This blog is a write up of our latest KIMRA Conversations podcast with Ludovic and Simon. Listen now on Spotify, Apple Podcasts or Amazon Music.

The blog expands on a podcast conversation between Ludovic Leforestier / Starsight and with Simon Burton / KIMRA. We discussed how AI changes research production, analyst-firm economics and the human value that survives both.

Analyst firms have an abundance problem. AI makes research quicker to produce and easier to interrogate. That sounds useful. It also weakens the traditional research business model.

“People do not want more content.” They want the right answer, from somebody they trust, for the decision in front of them.

Research escaped the research firms.

Information used to be scarce. Analyst firms collected privileged vendor and buyer insight, packaged it into reports and sold the same research repeatedly through subscriptions. The FIGs—Forrester, IDC and Gartner—built scale around that model.

The internet broke the first barrier. Vendor websites, blogs, forums, LinkedIn, Glassdoor, Crunchbase and Reddit made market information abundant. The creator economy gave independent experts their own distribution.

Peer reviews broke another barrier. Buyers could compare structured customer experiences alongside formal research. We explored that shift when B2B peer-review sites began moving into analyst territory.

“Research can be produced faster and interrogated faster.” AI can search, summarise and repurpose that abundance in seconds. The second point matters more than the first.

The delivery model is already changing. AskGartner provides tailored answers from Gartner research. Forrester AI lets clients converse with reports and takes research into Microsoft Teams and Microsoft 365 Copilot.

AI is the Napster of analyst research.

The music industry has seen this film before. Musicians once made money from recordings and used concerts to promote them. Streaming turned that model upside down and the recording became the marketing engine. The value moved towards live performance, communities, merchandise and privileged access.

Traditional analyst research is the album. Create it once. Sell it many times. Add advisory, consulting and conferences around the edges. I first developed the music-industry analogy in 2011, arguing that customers buy insights, not research—and that the experience is better live than reheated

“AI is the Napster of the analyst industry.” The barriers to producing competent-looking content are collapsing. We can generate a plausible market report in half a day. So can everybody else.

Plausible is not insightful. When similar models process similar public sources, they converge on the same polished average. Volume rises. Differentiation falls.

“If you have read it in an LLM, somebody else thought about it before you did.” Originality now requires proprietary evidence, human judgement or a genuinely new framework.

The long tail just gained leverage.

Independent analysts are not new. What changed is their leverage. They now have better data, AI acting as a junior researcher and direct distribution through LinkedIn, Substack and other platforms. The same unbundling is reshaping professional services: in my piece on why agencies must change I drew on the Harvard Business Review argument that expert networks and specialist providers disrupt integrated consulting firms.

Specialists can beat scale in a niche. IT-Harvest, for example, maps a very broad cybersecurity market quickly. A larger firm may still add customer evidence, methodological depth and board-level influence.

Choose analysts by the job required. A specialist can build the longlist. A major firm can validate a strategic decision. Our map of 100 analyst firms shows why firm size is no longer a proxy for relevance.

“You do not go to Crunchbase or McKinsey for the same thing.” Nor should you ask every analyst firm to solve the same problem. Coverage, depth, buyer access and influence differ.

Start-ups should exploit this choice. Engage niche analysts early to test category, positioning and product-market fit. Our ten truths for fintech start-ups explain why you should do that before buying expensive subscriptions.

AI is (just) a fast junior researcher.

“AI can do the work of a junior researcher much quicker and much cheaper.” Analysts record and transcribe briefings, summarise conversations and process public sources. Research chatbots let clients interrogate the resulting intellectual property.

“AI is just a tool.” Good analysts use it to frame questions, build mental models, categorise evidence and test contrary views. Weak analysts use it to make weak research faster.

The analyst still supplies the intelligence. Expert prompts help. Proprietary inputs help more. But differentiation ultimately comes from how an analyst thinks. Our scenarios for analyst firms in an AI-driven market found analysts treating AI as a force multiplier while reserving trust, context, influence and strategic sentiment for humans.

