Chalice AI and OpenX Launch AI‑Powered Video Curation Platform for Programmatic Video – The two firms announced a joint solution that embeds custom artificial‑intelligence models directly into OpenX’s supply‑side infrastructure, allowing advertisers to evaluate and bid on high‑quality video inventory in real time.
What the partnership delivers
Chalice AI, a platform‑agnostic AI media‑decisioning company, has integrated its proprietary models into OpenXBuild™, the real‑time bidding suite that powers OpenX’s supply‑side platform (SSP). The combined offering lets marketers run bespoke AI agents at the moment an impression is served, scoring each video slot for its likelihood to generate a high‑value conversion. In practice, the system replaces static deal tiers with dynamic, impression‑level pricing that reflects predicted business impact.
How the technology works
The core of the solution is “Custom Decisioning,” a container‑based execution environment that runs Chalice’s models inside secure, cloud‑native sandboxes hosted on OpenX’s infrastructure. When a video request reaches the OpenX bid stream, the container evaluates the request against first‑party data, contextual signals, and the advertiser’s own conversion model. The output is a bid price and a recommendation to serve or pass. Because the decision occurs within milliseconds, latency remains comparable to traditional programmatic bids, while the AI adds a predictive layer that standard SSPs lack.
Why it matters for marketers
Video continues to dominate digital ad spend—Gartner estimates that programmatic video will account for 63 % of all video advertising budgets by 2025. Yet the majority of programmatic video buying still relies on pre‑negotiated deals or coarse audience segments, which can obscure true ROI. By moving the optimization logic to the supply side, advertisers gain three concrete advantages:
- Cost efficiency – In a pilot with Hyundai, the AI‑driven workflow cut cost‑per‑action by 20 % and lowered CPMs by 67 % versus the next‑best alternative.
- Performance focus – Real‑time valuation aligns spend with the probability of a high‑value conversion, shifting the metric from impressions to outcomes.
- Privacy compliance – The model operates on first‑party and contextual signals, reducing reliance on third‑party cookies that are being phased out under privacy regulations such as GDPR and CCPA.
Enterprise marketing teams can now plug their own predictive models into a trusted SSP without building a custom bidding infrastructure from scratch.
Competitive context
OpenX’s move mirrors a broader industry trend toward “programmable supply.” Google’s Authorized Buyers and Amazon’s DSP have introduced server‑side bidding extensions that let advertisers run code at the exchange level, but those solutions are tightly coupled to their own ecosystems. Chalice’s container approach is platform‑agnostic, enabling brands to reuse the same model across multiple SSPs or even on‑premise exchanges. Compared with traditional data‑management platforms (DMPs) that feed audience segments into the bid request, AI‑powered curation delivers a per‑impression decision rather than a batch‑processed audience slice.
Implications for enterprise teams
For large advertisers, the new platform reduces the time and cost of building in‑house bidding engines. Marketing operations can now focus on model development—defining conversion goals, training on first‑party data, and testing hypotheses—while the OpenX infrastructure handles scaling and latency. The solution also dovetails with existing Customer Data Platforms (CDPs) and identity graphs from vendors such as Salesforce and Adobe, allowing a seamless flow of unified customer profiles into the real‑time decision loop.
From a measurement standpoint, the AI layer generates granular attribution data that can be fed back into performance dashboards, satisfying the growing demand for end‑to‑end ROI visibility that Forrester predicts 45 % of marketers will require by 2026.
Market Landscape
The adtech ecosystem is at a crossroads. Programmatic video spend is projected to reach $70 billion globally by 2027, according to IDC, while privacy‑first initiatives are forcing a shift away from third‑party cookies. SSPs are responding by opening their bid streams to custom logic, a capability once reserved for demand‑side platforms (DSPs). Simultaneously, AI adoption across adtech is accelerating; a McKinsey study shows that 38 % of advertisers plan to embed machine‑learning models into their media buying workflows within the next 12 months.
Chalice AI and OpenX’s collaboration sits squarely in this inflection point, offering a scalable, privacy‑compliant way to fuse AI with supply‑side operations. Competitors such as The Trade Desk’s “Koa” and Amazon’s “Unified Ad Marketplace” are also exploring programmable bidding, but the open‑container model championed by Chalice may provide the most flexibility for brands that already own sophisticated AI assets.
Top Insights
- Dynamic pricing beats static deals: The Hyundai pilot demonstrated a 20 % drop in cost‑per‑action and a 67 % CPM reduction, underscoring the financial upside of AI‑driven impression valuation.
- First‑party data becomes the new currency: By relying on proprietary signals rather than cookies, the platform aligns with emerging privacy regulations while preserving targeting precision.
- Enterprise agility improves: Marketers can deploy and iterate on custom AI models without building a full‑stack bidding engine, accelerating time‑to‑market for performance‑focused campaigns.
- Competitive advantage through openness: Unlike ecosystem‑locked solutions, the container architecture works across multiple SSPs, giving brands the freedom to negotiate inventory on their terms.
- Attribution granularity rises: Real‑time AI decisions generate detailed conversion likelihood scores, feeding richer data back into CDPs and analytics stacks for better ROI reporting.
Get in touch with our Adtech experts
