Sara & Nik’s Podcast:
Tommy Delorme and Ismael Girard

About this conversation

Tommy Delorme and Ismael Girard, two consultants based in Quebec, join Sara & Nik Podcast to discuss how they understand semantic SEO and how they use AI in their work without sacrificing quality.

The discussion covers several points, including their SEO journey, which took them from copywriting jobs to managing the SEO campaign of 500-employee-strong businesses.

The Consultancy Partnership Turned SEO Business

Tommy comes from a 10-year-long SEO background, while Ismael started his career as a copywriter for an insurance company and moved into the SEO field in 2020. They only recently partnered up to open a consultancy business, which they operate themselves.

They have no project manager and deal with the clients themselves, which helped them get more accustomed to how SEO and marketing works. That being said, they met their fair share of hurdles along the way. Pricing was one of them.

As Tommy explains, they weren’t satisfied with fixed-price packages, because websites and projects are all different, which meant that they had to sometimes “go custom”. This led to them finding a workaround in the form of creating a block-of-hours package deal. Simply put, clients would order an hour package with 23 hours, for instance, which would cover the entire workload.

This allowed them to focus on quality above all else, which helped them move to more high-profile clients, typically accounting, law, and insurance businesses with 50-500 employees.

AI Automation and Producing Quality Without Cutting Corners

It didn’t take long for Tommy and Ismael to realize the value of AI automation in the SEO field. Using n8n, Make, and Moonlit, organized in Notion, allowed them to speed up the SEO optimization process by automating repetitive workloads like SERP scraping, QA, content briefs, and keyword data gathering.

However, they soon realized two critical things:

First is that the human component was very much necessary, because AI can’t replace genuine human expertise; it’s just a tool that the professional uses to get the job done. All AI does is to speed up the process, when used correctly.

Second is the allure of speed, which can quickly become a double-edged sword. Fortunately, Ismail admits that they realized pretty fast that they don’t want to sacrifice quality for speed. From that point on, their work philosophy revolved around the question:

"How far can we push automations and AI without sacrificing the quality of a manual output?"

While the exact ceiling differs between projects, they knew there was a limit. A 10-hour project could drop to 6 hours with some AI involvement, but any lower than that and quality would suffer.

Sara then moved the conversation to keyword research, which prompted Tommy and Ismail to walk the guests through their internal process.

First there’s the client interview for information gathering and to lay out the strategy. Then comes the psychographic research, which looks at the most common pain points the audience deals with. Those pain points turn into 30 to 100-or-more seed terms, which go into DataForSEO for related searches, autocomplete, and keyword ideas. The goal is to create a larger keyword universe with relevant entries.

Then they cluster based on shared search results, which Tommy has a few words on:

“SERP-based clustering should be the only way to cluster”.

As he explains, the reason lies in the intent. “What is SEO” and “SEO consultant” may share a word there, but the constructions serve entirely different intents.

Why GEO is Mostly Semantic SEO

As Tommy explains, search engines want to understand the content as fast and with as little effort as possible.

“Everyone knows machines are dumb and cheap”.

This led to them creating their own tool, which they originally developed as a Python script in Google Colab, and which relies on Google’s Natural Language API to figure out which entity Google sees as important on any given page. They performed a live demo to showcase how the tool works, during which they showed how two sentences with the same meaning produced very different results.

Ismail explains that:

“The way it works is not about occurrence but more about placement. (...) It doesn't make sense that adding a keyword five times within a text randomly will improve your SEO”.

In his view, this beats standard SEO optimization which typically relied on keyword frequency.

The discussion then moves to GEO, which prompted Ismail to admit that, so far, AI’s been very good for business because nothing sells better than fear.

The way he sees it, the 2026 FOMO wave made it so every client now wants a GEO plan, which means all SEO professionals should learn how GEO works.

This should come relatively naturally, Tommy believes, because GEO is essentially semantic SEO. His argument included a test during which 10 guest posts called someone the best SEO consultant in a given city, which eventually influenced LLM answers in no more than two months.

As for what the future holds, Tommy and Ismail are now working to build an internal operating system to improve their process centralization.