Research & Insights Services: Agents Are Coming

Agentic purchasing is coming to the research, analytics and insights industry—according to many of its largest buyers and suppliers. Expectations are unevenly distributed, and the strongest believers appear to control a disproportionate share of the money, growth and outsourcing momentum that could make the expectation become reality.

Industry Expectations Differ Sharply

According to the 2026 GRIT Insights Practice Report, 66% of analytics-oriented insights professionals on the brand-side completely or mostly agree automated or agentic systems will mediate most research/insights services purchases within the next few years. Half of the largest suppliers of insights services (500+ FTE) echo this belief, as well as 43% of tech-led suppliers. Great expectations, indeed.

On the other hand, only 37% of research-oriented professionals on the brand side completely or mostly agree, and only about one-third of service-led suppliers with 21 to 500 FTE express the same belief in agentic purchasing’s ultimate dominance. Among smaller service-led suppliers, only 19% share this belief. Should they be concerned?

Believers Control More Current Purchasing Power 

Of those who believe agentic purchasing will one day dominate, 27% on the brand side have research project budgets in excess of $15MM compared to just 7% of those who agree very little or not at all. While large budgets skew toward believers, budgets of $3MM or less skew toward those who are more doubtful, 59% to 38%. Those who only somewhat agree are in the middle of both ranges: 48% have budgets of $3MM or less, and 15% have budgets in excess of $15MM.

Brand-side believers have both purchasing power and momentum. In the past year, 56% of believers saw research project budget increases compared to just 26% of skeptics. Also, those reporting staff size increases outnumbered doubters by 39% to 22%, and they also report more outsourcing increases by 39% to 14%. Further, 37% of believers expect to increase outsourcing next year, compared to 16% of doubters.

Supplier Believers are Larger and Commercially Stronger 

Among service-led suppliers, 32% of those who believe agentic purchasing will dominate have more than 1,000 FTE compared to just 12% of skeptics. Those with 10 FTE or fewer compose 37% of doubters but only 13% of believers. Belief in an agent-dominated future skews toward larger suppliers, which presumably account for a disproportionate share of industry receivables and payables.

In fact, 60% of the believers reported revenue increases last year compared to just 46% of skeptics, who had more revenue decreases (26% to 15%). In addition to leading in receivables, believers also lead in terms of expecting outsourcing to other suppliers to increase, 46% to 13%. Revenue and spending momentum is clearly on the side of those who expect most purchases to become agentic. If your business depends on outsourcing from larger suppliers, this might be a good time to audit your technological compatibility with them.

Present Purchasing-Agent Use Remains Limited 

GRIT asked participants if most research/insights services purchases would be made by agents “in the next few years,” but we also asked what they are doing now. When asked whether they currently use AI agents for selecting or shortlisting suppliers or partners for research or analytics, just 28% of believers on the brand side said they do, and 17% of service-led supplier believers agreed. Time will tell us whether or not that expectation-practice gap closes and how quickly it does.

Current Adoption Suggests Evolution from Lower-Risk to higher-Risk Applications

Use of agents for making purchases isn’t common today, but that doesn’t mean use of agents is also rare. On the brand-side, more than 60% of believers currently use AI agents for analyzing or modeling data (77%), preparing and integrating data (69%), and creating or updating reports, dashboards, or summaries (69%). Among service-led supplier believers, most currently use agents for analyzing or modeling data (69%) and creating or updating reports, dashboards, or summaries (57%); nearly half use it for preparing and integrating data (48%). Agentic AI is common for tasks that can be checked and redone with relatively little risk, but less common where the stakes are higher.

For comparison, of the three tasks that exceed 60% use of agents among brand-side believers, only creating or updating reports, dashboards, or summaries (53%) and analyzing or modeling data (52%) reach a majority among skeptics. Among service-led supplier skeptics, only creating or updating reports, dashboards, or summaries (44%) eclipses 40%. The doubters are also adopting agents, just at a lower current rate than the believers.

Supplier Discoverability and Shortlist Status May Increase in Importance

Agentic purchasing may not simply automate existing selection processes; it may privilege the criteria that are easiest to formalize. According to the 2026 GRIT Insights Practice Report, the key criteria for supplier selection generally include all or some of:

• Data quality
• Service quality
• General pricing 
• Reputation 
• Relationship 
• Use of tech in research/analysis

In particular, data quality, service quality, and general pricing are universal criteria. Of these, pricing and technology use can be specified relatively cleanly. Data quality may sometimes be reducible to measurable standards. However, service quality, reputation and relationships are more abstract and context-dependent.

In an open discovery scenario, the agent searches broadly and evaluates possible suppliers that have been previously indexed. Under constrained selection, the agent evaluates or transacts among an approved shortlist. If most enterprise agents operate within approved vendor lists, relationships and reputation remain crucial because they determine who gets prequalified. If agents search openly, structured discoverability becomes much more important.

The most sure-handed way to be found and to be qualified is to be on the buyer’s shortlist to begin with. This is not a new idea, but it may be more urgent in an agent-dominated market. A supplier cannot assume an agent will find them or that it will evaluate them positively on abstract criteria, but this issue becomes irrelevant if the buyer has already shortlisted you and told the agent you are pre-qualified on service, reputation, and relationship. 

What Can We Make of This?

We can’t assume just because someone thinks agentic purchasing will be common that they will use it themselves – they may just be super-cynical about their own preferences. However, we can hypothesize that agentic purchasing of insights services will become common, especially if believers become users and usage follows the money. Belief is stronger where current use is higher, and believers likely control more money and volume.

We can also hypothesize that agents might not affect all purchases equally, at least until the practice diffuses into some future state of equilibrium. We see some evidence that agents are adopted first in lower-risk situations, suggesting purchasing agents may spread more quickly for offerings that are standardized, easily specified and readily compared than for highly customized services. GRIT evidence suggests the time horizon for agents to dominate the purchases of complex offerings may be farther out than for “off-the-shelf” purchases, although we can’t say how much farther.

Across GRIT segments, we see varying levels of belief and skepticism about the future of purchase agent dominance, but it looks like the big money is expecting to back it eventually. For now, even skeptics among suppliers might want to prepare agent-facing offering pages that describe their wares and document criteria and parameters buyers would be likely to specify. And every supplier should continue to do what they’ve always had to do: know your clients and potential clients as well as you can and position yourself to be on the shortlists that are given to agents.

The next GRIT Business & Innovation survey will build on these findings, probing issues such as approved vendor lists, transaction autonomy, the types of purchases handled by agents, and so on. Stay tuned.