Picture this (an illustrative example): Mindaugas runs a three-person sales team at an IT consulting company in Kaunas. He gets about 160 new leads a month, but the team actually calls only 45 of them: the ones someone happened to notice or still remembers from last week. The other 115 slip through: nobody has time to read them all, and there is no system that says which ones to work on first. Every Monday morning the team spends 40–60 minutes going through the CRM, trying to remember who on the waiting lists has not been contacted yet.
According to Salesforce's “State of Sales” research (2023, a survey of more than 7,700 sales professionals), salespeople spend only about 28% of their work week actually selling (source). The rest goes to deal administration, data entry, reports and other non-selling tasks. Sales automation with AI changes that equation: a lead scoring algorithm separates warm leads from cold ones automatically, and follow-up sequences send a contextual email a minute after the demo, without the salesperson having to remember it the same morning and write it from scratch. In this article I explain how these technologies work in practice in the Lithuanian B2B market, which tools you need and what you can realistically expect in the first four months.
Why about 70% of a salesperson's time goes to things other than selling
If a salesperson sells about 28% of the time, that is roughly 2–2.5 hours of an 8-hour working day. Where do the other 5–6 hours go? The most common time sinks: updating CRM records after calls, preparing reports for management, planning follow-ups (who to write to, what to write, in what order), and manually sorting and prioritizing leads. Each of these often eats anywhere from a quarter of an hour to an hour a day.
The math is simple: a team of 5 salespeople × 90 minutes of admin work a day × 22 working days = 165 hours a month spent on things other than selling. That is almost a full-time employee. Sales automation with AI goes after exactly these losses first: lead scoring removes the time spent sorting by hand, CRM webhooks create records automatically after a call, and follow-up sequences mean the salesperson no longer starts every message from a blank page.
How AI lead scoring works in practice
Lead scoring means automatically gauging how “hot” each lead is, based on many signals in real time. Not a 1–10 score in the CRM that a salesperson assigns by gut feeling, but an AI algorithm that weighs 15–30 factors and gives every contact a probability score automatically.
Signals the AI analyzes
Firmographics: company size and revenue via the Rekvizitai.lt API (a Lithuanian company register directory) or Apollo.io, NACE activity code, company age, registered location. For IT consulting services, the ideal client is a private limited company (UAB) with 20–100 employees, NACE code 62.xx, in business for 3–15 years.
Behavioral signals: which website pages were visited and for how long (pricing page +20 points, service description +10, blog +5), email open rate, and reply speed: a lead who replies quickly is usually far more interested than one who replies after a few days.
Decision-maker's title: a CEO or head of sales contact gets +25 points. An accountant or administrator gets +5. This reflects whether you are talking to someone who can approve the purchase or to an intermediary gathering information for their manager.
How the scoring system works
All signals are combined in an n8n workflow: data is collected from Pipedrive, website analytics, the email tool and Rekvizitai.lt. An OpenAI model trained on your historical deal data gives every lead a score from 0 to 100. The result is split: the top 30% of leads (75–100 points) are automatically assigned to a senior salesperson with a “hot” tag and a “call today” task. The middle 40% go into a nurturing email sequence. Those below 30% stay under monitoring until their situation changes. With a well-calibrated model, the “hot” leads it selects usually convert noticeably better than unprioritized lists.
Automating follow-up sequences: when AI writes better than a salesperson
Follow-up emails after a demo are one of the biggest time drains for sales teams. After the meeting the salesperson sits down and writes: recaps the key points discussed, shares useful resources, closes open technical questions. This takes 20–40 minutes per email and often slips to the next day, so the moment when the client is still “hot” is lost.
A contextual follow-up after the demo
The system connects to Calendly (for booking demos) and Pipedrive (the demo notes field). When the salesperson finishes the demo and types a few words in the CRM (“module discussed, main question, open issue”), n8n triggers OpenAI, which generates a personalized follow-up email with specific details. The salesperson approves or edits it in 30 seconds, and the email goes out a minute after the meeting. Speed matters: according to Harvard Business Review (2011, “The Short Life of Online Sales Leads”), companies that contacted a potential customer within an hour of the inquiry were nearly 7 times more likely to have a meaningful conversation with a decision-maker than those that contacted them later (source).
