteria and to compile references.

With AI

Every day, the agent checks new notices on simap.ch against your CPV codes and criteria, retrieves the documents for relevant tenders, summarises the specification and creates a SharePoint folder with the list of required documents, the deadlines and draft answers drawn from your previous bids. Management decides whether to bid; the bid owner reviews and signs.

What to measureThe number of relevant tenders spotted in time and the time needed to prepare a complete bid.

Tools simap.ch, Microsoft Copilot Studio, SharePoint, Planner, ChatGPT Business or Claude with connectors·Sectors Construction, IT services

32

Engagement proposals and cover letters

Today

Each new proposal starts from an old Word file: pieces from another client get left in, and the text barely reflects what the prospect actually asked for.

With AI

On request, using the notes from the first meeting and your templates, the assistant writes the understanding of the brief, the proposed approach and the deliverables, in your firm’s tone. The partner or account lead sets the fees, reviews and signs.

What to measureThe time between the first meeting and sending the proposal, and the prospect’s feedback on how well the request was understood.

Tools Microsoft 365 Copilot (Word), ChatGPT Business, Claude Team, proposal templates·Sectors Agencies, Fiduciaries

In short

In a Swiss SME, sales often rests on two or three people who prospect, price, write quotes and follow up between meetings. An AI agent starts from a trigger, such as an enquiry received, a meeting in the calendar or a tender published on simap.ch, then works through the steps on its own: it reads the mailbox and the CRM, looks up items and prices in the ERP, prepares the quote as a draft and updates the customer record, logging every action. It does not set prices or discounts, it does not negotiate and it sends nothing: someone on your sales team approves every quote and every e-mail.

Points to watch

Prospect and customer data are personal data under the revised Federal Act on Data Protection (nFADP): we agree with you on where they come from, what they are used for and how long they are kept, and we select tools that do not use them to train models. An agent only gets the access its task requires, and its actions are logged. For mass e-mail advertising, the Unfair Competition Act (UCA) requires the recipient’s prior consent, correct sender details and an easy way to opt out. Prices, discounts and contractual commitments are decided and reviewed by an authorised person.

Can an AI agent send our quotes and e-mails by itself?

Technically yes, but we do not set it up that way. The agent prepares the quote in the ERP and leaves the message as a draft in the mailbox or CRM; the salesperson reads, adjusts and sends it. Its permissions are limited to the agreed actions and every step is logged, so you can check what it did. That keeps you within the Unfair Competition Act for e-mail advertising and the data protection rules on where contact data comes from, and you stay in control of what is promised to the customer.

Do we need a new CRM to use AI agents?

No. HubSpot, Pipedrive, Salesforce, Dynamics 365 and the Bexio contact module offer interfaces an agent can connect to, and some already include their own agents. We start from the tool your sales team actually uses. If the data is incomplete or scattered across Excel files, we tidy it up first, because an agent is only as useful as the CRM it reads and updates.

Can we share our price lists and quotes with an AI tool?

Yes, provided the tool’s contract excludes training models on your data, hosting is documented and access is limited to the sales team. We check these points with your IT before any configuration. The prices themselves stay in your ERP or reference file: the agent reads them to prepare the quote, it does not change them.

Choose your first use cases.

In half a day or a day, the AI Workshop sorts your ideas and keeps three, tested on your own documents.