Information Matters - Agentic AI News and Market Forecasts

The Agentic AI Revolution: what it means for business and the rules of competition

Our Methodology

Our Methodology

How Information Matters covers the agentic AI sector — primary-source research, transparent scoring, accountable analysis.

Information Matters is an independent analyst publication covering the agentic AI sector. We track and assess around 200 named companies. We follow the same approach a sophisticated buyer or investor would — primary sources, structured scoring, transparent reasoning — and we publish how we work so you can judge it.

The Framework

We assess every company we cover against three questions. Together they make up the Information Matters Framework.

DEFENSIBILITY How defensible is this business today? Customer entrenchment Knowledge & data Distribution channels Strategic resilience Platform / network effects DISRUPTION POTENTIAL How much disruption potential does it carry? Momentum Novel capability Team velocity Category leadership TRAJECTORY Where is the market headed, and how is this company positioned for it? One of eight named futures for the agentic AI sector — see below.

The Information Matters Framework — three layers of company assessment.

The first two questions produce numeric scores from 0 to 10 on each sub-dimension, which combine into composite scores. The third question produces a strategic read — which way the market is heading, and how a given company is positioned for it.

The Information Matters Compass

When we plot every covered company on the two composite scores, we get the Information Matters Compass — our primary visual for showing where each company sits in the market.

Defensibility → Disruption Potential → Disruptive Challengers high momentum, defensibility still being filled in Dominant Innovators high on both axes — defensible AND disruptive Emerging Players mid-pack on both axes Established Incumbents defensible, lower disruption potential

An illustrative Compass — each dot is one company. Quadrant labels describe the competitive position.

The four quadrants split companies into five competitive positions:

  • Dominant Innovator — high on both axes. Companies that have built a defensible business and are still pushing the category forward.
  • Disruptive Challenger — high disruption potential, defensibility still being filled in. Often the pure-play AI-native operators making the most noise.
  • Established Incumbent — defensible position but lower disruption potential. Their moat is real but they aren’t reinventing the category.
  • Emerging Player — mid-pack on both axes. Most companies sit here.
  • Wound-Down — a separate position for residual entities post-acquisition or wind-down.

Eight Futures for Agentic AI

Most analyst publications stop at “where is the company today?”. We add one step: where is the market headed, and how is this company positioned for it?

The agentic-AI sector could evolve in several different shapes. We named the eight most plausible.

Frontier Foundation-model capability keeps compounding Plateau Capability gains level off, market normalises Verticals Vertical specialists beat generalist AI Rewire Operating models get rebuilt around AI agents Low-Cost Compute Inference costs collapse, density of use cases explodes Expensive Compute Compute stays expensive; usage concentrates at scale Inertia Enterprises adopt slower than the hype suggests Borders Sovereign and geopolitical splits fragment the market

The Eight Futures — named scenarios for how the agentic AI sector could evolve.

For each company we identify which one or two of those scenarios its strategy depends on, and how confident we are in that read. Together those answers make up the company’s Trajectory Profile.

How we research

Every figure on our site ties back to a public source. Specifically:

  • Earnings transcripts, SEC and other regulator filings, first-party surveys
  • Vendor disclosures — company blog posts, investor announcements
  • Named secondary coverage from publications we’ve judged reliable — Reuters, Bloomberg, Artificial Lawyer, Law360, Above The Law, FT, WSJ, NYT, The Information, TechCrunch, named industry publications
We do not use paid analyst-database content (Tracxn, PitchBook, Crunchbase Premium, CB Insights premium, Sacra premium, etc.) as our source of record. That content isn’t re-verifiable by a reader following a link. We may reference it as a triangulation check, never as the source.

Where we don’t have publishable evidence, we decline to publish. Our coverage entries explicitly mark cells where we don’t have a confident figure rather than guessing.

How we audit

Before any score reaches a published page, it goes through two checks.

RESEARCH Primary sources, two sources per claim ANALYSE Apply the Framework, score each dimension AUDIT Both checks must pass. Programmatic 20 deterministic checks — math, schema, source policy Qualitative Does prose support the score? PUBLISH

The two-layer audit. Programmatic checks catch what humans miss; qualitative review catches what machines can’t see.

Programmatic validation — every score file runs through 20 deterministic rules covering arithmetic, schema, source policy, label consistency, and more.

Qualitative review — a second pass that no automated rule can do. The reviewer reads the score alongside the prose: does the reasoning actually support the number? Are citations load-bearing? Does the score look plausible given the company’s stage and competitive position?

Both checks must pass. Nothing publishes that fails either.

Who we are

Information Matters is led by Martin De Saulles, Principal Analyst. Martin has over 20 years’ experience as a technology analyst, technology columnist and author of multiple books, reports and academic papers on the technology sector.

The team works alongside Martin as a set of named AI personas with defined research, analysis and communications roles. The personas operate with the same source discipline as a human analyst would — primary-source citations, decline-to-publish over guessing, structured framework application.

Updating the Framework

The Framework is a living methodology. When we find a sub-dimension that doesn’t capture what we mean, we update it and recalibrate the affected scores. When we find we’ve been weighing something incorrectly, we say so and re-baseline. Each material change is flagged on individual company pages with the date of the revision.

We do not silently change scores. Every revision has a history.

Corrections

Spotted a figure you think is wrong? Send corrections to info@informationmatters.net with the page URL and what you’d like us to look at.

If you’d like to talk about how to use the Framework for your own decisions — investment, partnership, vendor due-diligence — get in touch at the same address.