- 1Demo is bookedCalendly
- 2Salesperson types a few wordsPipedrive
- 3AI writes a personal emailn8n + OpenAI
- 4Salesperson approves in 30 sFor the first two weeks, every email.
- 5Email is sent and trackedLemlist
Answering informational questions
When a lead sends a technical or informational question by email, the n8n flow checks the type of question. If the answer is in the knowledge base (prices, list of integrations, technical specifications), the AI drafts a reply and the salesperson approves it. If the question involves deal negotiations or special terms, it is passed straight to the salesperson with an alert. This logic lets you respond to technical questions within 10–15 minutes without burdening the salesperson with digging up details.
A re-engagement email after 30 days
Leads that have been inactive for 30 days go into a re-engagement sequence. The AI checks public company information (LinkedIn, changes on Rekvizitai.lt, published news) and generates a personalized re-engagement email with a new value proposition. Not a templated “still interested?”, but “I see your company has grown its team to 45 people, which is exactly the growth trigger you mentioned.” Results of re-engagement sequences depend heavily on the quality of your database, so it is worth measuring them from the first week.
A combined stack for Lithuanian B2B: Pipedrive + n8n + OpenAI API + Lemlist
A sales AI system is not a single tool. It is an ecosystem of four components tied together by n8n orchestration:
Pipedrive: the sales CRM and database
Stores the full contact history, the pipeline and deal stages. n8n reads from and writes to Pipedrive via the API. Pipedrive has its own AI features, but they do not connect to Lithuanian firmographic sources, so it is better to build custom scoring in n8n with Rekvizitai.lt data.
n8n: orchestration and automation logic
Connects all the tools: pulls firmographics from Rekvizitai.lt and website behavior data from analytics, triggers OpenAI before sending, sends through Lemlist and updates Pipedrive. Self-hosted on a VPS: from about €10–15/month (prices change). More on n8n automation in Lithuania with typical use cases.
OpenAI API: the lead scoring model and copywriting
A separate agent for each function via the OpenAI Responses API (OpenAI shut down the Assistants API on August 26, 2026): lead scoring “talks” to your historical sales data (RAG or fine-tuning), and follow-up copy follows your brand voice. Token costs are roughly €80–150/month at moderate volume (depending on the model and volume).
Lemlist: sending and tracking email sequences
Email sequences with personalized fields, delivery and open tracking, automatic reply detection. In this stack the AI copywriting is done by OpenAI, and Lemlist is used as the sending platform with analytics. Plan for email sequences: from about $70–100/month per user (prices change).
Rekvizitai.lt API or Apollo.io: firmographics for Lithuania
Rekvizitai.lt collects publicly available company information (size, revenue, CEO, line of business), an essential ingredient of lead scoring in the Lithuanian market. For international leads, use Apollo.io. Cost: from about €50–80/month depending on the API quota (prices change).
Illustrative example: how an IT consulting company could close more deals in 4 months
Say the starting point is this: 3 salespeople, 150 leads a month, 30% of leads do not respond to the first contact, and the deal conversion rate is 8%. The team does not know which leads to prioritize, so they call in the order the leads were entered in the CRM or based on a salesperson's “feel”.
What typically happens in the first month:
- A lead scoring model is built on Pipedrive history + Rekvizitai.lt firmographics + website analytics. The model is trained on, for example, 18 months of sales history (a few hundred deals).
- Follow-up sequences: a contextual email within 30 minutes of the demo, re-engagement after 7 days, and re-engagement after 30 days with a new value proposition based on public company information.
- n8n orchestration: Calendly → Pipedrive → OpenAI → Lemlist, at first with the salesperson's approval and, after calibration, automatically.
