AI Agent Consultant & Freelancer

Felix Erkinheimo

Building AI agents that do real work for your business

I have spent a decade scaling and leading sales and marketing teams. Now I build AI agents and workflows that actually move the pipeline.

Past projects

AI agents I've built

Seidat

Sales database and market signals

Seidat set out to find international growth in the hotel industry. Virtual tours are part of Seidat's offering to hotels. First a company database was needed for sales. Then information was needed on whether a company already has a virtual tour, in order to personalize the sales outreach. Using Clay and the OpenAI API, an AI agent was built that finds all potential customers in each target country and checks for each one whether information about an existing virtual tour can be found online. This was done by searching the company's website for keywords, such as the names of virtual tour providers.

Background

What

Using Clay and the OpenAI API, an AI agent was built that finds all potential customers in each target country and checks for each one whether information about an existing virtual tour can be found online. This was done by searching the company's website for keywords, such as the names of virtual tour providers.

Result

The company database grew by 6000+ companies. Sales can focus resources on the companies with the greatest need for the solution, and personalize sales messaging accordingly.

ClayOpenAI APIWeb research

Discuss a similar setup

High-ticket B2B sales team

Sales outreach portal

The team sells high-ticket B2B solutions. The market is small, and mass email and automated messaging do not suit the target group. At the same time, sending emails manually is too slow. I built a custom UI element into the Attio CRM for sending sales messages. It filters people by market segment and writes a personalized message draft for the segment. The salesperson edits and reviews every message before sending.

Background

The team sells high-ticket B2B solutions. The market is small, and mass email and automated messaging do not suit the target group. At the same time, sending emails manually is too slow.

What

I built a custom UI element into the Attio CRM for sending sales messages. It filters people by market segment and writes a personalized message draft for the segment. The salesperson edits and reviews every message before sending.

Result

This ensures that all sales follow-up messages go out on time and are of high quality. That is critical for the company, because there are fewer sales cases and the average deal size is high.

AttioAttio SDK AppClaude Code

Discuss a similar setup

Futures Platform

Automated LinkedIn sequence for inbound leads

Futures Platform's sales team manually sent a LinkedIn message to every website lead by copying an AI-written message into the message field. Sending dozens of messages every week was a significant burden on the team. In addition, the manual LinkedIn messaging and its outcome left no record in the company's CRM. We built an automation with HubSpot's AI and the FirstTouch automation platform that sends inbound leads a connection request, a personalized message, and a follow-up message.

Why

What

We built an automation with HubSpot's AI and the FirstTouch automation platform that sends inbound leads a connection request, a personalized message, and a follow-up message.

Result

The time spent on manual messaging can be used on other sales work, and LinkedIn messages leave a record in HubSpot, so they can be reacted to quickly and the whole team stays up to date on the communication.

HubSpotFirstTouchLinkedIn

Discuss a similar setup

Seidat

Sales meeting analyzer

Seidat needed the pain points, objections and competitor mentions from its sales meetings in one place, to develop sales messaging and product and to target marketing. Built an n8n automation that sends each meeting transcript to the OpenAI API, which extracts the pain points, objections and competitor mentions. The analysis is saved to Google Sheets.

Background

Seidat needed the pain points, objections and competitor mentions from its sales meetings in one place, to develop sales messaging and product and to target marketing.

What

Built an n8n automation that sends each meeting transcript to the OpenAI API, which extracts the pain points, objections and competitor mentions. The analysis is saved to Google Sheets.

Result

The tool helps Seidat develop its sales messaging and product, and target marketing at customers' real needs.

n8nOpenAI APIGoogle Sheets

Discuss a similar setup

Videopalat

Email campaign builder agent

Videopalat wanted to test email as a sales channel for its short-video service. Each campaign needed a target audience and message copy built by hand. Built an AI agent that selects the target audience from the CRMs and drafts the campaign messages. The agent reads the CRM data to pick the audience and writes the message drafts, and outreach runs on Clay and Instantly.

Background

Videopalat wanted to test email as a sales channel for its short-video service. Each campaign needed a target audience and message copy built by hand.

What

Built an AI agent that selects the target audience from the CRMs and drafts the campaign messages. The agent reads the CRM data to pick the audience and writes the message drafts, and outreach runs on Clay and Instantly.

Result

The outreach project produced 11 positive conversation openings and a new customer acquisition channel for Videopalat.

CRM dataClayInstantlyLLM drafting

Discuss a similar setup

Pricing

3 ways to work together

AI and automation projects

From€500per day

Defined-scope delivery of AI agents and automations, from idea to production

Dedicated AI & Agentic Developer

From€2000per month

Continuous co-development, AI agent development, and process ownership

Fractional Head of AI

From€4000per month

Strategic AI vision, architecture oversight, and team leadership.

Let's talk about what your business needs.

Contact

Let's work together

Email me Connect on LinkedIn