Agentic commerce. Human insight.
I help brands and retailers understand how AI shapes what customers discover, compare and choose, and what shapes the recommendation.
Three positions from what I have written and investigated in public, each linked to its source.
Visibility is the start of the question. What matters commercially is which sources, product facts and offers led the answer there.
LinkedIn, 7 September 2026 ↗Go deeper: AI product discovery →The storefront an agent sees is the structured information behind the page. A new platform does not help if that information is incomplete.
“The Invisible Storefront”, LinkedIn, 6 August 2026 ↗Go deeper: Product data readiness →AI answers vary from run to run. Conclusions need the same question asked many times, under documented conditions.
Hyper Agency study on AI election compasses, 2 September 2026 ↗Go deeper: LLM answer stability →Events, studies and essays where I have pushed the question of how AI recommends products.
Conversations with brands and retailers at IFA about how AI answers recommend products, and why visibility alone does not explain the outcome.
Read the post on LinkedIn ↗A study based on 60,000 API calls, showing large differences in party sympathies between AI-built election compasses. Covered by Resumé and Dagens Media.
Read the study ↗Go deeper: LLM answer stability →Coverage: Resumé ↗ · Dagens Media ↗
Why the re-platforming panic misses the point: AI agents read product data, not page layout.
Read the essay on LinkedIn ↗Go deeper: Product data readiness →Hosted sessions at the C2 event together with Mike Silver, during two weeks of meetings in North America.
Read the post on LinkedIn ↗Each chapter added a question I now ask of AI: who is buying, what is sold, and why that product.
How selling is taught in a store.
How store standards carry into a district.
How a store network and a consumer business develop across functions.
How a regional store organisation is led day to day.
How a product company becomes useful through partnerships.
How client programmes develop across markets.
Bringing the commercial questions to AI: what shapes a recommendation, and how to measure it.
Three recommendations from people I have worked with, in full and linked to the original.
Efter att ha jobba nära Fredrik i över tre år har jag sett hans förmåga att driva, leda och genomföra både affärs- och organisationsutveckling. Fredrik har en härlig kombination av starkt ledarskap och en fingertoppskänsla och spetskompetens när det kommer till retail och utvecklingen av retail som fenomen. Det som som kännetecknar Fredrik i allt detta är hans trygghet, ödmjukhet och engagemang i allt han gör. Grattis till dig som får jobba nära Fredrik!
Translated from Swedish
After working closely with Fredrik for more than three years, I have seen his ability to drive, lead and carry through both business and organisational development. Fredrik has a fine combination of strong leadership and a sure feel for, and deep expertise in, retail and the development of retail as a phenomenon. What characterises Fredrik in all of this is his steadiness, humility and commitment in everything he does. Congratulations to anyone who gets to work closely with Fredrik.
Fredrik och jag har arbetat tillsammans under flertalet projekt rörande utvecklingen av 3s butikskedja med ca 70 butiker.
Fredrik är väldigt enkel att arbeta med och man vet att han alltid kommer leverera. Han har ofta många bollar i luften men som motpart känner man sig alltid trygg i Fredriks leveranser.
Fredrik besitter en bra strategisk höjd och lyfter ofta blicken från det operativa för att se hur arbetet fogar in i den övergripande strategiska planen.
Därtill besitter Fredrik en stor kunskap om Retail från flertalet branscher vilket har varit till stor nytta för 3.
Translated from Swedish
Fredrik and I have worked together on a number of projects concerning the development of 3’s store chain of about 70 stores.
Fredrik is very easy to work with, and you know that he will always deliver. He often has many balls in the air, but as a counterpart you always feel confident in Fredrik’s deliveries.
Fredrik has a good strategic altitude and often lifts his gaze from the operational work to see how it fits into the overall strategic plan.
In addition, Fredrik has extensive knowledge of retail from a number of sectors, which has been of great use to 3.
Fredrik is a biz dev and client service pro for the digital age. His ability to marry data with superb client relations allows him to craft solutions truly tailored towards his clients needs. His commitment is unwavering across multiple timezones and geography. I learn from him everyday and it’s been an absolute pleasure growing client programs and accounts together!
The guides, method, study and tools I use to reason about agentic commerce.
A practical guide to the decisions, data and handoffs behind AI-assisted commerce. Start with the customer’s task. Follow it all the way to a valid offer.
02 / FIELD GUIDEInvestigate how a customer’s situation changes the products, brands and sellers that enter an AI-generated shortlist.
03 / FIELD GUIDEConnect product identity, verified attributes and live commercial terms. Find the gaps between a compelling description and a purchasable offer.
04 / METHODAn explicit approach to testing AI recommendations: clear definitions, comparable conditions, repeated observations and honest limits.
05 / RESEARCHA reading of Hyper Agency’s published election-compass study, with a clear distinction between its reported findings and the lessons for measurement practice.
06 / INTERACTIVEAn interactive explanation of eligibility and suitability. Explore how budget, buyer priorities and seller constraints shape a fictional buying decision.
07 / WORKING TOOLA working checklist for a single market and product category. Mark what you can substantiate, then download the open items as an action plan.
08 / PERSPECTIVEMake the seller, local assortment and fulfilment conditions part of the investigation. A product recommendation only becomes your opportunity when a relevant offer connects to it.
09 / PERSPECTIVEConnect the source specifications, the customer’s use case and the retailer representation. Find where meaningful differentiation gets lost.
10 / FIELD NOTESOriginal perspectives on AI discovery, recommendation measurement and the product information that connects them.
11 / REFERENCEA decision-oriented introduction to ACP and UCP. Use the official specifications to investigate capabilities, eligibility and responsibilities for a specific integration.
12 / REFERENCEPlain-language working definitions for the conversations between commerce, marketing, product information and engineering.
FREDRIK KRIDAHL / STOCKHOLMAdvisory, a talk or an interview about how AI recommends products. Tell me about the market, the category or the question.