Knowledge managers face the real challenge. They must separate the wheat from the chaff: true insight from an expanding mountain of credible-looking rubbish. Forrester argues that trust must be designed into AI research delivery.

AR risks an AI arms race.

AI can improve analyst relations. It can summarise subscriptions, prepare tailored inquiry questions, maintain shared knowledge bases and accelerate responses to RFIs.

But AI can also industrialise pointlessness. Vendors use AI to write longer RFIs. Analysts use AI to read them. “In the end, nobody reads what anyone is writing.”

The right human must remain in the loop. Not somebody who merely approves the output. Somebody who understands the market, the question and the evidence behind the answer.

Briefings still need people. Analysts consistently ask for dialogue, specificity, honesty, customer evidence and brevity. AI-generated slides do not replace those qualities.

The premium moves from knowing the answer to framing the problem. Junior talent still matters. But curiosity, categorisation and judgement matter more than basic information processing.

AR becomes the human translator. It connects the vendor’s reality, the analyst’s model, the buyer’s problem and the machine-generated version of all three.

Clients want answers, not research volume.

The underlying need has not changed. Clients want trusted answers that help them make a decision. Their expectations of speed, brevity and personalisation have changed completely.

“Those millennials will not read a 15-page academic-looking report.” I made that argument to a Gartner research leader fifteen years ago. Today it is no longer a prediction.

People expect immediacy and brevity. They want an answer with evidence, context and the option to go deeper. Forrester says clients use its AI to cut through research volume and reach relevant material faster.

Personalisation creates blind spots. An algorithm serves what appears relevant. But “you do not know what you do not know”. Research still needs to challenge the question, not merely answer it.

“Clients want answers that are pertinent, personalised and relevant to them.” They also yearn for thought leadership, expertise and showmanship.

Speed gets you into the conversation. Trust determines whether your advice is used. The winning model combines immediate access with credible expert interpretation.

The experience becomes the product.

Research will remain essential. It establishes credibility and creates the evidence base. But the report will increasingly behave like a database, not a destination.

Value moves towards experiences. Advisory conversations, consulting, peer communities, roundtables and live events apply research to a client’s reality.

“Research becomes the credibility. The experience becomes the product.” The album supports the concert. The report supports the advisory relationship.

Analyst events reach where reports cannot. They occupy mindshare, build association, deepen relationships and expose buyers to unfamiliar questions.

In-person connection becomes more valuable because everything else is automated. It is difficult to commoditise charisma, debate, empathy or a well-timed challenge. “We all need that to exist as humans.”

Analyst firms must market their analysts.

“Research is a people industry that markets itself as a data or content industry.” That becomes untenable when machines can manufacture content cheaply.

Show the expertise behind the output. Who did the work? What have they done before? Where are the battle scars? Phil Fersht argues that the industry must stop publishing “analyst slop” and return to battle-scarred practitioners who can challenge customers. That is the pedigree clients trust—not a plausible answer from a generic model.

Great industry analysts do what AI cannot.

First, they create trust. Clients want advice from somebody with credibility, context and accountability.

Second, they notice weak signals. Experienced analysts detect uncertain changes in conversations and behaviour before those changes become clean, structured data.

“AI is bad at finding things people are not sure about.” It can identify statistical outliers in structured data. Human analysts can hear doubt, hesitation and changing sentiment in an imperfect conversation.

Third, they judge ambiguity. They can attach sensible confidence to incomplete evidence instead of producing a certain-sounding answer.

Finally, they remember. Humans carry decades of context. They connect today’s developments with patterns they saw twenty or thirty years ago.

Prediction is allowed to be human. Analysts will sometimes get the future wrong. That is also what makes their thinking interesting.

Do not compete with AI on volume.

Analyst firms should compete on trust, weak signals, judgement and experience. Vendors should bring better evidence and create better conversations.

AR must keep both sides human. Its value is not moving information between organisations. It is translating context, confidence and intent. Mark Smith reaches a complementary conclusion: research must become continuous, transparent and embedded in enterprise AI. The business is not disappearing. The old version should.

AI can summarise what is already known. Great analysts help clients understand what is not yet obvious.