To find influencers for a tech or AI brand, start from your buyer and work outwards: search X for who already posts about your category, read who quote-posted your competitors' launches, check who carried launches like yours in a public directory, then look at LinkedIn, newsletters and podcasts. Vet each name on median views, audience fit and engagement that looks real before you think about followers.
This guide is the method we use. Doomers AI books founders, engineers and creators for AI launches on X from a network of 1,000+, and we vet new accounts this way every week. It takes an afternoon and no paid database.
Start from the buyer
Write down who you need to reach before you search for anyone. "AI people" is a category. "The engineer who owns evals at a Series A company" is a buyer, and the search changes completely once you name them.
Then work out where that buyer reads. Developers, founders, AI researchers and investors are mostly on X. HR, sales and finance leaders are mostly on LinkedIn, and educators who teach a tool live on YouTube.
The shortlist you want is the accounts that buyer already chose to follow. Tech influencers on X covers the four types worth casting: founders and operators, engineers and builders, research accounts and tech generalists.
Where to find influencers for a tech brand
X search
Search your category terms, your product name and your competitors' names, then switch to the Latest and Top tabs. X's search operators narrow it fast: min_faves:200 keeps posts with at least 200 likes, and from:handle reads one account's history.
Note who posts about the category with an opinion, and who your buyers reply to. The people already arguing about your space are the shortlist, and some of them may already use your product.
Who quote-posted competitor launches
Open a competitor's last launch post on X and read its quotes. Whoever amplified it is reachable, already knows the category, and has shown they will post about products like yours.
Weigh the quotes that carry a take over the ones that just say congratulations. A creator who adds an argument moves a launch, and one who adds an emoji does not.
The launch directory
The Doomers AI launch directory holds 2,300+ AI and tech launches on X with their views, bookmarks and the size of the account that posted. Filter by vertical and account size to find launches like yours, then open the post on X and read who quoted it.
This shows you who carries launches in your category in practice. A founder who quoted the last three model launches in your space with an opinion is a strong candidate for yours.
Search LinkedIn posts for your category and sort by engagement. Founder-operators and most of the buying committee are here, and posts convert better and start slower. LinkedIn creator partnerships covers how they price.
Newsletters and podcasts
The writers and hosts your buyers already read are often on X too, and their guests are a ready-made list of people the audience trusts. Start with the best AI newsletters and the best AI podcasts, then check each host's X account with the vetting below.
Your own team and users
Ask your engineers who they read. They have followed these people for years, and it takes ten minutes. Then search for customers already posting about you, since they are the most credible carriers you have.
How to vet an influencer in five minutes
Run four checks on every name, in this order. An account that fails one of the first two is out, however large it is.
| Check | How to run it | What good looks like |
|---|---|---|
| Median views over followers | Note the views on their last 20 to 30 posts and take the middle one. Ignore the one that went viral | A steady median. Thirty posts between 40K and 90K beats one at 2M and the rest at 4K |
| Audience fit | Read the replies on their last ten posts and look at who is answering | Engineers, founders and people with company names in their bios that match your buyer |
| Engagement that looks real | Compare likes, replies, reposts and bookmarks against views, and read the reply text | Replies that argue or ask questions, in proportion to the views |
| Past paid posts | Find their sponsored posts and see how they were labelled and how often they appear | X's paid partnership label every time, and sponsored posts a small share of the feed |
One more check that most teams skip: overlap between the creators you pick. Thirty accounts drawing on one shared audience reach fewer distinct people than ten that do not. What we check before we pay a creator is the full buy-side list.
Red flags
- A big following and a flat median. 300,000 followers and 8,000 views a post means you are paying for an audience that has stopped reading.
- Replies from bots, emoji or the same few accounts. Engagement pods and bought engagement show up in the reply text, and nothing reported from that account can be trusted.
- A feed of ads. An account that posted about forty tools in six months is a billboard, and its readers treat it as one.
- Undisclosed paid posts, or a request to skip the label. The exposure lands on your brand. The paid partnership label covers why.
- A media kit built on followers. Creators who lead with follower counts and reach summed across platforms are counting the same reader up to four times.
- A promised view count. Nobody controls the algorithm. A guarantee means guessing, or buying engagement when the post falls short.
How many influencers a launch needs
One post is an experiment. A launch is cast across several types, so a reader sees it from a founder, then a builder, then a generalist they follow, inside the same few hours.
TypeSafe AI came out of stealth from a founder account of about 1,500 followers, carried by 80 to 100 engineers, founders and creators, and the post did 40.1M views. The TypeSafe AI case study shows how the launch was carried.
Smaller launches need fewer people, picked more carefully. Micro vs macro influencers covers the trade between a few large accounts and many small ones.
Doing it yourself, or using a network
Creator databases sell the same scraped follower counts, and the selection, briefing and timing are still left to you. The manual method above beats them for a specific launch because it starts from your buyer.
A launch agency with its own creator network has already done the vetting, knows each creator's median and rate, and can book them for a fixed window. B2B creator rates covers what the posts cost either way.
Questions we get asked
How do I find influencers for my brand?
Start from your buyer, then search the platform they read for accounts already posting about your category. On X, read who quote-posted competitor launches and who carried similar launches in a public launch directory. Check LinkedIn, newsletters and podcasts your buyers follow, then vet every name on median views, audience fit and real engagement.
How do I find tech influencers on X?
Search your category and competitor names with operators like min_faves to surface posts that travelled, then read who replied and who quoted them. Open competitor launch posts and read their quotes. The Doomers AI launch directory lists 2,300+ AI and tech launches you can filter by vertical to see who carries launches like yours.
What should I check before paying an influencer?
Their median views across the last 20 to 30 posts, who replies to them, whether engagement is in proportion to views, and how past paid posts were labelled. Follower count comes last. An account whose posts typically do 60,000 views is worth more than one with ten times the followers whose posts typically do 8,000.
Are influencer databases worth paying for?
For a single launch, rarely. Most sell scraped follower counts and leave the selection, briefing and timing to you. Searching from your buyer outwards takes an afternoon, and a launch agency with its own vetted network already knows each creator's median and rate.
What are the red flags when choosing an influencer?
A large following with a flat median, replies from bots or engagement pods, a feed that is mostly sponsored posts, undisclosed paid posts, a media kit built on followers, and any promised view count. Each one means the numbers you get back cannot be trusted.