Publisher-owned yield intelligence
The intelligence layer moved to the buy side. bidBrain moves it back.
bidBrain is auction-speed decisioning that you own. It reads live session context, adapts your request strategy inside the break, and shows you what it changed and why.
Why now
Your demand partners got smarter. Your sell side didn't.
Buy-side intelligence has been compounding for years. Bidders model your inventory, price it, and adjust continuously. Supply-path optimization keeps narrowing the list of sellers they bother talking to, and agentic buying tools now sit on top of all of it.
The sell side answers with static waterfalls and rules somebody wrote last quarter. A static rule is a decision made once and then repeated, whether or not the market still agrees with it. That gap is where yield leaks, quietly, every break.
The answer isn't a better rule. It's a system that senses, decides, and learns at the speed the auction actually runs.
Buyers model you in real time
Bid-side systems price your inventory continuously. Your side of that conversation is a configuration file.
Supply paths keep narrowing
Buyers are consolidating who they buy through. The sellers who can demonstrate decisioning quality are the ones who keep the volume.
Rules go stale without telling you
A strategy that worked in March can be losing money in June. Nothing in a static stack surfaces the drift.
What it is
Auction-speed decisioning you own and can inspect.
bidBrain sits in your request path and decides the way your sharpest ad ops person would, on every request, at machine speed. Strategy lives in artifacts you author: demand sources, request strategies, optimization rules. Each one describes itself and adapts as conditions move.
One shared engine handles expressions, rules, mappings and context, so those artifacts compose into the configurations your team already thinks in. Values pass through unchanged unless you explicitly override or suppress them, and every change the system makes is visible after the fact.
Demand sources that describe themselves
Model each partner once. Static, macro and path mappings source the values, and expressions compute on top of them. The template library auto-matches industry-standard macros per SSP.
Request strategies you can reason about
Compose rounds, overrides and suppression into strategies your ad ops team edits directly. No ticket, no release train.
Full session context
Path lookups draw on the whole session, live ad-pod state included, so a decision late in a break knows what already happened in it.
A control group that never stops running
A share of live traffic runs as pure pass-through. Results are measured against the market as it is right now, not against a projection made last quarter.
Yield testing without guesswork
Run blind requests that strip content signals to find out what your inventory is actually worth to each buyer.
oRTB on both sides
Parse and template oRTB in and out, down to leaf-level mapping when a partner needs something specific.
What changes when the intelligence is yours
Yield
Decisions that move with the market
Strategies adapt inside the break instead of waiting for the next config change. The distance between what your inventory is worth and what you ask for it stops widening.
Control
Your logic, your data, your call
Decisioning lives on your side of the auction. Read any rule, change it, and watch the effect. It isn't a black box you rent.
Speed
Ad ops ships without engineering
Artifacts are authored in the product, not in a codebase. The people closest to the problem are the ones who get to solve it.
See it working
Real screens from the product. Try the demo to move through them yourself.
Request strategies
Strategies are objects your team owns. Parallel, weighted and sequential patterns, each with its own call count and timing.
Per-round plans
Each round declares its floor, timing and source weighting. Later rounds inherit from the strategy unless you say otherwise.
Demand sources
Model a partner once. Readiness comes from an actual test against the source, not from the configuration looking complete.
Field mapping
Static, macro and path mappings source the value. Expressions compute on top of them. Path lookups reach the whole live session.
Optimization strategies
Fill, max value, max yield and rule-based optimizers, assignable per strategy rather than set once globally.
Baseline vs optimized
A pass-through cohort runs alongside the optimized traffic. Upside is the difference, measured against the market as it is now.
Where this came from
Built on one conviction
Complex systems don't reward rigidity. They reward adaptation. Ad markets are complex systems, and static rules are how you lose to them slowly. bidBrain is that idea applied to the auction, built by the team at Adaptive Chaos Labs.
Try it
Run it against your own thinking.
The demo is the product, loaded with sample demand and live decisioning. Bring a strategy you already run and see what it looks like when it can adapt.