by Adgentek
Ad server for AI chatbots and agents that sorts each conversation into 9 intent buckets and serves labeled native ads; free to start, a few lines of JavaScript.
Use it if you run a free browser-based AI chatbot and want ad revenue toward inference costs without a traffic minimum. Skip it if you build native mobile apps or headless agents, where the SDK and CPM demand are limited. Fill follows purchase intent, so expect gaps.
Intent classification of the conversation itself, with four demand tiers bidding on that signal.
Founders with prior DSP and ad-server businesses (Beachfront, Simpli.fi) give it demand relationships, but the buyer side is shared with every rival AI ad network and the SDK is a few lines to swap.
Main trade-off
Web-only SDK today, with fill that depends on how many purchase-intent conversations the app has.
| Plan | Monthly price | What it includes |
|---|---|---|
| Publisher account | Free | Signup at no charge, JavaScript Web SDK and a publisher dashboard with placements, reporting and earnings, Publisher controls: formats, category and brand blocklists, frequency caps, relevance floors, Earnings accrue net of a platform fee |
| Advertiser access (Spark, ORCA beta) | Custom | Spark interactive conversational ad format, Adgentek ORCA agentic media buying, listed as in beta, Pricing not published, contact hello@adgentek.ai |
Publisher account: no charge. Advertiser access (Spark, ORCA beta): no published price, contact hello@adgentek.ai.
Adgentek Ad Server is an ad server for AI apps, chatbots and agents. It sorts each conversation into one of nine intent buckets and serves a labeled native ad in a slot the publisher places, using a JavaScript SDK. Four demand tiers compete for each request, and publishers can start for free.
JavaScript Web SDK
Two snippets load the SDK and request an ad after each assistant response; the ad renders in a slot the publisher places and styles around.
Conversation intent classification
Each exchange is sorted into one of nine buckets; support, entertainment and low-confidence turns are deliberately served no ad.
Four-tier demand waterfall
Direct campaigns, programmatic, CPC feeds and CPA budgets compete for each request, so a request that misses the top tier falls to the next.
Publisher controls
Standard, Strict or Custom brand safety, category and advertiser-domain blocks, a minimum relevance threshold and a minimum conversation turn are set under Ad Policies.
Derived-signal privacy model
Demand partners receive intent category, decision stage, topic and IAB categories, not raw conversations, according to Adgentek's privacy architecture.
Labeled native formats
The dashboard exposes inline cards, native listings, sidebar cards, banners, voice slots and video, each carrying a visible Sponsored or Conversational Ad label.
Docs written for coding agents
docs.adgentek.ai publishes an llms.txt and a skill.md so a coding agent can do the integration, and an AdsMCP endpoint serves MCP-capable agents.
Most users expect these and they are not currently supported.
The nearest AI tools on Hokai for the same job.
| Tool | Price from | Free plan | Runs on | Rating |
|---|---|---|---|---|
| -- | Yes | Web | 3.1 | |
| Free | Yes | iOS · Android | 3.8 | |
| $3.92/mo | Yes | Mac · Windows | 3.6 | |
| -- | Yes | Mac | 3.2 |
Proprietary
A publisher account is free to start, with no traffic minimum and no exclusivity, according to Adgentek. Publishers earn net of a platform fee whose percentage is not published on the pages Hokai read, and a directory listing says they earn on CPM. Payment is made within 75 days after the end of each calendar month in US dollars, and balances under $100 may be deferred. Advertiser pricing for Spark and ORCA is not published, so you contact hello@adgentek.ai.
Yes to start: you create a free publisher account, add the JavaScript SDK and place a slot. Free here means no signup cost, not guaranteed income, because an ad only serves when intent and demand match and the dashboard's payout setup is marked coming soon. Test mode exists in the docs for trying the integration before live traffic.
Adgentek's FAQ says this is by design. The ad server serves only when intent and demand genuinely match, so support conversations, entertainment turns and low-confidence classifications intentionally return no ad. Conversational inventory is meant to earn on relevance and engagement depth rather than on fill rate.
Publishers send the conversation messages needed for contextual matching and a stable session_id. Adgentek says demand partners receive derived signals only, such as intent category, decision stage, topic and IAB content categories, and that raw prompts, chat history and personal data are not sent to buyers. Its docs add that standard advertising request fields may be included in the demand path.
Partly. Adgentek's docs say impression-priced demand, meaning direct and programmatic, needs client-side proof that a real user saw the ad, which a headless surface cannot give, so only CPA demand applies there. Products without a browser UI are told to contact hello@adgentek.ai for direct setup, and native mobile SDKs are not out yet. Pick Adgentek for a browser-based chatbot; for a native mobile app or a headless agent, ask Adgentek first.
Adgentek Ad Server is sell-side ad decisioning for AI apps, chatbots and assistants, built by Adgentek. Adgentek's pitch is that a classic ad server decides on page URLs, device IDs and cookie audiences, none of which exist inside a chat. This one reads the latest user and assistant turns, sorts the conversation into one of nine intent buckets (purchase, transactional, consideration, comparison, research, informational, navigational, support, entertainment), and serves a clearly labeled native unit in a slot the publisher places in its own interface. Adgentek's team page lists Frank Sinton, who founded Beachfront, as CEO, and Paul Harrison, co-founder and CTO of Simpli.fi, as President.
Per Adgentek's documentation, the ad request runs in parallel with the model call, and if no decision comes back in time the slot stays empty and the conversation is unchanged. Adgentek says a coding agent such as Cursor can complete the integration from its public docs.
It fits developers of free chatbots and assistants who need ad revenue to cover inference costs, and advertisers who want interactive units inside AI conversations. Advertisers reach the same inventory through Spark, the guided question-and-answer ad format, and Adgentek ORCA, its agentic media-buying platform, which the company's llms.txt lists as in beta. Hokai read the public automotive demo on 11 November 2026: it shows one card for a fictional car company, labeled "Conversational Ad · Sponsored", with guided answer chips, "Build & Price" and "Find a Dealer" buttons and a "Powered by Adgentek" footer. Hokai did not run the SDK itself.
Pick something else if your product is not a browser app. Adgentek's documentation says React, iOS and Android SDKs are still in development, and that headless agent surfaces cannot earn impression-priced demand, only CPA. Adgentek's own llms.txt says it is not the right fit for display banners on ordinary web pages, DSP seat management or search marketing. Koah, whose homepage (read 11 November 2026) calls it the ad network built for AI with real-time intent matching, is another AI-chat ad network that Hokai has not yet listed. Adgentek is also not Adgentic, a separate affiliate-API product at adgenticplatform.com that a GetChatAds roundup, read on 11 November 2026, describes as a managed affiliate platform for AI agents. For the wider list, see emerging AI software.
Product facts on this page (pricing, plans, integrations, compliance) are read from Adgentek's own pages and documentation. The verdict is Hokai's reading of them, not the vendor's claim.