Expected results after 4 months (a hypothetical scenario, not a specific client's results), compared with the same quarter of the previous period:
An important note: for the first two weeks the team checks every AI-generated email before it is sent. This is the essential calibration phase, and the model improves from the salespeople's corrections. In this example, from the third month most emails go out without manual editing: the salesperson just reads and approves them.
BeforeAfter 4 months
First email to the client after the demo
Top 30% lead conversion
Figures from the table above (4 days = 96 hours), not a specific client’s results.
How much does it cost to launch in 30 days?
A sales AI system is not an expensive investment compared with hiring more salespeople, but it is important to understand clearly where the cost comes from:
One-time costs
Monthly costs
The ROI logic: a team of 5 salespeople with an average deal of €3,000. If the system helps close 2 extra deals a month (a 22% increase on a baseline of about 9 deals/month), that is €6,000 in extra revenue vs. about €345/month in platform costs. In this hypothetical calculation the investment could pay for itself within the first few months, but the actual payback period depends on your margin and the real increase in deals.
Mistakes worth avoiding
Sales AI projects fail more often because of organizational mistakes in the rollout than because of the technology. Here are the four most common:
The AI writes too generically
The most common mistake is an OpenAI prompt without context. “Write a follow-up after a sales demo” produces a templated email the client recognizes from the first sentences. “Write a follow-up for an IT consulting service; the call covered a Dynamics 365 rollout for 25 seats, and the client had a question about SAP integration via the API” produces a personalized email with specific value. Prompt engineering and passing context account for most of the success.
Lead scoring without real-world testing
A scoring model built without analyzing historical sales, or not tested with the actual team, can prioritize the wrong leads. A typical example: the model rates “large” companies (200+ employees) highly, but your ideal client is the 20–50 employee segment, because large companies often only want pilot projects. The first month must be spent on continuous validation: do the top 30% of leads actually convert?
Automating without human approval in the first 2 weeks
An untested flow can send the wrong email to the wrong client. For the first two weeks, every AI-generated email must go through an approval step by a salesperson, not out automatically. This is not a weakness; it is the calibration phase. The system is refined based on the salesperson's corrections, and after calibration a quick read is usually enough for most emails.
Ignoring GDPR requirements in automated sequences
Automated marketing sequences must comply with Article 81 of the Lithuanian Law on Electronic Communications and the GDPR. Since April 22, 2026, direct marketing emails to legal entities (including work addresses such as firstname.lastname@company.lt) can be sent without prior consent, but every email must include a clear, free and easy way to opt out, and sending must stop immediately once someone opts out (explanation by the VDAI, the Lithuanian data protection authority, May 19, 2026). For individuals (for example, private addresses) you still need prior consent, except when you offer your own customers similar goods or services. Replying to an inbound lead who filled in a form themselves is fine. An n8n flow can automatically check each lead's source and type, insert an unsubscribe link in every email and stop the sequence when someone opts out.
The Aigentas sales AI package: audit → setup → training
Aigentas.lt builds sales automation systems for B2B companies in Lithuania. I do not implement “AI in general”; I set up specific sales flows with measurable metrics and a payback calculation before any investment.
Initial discussion (free). You write where your leads come from, how many of them turn into deals and where your salespeople lose the most time. I read it and reply by email with where I see the biggest potential; if we need to look at the pipeline together, we arrange a call. We agree on scope and price before you decide on the investment.
Setup package (from €2,500). A first, simpler solution (for example, one follow-up sequence) starts from €450. The full package includes building a lead scoring model on your historical sales analytics, automating three main follow-up sequences (after the demo, re-engagement after 7 days, personalized re-engagement after 30 days), the n8n + Pipedrive + OpenAI + Lemlist integration and two weeks of calibration support. Setup takes 30 days. More on building AI agents for sales and business automation.
Training and support (from €150/month). A four-hour training session for salespeople: how to interpret the scoring recommendations, how to edit AI emails, and when to override the system based on the relationship. Monthly support: model updates, new sequences, analytics reviews. More on business process automation with n8n in Lithuania.