Do I have to join eXp to learn at KTO?

No. Open examples and paid learning are available to agents at any brokerage. Each offer explains its own access and support terms.

Is KTO Community the same as a CGP sales team?

No. KTO is the learning and community relationship. Sponsorship with Khai at eXp and joining a CGP sales team are separate decisions.

Where should I start with AI?

Try the fictional meeting-prep example on the AI page. It shows how to supply context, keep unknowns visible, and review the output before using it.

Does the AI Partner Kit include Khai’s full AI system?

No. The Kit teaches the Foundation. The Kit course is permanent; the wider Academy library and community are included for 90 days. Later-layer stories describe Khai’s own system, not included implementation services. Check the offer for the full scope and costs.

Where should I start with commercial real estate?

Try the first-conversation example on the CRE page, use the free CRE guides, and review the CRE Transition course. Actual property suitability and transaction terms need the appropriate verification and professional input.

Where can I compare eXp’s model?

Read the eXp Model Explained page and its official source links. Use the partnership page for a conversation about Khai as a sponsor and your needs. Sponsorship does not include a CGP sales-team role or KTO products; those are separate decisions and separate purchases.

Back to Blog

AI and the Interest Graph: Instagram Reach Without Followers

By Khai Tran · September 28, 2026 · 5 min read

Instagram now decides who sees your posts by interest, not by your follower count, so a small account can reach the right buyers. Here is how the interest graph works, and how to build a simple AI content engine that keeps it fed.

An account with zero followers can outreach one with ten thousand. That is the part most agents miss when they complain the algorithm buried them. Instagram reach without followers is not a glitch or a growth hack. It is how the platform decides who sees what now, and it rewards a different input than the one agents were taught to chase.

Follower count used to be the meter. You posted, you gathered followers, and your next post reached a slice of them. That was the social graph: distribution ran along the connections you built. The lever has moved from you to a recommendation engine, and the engine reads signals, not your subscriber list.

What is the interest graph, in one paragraph?

The interest graph is distribution based on what a viewer engages with, not who they follow. When someone watches three videos about first-time buying, the engine files that interest and serves them a fourth from an account they have never heard of. Your follower count is not the gate. The match between your content and a stated interest is the gate.

That is the mechanical shift. Reels, Explore, and the main feed all pull from recommendations now, so a strong post can travel far past your own audience. A weak post stays home no matter how many followers you have.

Can you really get Instagram reach without followers?

Yes, and it is the normal case now, not the exception. Instagram has said publicly that it recommends content from accounts you do not follow, and that reach from unconnected accounts is a core part of the feed. A new account that posts one genuinely useful video about closing costs can reach thousands of the right people, because the engine is testing the content against interest signals, not against a friends list.

The catch is that reach without followers is earned per post. You do not bank it. Each piece starts a fresh test, which is freeing if you are early and unforgiving if you got lazy.

Why is my reach up but my engagement down?

Because the engine is showing your work to strangers, and strangers are a colder audience than followers. When distribution runs on interest instead of connection, more of your viewers have no relationship with you yet. They watch, they judge in two seconds, and most keep scrolling.

This is not a failure signal on its own. A wider top of the funnel with a lower engagement rate can still produce more saves, more profile visits, and more DMs than a small warm audience did. Rate goes down while volume goes up. Read both numbers before you decide a post failed.

What should you optimize for now?

Optimize for the first two seconds and for a clear reason to save. The engine decides whether to keep pushing a post based on early retention and meaningful actions, so the opening frame and the payoff matter more than your handle.

Three concrete levers:

A specific hook aimed at one interest, not a greeting. "Closing costs on a 400k house in Cypress" beats "Hey guys."
One idea per post, delivered fully, so a viewer finishes it and saves it.
A reason to act at the end that fits the interest you targeted, not a generic follow request.

Followers still matter as social proof and as a warm list you own. They are just no longer the thing distribution runs on.

What the interest graph does not fix

It does not fix vague content. A recommendation engine can only match your post to an interest if the post is about something specific. Broad motivational posts have no interest to match, so they sit unshown while you blame the algorithm.

It also does not replace trust. Reach gets a stranger to watch. Your consistency, your profile, and your follow-up are what turn that watch into a client. The interest graph fills the top of the funnel. The rest of the funnel is still your job, and it is still the harder one.

Where AI fits, without the hype

The interest graph rewards a steady volume of good, specific posts, and volume is exactly where a small operator runs out of hours. This is the part of the stack to automate. AI does not write the reach for you. It removes the friction between an idea and a finished, specific post.

A workable content engine looks like this. Keep a running ledger of the real questions clients and leads ask you. Feed one question at a time into a drafting prompt that knows your voice and your market. Get back a hook, a script, and three caption variants. You edit for truth and tone, then record. The system is the spoke that keeps the interest graph fed while you work your actual pipeline.

Two rules keep this safe. Never let a model state a number, a market stat, or a legal or lending claim you have not checked. And keep your own hand on the final edit, because a post that sounds like a template gets scrolled past faster than one that sounds like a person.

A walkthrough (labeled hypothetical, not a specific result)

Picture a newer agent with 180 followers who feels invisible. Instead of posting one polished market recap a week, they pull five real buyer questions from their last month of texts. Each becomes a 30-second video that answers one question with a specific number and a named neighborhood.

They open each with the question itself, not a wave. They close with one line: save this for when you start looking. Some posts do nothing. One reaches well past 180 people because it matched a live interest, and two of those strangers send a DM. Nothing here is guaranteed, and the point is the process, not a promised number. The engine tested five pieces of specific content against interest, and specificity is what it can reward.

The decision

Stop treating follower count as the scoreboard. Build a small content engine that turns your real client questions into specific posts, let AI carry the drafting load, and judge each post by saves and profile visits rather than likes. If distribution now runs on interest instead of connection, you already know which input to feed.

For more on the distribution side, read Coaching: Why My Reels Get No Views. For the systems side, see AI Marketing for Real Estate Agents and Social Media Without the Dance: Quiet Content Strategy.

Download the free 5 AI Automations Guide → https://khaitranofficial.com/ai-ops