16 Nov 2026 · 3 min read

Newton Research vs Alkimi: Predictive Analytics Agent vs Deal Governance Layer

TL;DR: Newton Research builds agents that analyse marketing data and predict campaign outcomes before and during a buy. Alkimi's marketplace holds the record of what a buying agent and selling agent actually agreed once a deal closes. One is a forecasting layer, the other is a governance layer, and a team evaluating agentic advertising tools needs to know which problem each one is actually solving.

What Newton Research's agents actually do

Newton Research builds what it describes as a team of marketing analytics agents: software that ingests campaign and marketing data and surfaces predictive signal, flagging where a campaign is likely to underperform or where an opportunity is being missed, before a human would otherwise notice. The company has demonstrated this publicly as a predictive layer sitting on top of existing marketing analytics, aimed at media buying and analytics teams who want an earlier signal than a standard dashboard gives them.

This is fundamentally a forecasting and analysis problem. The agent's job is to look at data and produce an informed prediction or recommendation. It does not negotiate a deal, and it is not designed to hold a record of what two parties agreed.

What a neutral deal record does instead

Once a buying agent and a selling agent have agreed terms for a media deal, something has to hold the record of exactly what was agreed, so neither side has to rely on the other's account of it afterwards. Alkimi's marketplace is built around exactly that job: a shared record, written at the moment of agreement, timestamped, and held independently by both sides. The marketplace is also neutral, meaning it does not take a side in any negotiation and does not own the planning, inventory or analytics layers around the deal.

Why these aren't competing products

A team could use Newton Research's predictive agents to decide what to buy and at what price, and then use a neutral marketplace to actually transact that buy with a record both sides can trust. The two problems, predicting a good decision and proving what was agreed, sit at different points in the same workflow. Treating them as competing categories misses that a prediction, however accurate, is not a record, and a record, however trustworthy, does not predict anything.

Frequently asked questions

Could a predictive analytics agent eventually negotiate deals directly?

Possibly, but that would be a different product with a different set of guarantees needed: a negotiating agent has to operate inside a mandate and be accountable for what it commits to, which is a governance problem, not an analytics one. Prediction and negotiation are different capabilities even if the same company eventually builds both.

Does a buying team need both kinds of tool?

Not necessarily at the same time. A team early in adopting agentic buying might only need the governance layer, trusting their own analysts for now. A team with mature predictive tooling already in place still needs somewhere for the resulting deals to be recorded and verified once agents start transacting directly with counterparties.

What should a buyer ask before assuming a tool does both jobs?

Ask directly whether the agent's output is a recommendation a person still has to act on, or a committed transaction. A predictive signal and a signed deal carry very different levels of accountability, and a vendor should be able to say plainly which one their product produces.

All articles

Speak to the team

Request a demo

You don't need to plug in your agent, the demo walks you through the steps.