The Morning Edition · Every Tuesday at 7:30 AM

Know, don’t guess.

A research brand for the people who actually have to ship AI inside enterprises. We publish what we wish someone had written for us when we were doing the same job. No hype. No hot takes. Numbers we can stand behind.

Free· About five minutes· Lands Tuesdays at 7:30 AM

What we publish

A few products. One standard.

Every surface here has to carry something a CIO, CoE head, or innovation lead can use the week they read it. If a piece doesn’t pass that test, it doesn’t go out.

Live

The Morning Edition

A weekly brief that lands every Tuesday at 7:30 AM. Eight hundred words on what actually moved in enterprise AI, with the survey data or teardown to back it up.

Q3 2026

Inference Research

Quarterly flagship reports on enterprise AI in production. Our methodology is in the appendix. Our raw data is downloadable. Sponsors fund distribution; they don’t shape the findings.

Q3 2026

Inference Roundtables

Invite-only sessions of fifteen practitioners. Chatham House rules. One vendor underwrites the room. Everyone else is a peer at the table.

Q3 2026

The Inference Podcast

Biweekly conversations with the people actually shipping AI inside large companies. Forty minutes per episode. No mid-roll ad reads cutting into the answer.

2027

Inference Quarterly

A premium print magazine for our paid subscribers. Sixty-four pages of long-form research, every quarter, mailed to your office. Think of it as the part of the year that earns a coffee table.

Why this exists

There’s plenty of AI noise out there. We’re here for the signal.

In a survey we ran last quarter, 87% of AI leaders told us they can’t close the gap between pilot and production. If you’re one of the people who has to close it, you probably already know the existing research wasn’t written for you. That’s the gap we’re filling.

01Evidence first

Every number you read in our work is sourced. Every claim is testable. When the data is too thin to draw a conclusion, we say so out loud rather than dressing it up.

02Written for operators

The people accountable for shipping AI systems are the audience. If a paragraph wouldn’t survive a Monday morning standup, it doesn’t make it into the brief.

03Transparent on money

Sponsors fund our distribution. They don’t shape what we cover. Once a quarter we publish exactly who paid us and what they paid for. The Morning Edition itself stays free for readers.

Latest issue

Issue 3 · 2 June 2026

The Evaluation You Skipped

What the 13% who ship actually run in production, and what the 87% wish they had built in Month 2. The evaluation harness is the cheapest line on the roadmap and almost always the last thing anyone builds.

Read Issue 3

THE NUMBER

200+

real production failures in your golden dataset before you ship, wired into CI as a merge gate. The median pilot ships with twenty hand-checked rows in a spreadsheet.

Also recent · what's coming

Issue 2 · 26 May

The Nine-Month Shape

Three enterprise LLM platform builds, reverse-engineered. The same nine-month timeline shows up in all of them, and one decision compresses it by a quarter.

Read Issue 2

Friday Supplement · 22 May

The AI eval tools shortlist

Fourteen tools currently shipping. Eight criteria that decide which one fits. One printable scorecard. A follow-up to Issue 1's Move 3.

View on LinkedIn

Coming next Tuesday

Issue 4 — Unit economics of a shipped feature

What an AI feature actually costs per user once it leaves the pilot, why the pilot cost model is usually wrong by an order of magnitude, and the three line items that blow up at scale.

Research briefings

Longer-form research, in between the weekly issues.

Quarterly briefings that go deeper than the Tuesday morning newsletter. Free to read. Free to share.

Briefing · June 2026 · 9 min read

AI in FinTech: The Race for Sovereign Infrastructure

A $99B market becoming a $500B+ one. A 1,210% surge in AI-enabled fraud. Three nation-states quietly building the infrastructure that will decide who runs financial AI for the next decade.

Read the briefing

$500B+

market size by 2034

1,210%

AI-enabled fraud surge

3 bids

UAE · Saudi · EU

Subscribe

Get the Morning Edition in your inbox every Tuesday.

It’s free. About five minutes to read. One-click unsubscribe whenever you’re done. A new issue lands every Tuesday at 7:30 AM IST.

Subscribe on Substack

No spam. No tracking pixels. No drip sequences.

Founder

Who’s writing this.

A publication is only as honest as the person doing the writing. Here’s who that is at Inference.

Aditya Patro

Founder & Editor

I started Inference because the research I wanted to read about enterprise AI didn’t actually exist. Plenty of vendor blogs. Plenty of analyst reports priced for boards, not operators. But not much in between for the people who own the actual production roadmap.

Before this, [edit: your prior role and one specific thing you did there]. I write every word of the Morning Edition myself. If a brief sounds like it came out of a content factory, it didn’t come from us.

Connect on LinkedIn