AI‑driven SEO revolution: the future of discoverability with Bernard Huang @Clearscope

saas.unbound is a podcast for and about founders who are working on scaling inspiring products that people love, brought to you by https://saas.group/, a serial acquirer of B2B SaaS companies.

In episode #48 of season 5, Anna Nadeina talks with Bernard, co-founder of Clearscope, an AI-driven SEO content optimization tool.

I started in SEO when growth hacking was trendy. Back then the game was: write strong content, earn links, and hope user signals do the rest. The arrival of large language models changed the field faster than any update I had seen before. Today the problem is no longer just ranking in the blue links — it is influencing the answer that an AI hands to a user.

What changed and why it matters

Two developments rewired search. First, models like ChatGPT made it trivial to generate readable content at scale. Second, search engines reacted by building their own answer engines and validation layers. The result: people increasingly expect concise, conversational answers, not lists of links. If the goal is to be the answer, your content strategy must change.

the new world … is influencing the answer

That sentence captures the shift. Instead of optimizing only for clicks, you now have to design content that AI systems will select and cite when constructing responses. That means thinking about where the models look, what they validate, and how they choose passages.

The three-layer model for AI search

Think of modern AI-driven retrieval as a stack with three layers. This framework clarifies where content helps and where it is vulnerable.

1. Training data — the corpus used to build the model. It has a cutoff date and is expensive to influence directly.
2. Validation layer — when a model needs to check recent facts it performs web searches and uses current pages to validate responses.
3. Memory and context — personalization based on prior interactions and how the model frames a specific conversation.

Targeting the validation layer is the fastest way to influence answers. That is where your up-to-date, well-structured content can be used as a factual source for models that want to avoid hallucinations.

Why AI-generated content is not the enemy

Yes, you should use AI to scale content. The reality is simple: if you are not using AI to create and refresh content, you will fall behind. AI enables you to produce the breadth of content required to show up across the many different web searches models perform when they validate answers.

That said, AI content alone is not enough. High-quality signals still matter: user engagement, topical depth, and authoritativeness determine whether a model will select and cite your page. Generating thousands of low-value articles will get weeded out by engagement signals and by platforms trying to avoid polluted training data.

Practical playbook: how to win in AI-driven discoverability

Here is a repeatable workflow that works for product teams, founders, and content owners:

1. Pick the topic you want to own. Narrow down to one core area (for example, SEO tools or how to sell a B2B SaaS business).
2. Map subtopics and synthetic prompts. For each subtopic write the questions an AI might validate: how-to prompts, “best of” prompts, edge-case queries, and industry-specific angles.
3. Collect AI web searches. Record the actual web searches models perform when answering those prompts. These searches are your new keyword list.
4. Perform entity analysis. Ensure your content contains the entities and concepts models expect. Include competitor names, product features, and context signals that align with the validation queries.
5. Refresh and prioritize pages already cited by AI. If a model already uses one of your pages to inform answers, update that page first — align its language with the validation queries and add fresh, factual passages.
6. Use AI to scale smartly. Generate drafts and outlines with AI, but add human editing, examples, and up-to-date facts before publishing.
7. Measure AI citations, not only clicks. Track when models cite your pages and use those insights to influence future content and product messaging.

Target AI web searches, not just keywords

Target AI web searches, not keywords. Traditional keyword research focused on volume and difficulty. Now you must ask: what web searches will the model run to validate this answer? Some of those queries will look normal. Others will be oddly specific or unexpectedly niche. That long tail is where you can gain influence quickly.

Example: a model answering “best SEO tools” might run searches for “free SEO tools,” “essential SEO tools,” or even industry-specific queries like “SEO tools for manufacturers.” Create content to match those exact validation searches.

Entity coverage and content chunks

Large language models often surface small passages or chunks from pages to compose answers. Make those chunks count by:

• Including the expected entities (tools, features, company names).
• Writing clear, self-contained passages that answer a single validation query.
• Ensuring freshness and facts so the model prefers your passage over older sources.

Also pay attention to the size of the context you provide internal tools. Small fragments can lose meaning. Many practitioners find that passages of roughly a minute of reading provide better coherence for AI consumption than tiny snippets.

AI citations: influence the answer, don’t chase traffic

Winning an AI citation is less about immediate click traffic and more about being the evidence an AI uses. When a model cites your page, that page is influencing the answer. Keep those pages accurate, aligned with key messages, and tuned to your brand strengths.

If an AI frequently cites your “about” page for queries on how to buy a SaaS company, make sure that page clearly communicates your positioning, differentiators, and up-to-date examples. Use citations strategically to shape the narrative models will present.

What to avoid: spammy shortcuts and the cat-and-mouse game

Many old tricks still exist in new clothes: comment spam, fake endorsements, coordinated posting to game signals. These tactics can work briefly but lead to penalties or model distrust. Expect search and AI providers to plug loopholes over time. The durable play is high-quality content that genuinely helps people and aligns with the AI validation paths.

One concise action plan

If you can do only three things this quarter, focus on:

• Identify one topic to own. Make it narrow and meaningful.
• Extract AI web searches for that topic. Build content specifically to satisfy those validation queries.
• Update the pages AI already cites. Make them factual, fresh, and aligned with your brand messaging.

Lessons learned from building for the new search

Change is constant. Products and channels have life cycles. A big win is embracing a durable framework that targets the validation layer and uses entities to guide content. A hard lesson is humility: what worked for years can be sidelined quickly by a paradigm shift.

Surround yourself with a strong team, prioritize sustainable workflows over hacks, and stay connected with peers who are testing tactics in real time. Trading notes with other practitioners accelerates learning more than trying every tactic alone.

The future of discoverability is not about gaming a single ranking factor. It is about shaping the factual context that AI systems use to answer questions. Target the validation searches, design content that contains coherent, entity-rich passages, and keep your most-cited pages accurate and persuasive. Do that consistently and you will be part of the answers people receive from the next generation of search.