Every procurement conference now has an agentic AI keynote. Vendors demonstrate agents that draft RFPs, score bids and negotiate with suppliers. Analysts publish adoption forecasts. It is easy to leave the room believing that autonomous sourcing is already mainstream.
The data tells a more sober story. The Hackett Group’s The Autonomous Sourcing Index scores organizations on a 10-point scale, and the median score is 3.6.
How the Index works
The Index measures four things: deployment maturity (adoption from experimentation through scaled use), sourcing and orchestration capability (breadth and coordination across nine sourcing activities), readiness (governance, data, skills and change capacity) and success factors (practices that sustain adoption and measurable value).
Scores fall into four stages:
- Nascent (0–2.5): awareness and pilots
- Developing (2.6–5.0): repeatable use, with foundations forming
- Scaling (5.1–7.5): integrated delivery expanding
- Highly evolved (7.6–10.0): enterprise optimization and impact
The respondents are senior sourcing leaders, and the sample leans large: 78% report $1 billion or more in annual revenue, 49% have sourcing teams of 100 or more full-time employees and 45% are executive-level leaders. They are spread across Europe, North America and Asia Pacific, and across industries from manufacturing to healthcare to financial services.
The number that should make you pause
The bottom quartile scores 2.2. The median is 3.6. The top quartile is 4.7.
Read that last figure again. The reported top-quartile benchmark is 4.7, which is still inside the “developing” band. Put differently, three-quarters of the market scores below a threshold that the framework itself describes as foundations forming.
In total, 78% of organizations fall within the nascent or developing stages. Most of the remainder are scaling, and very few are highly evolved.
If your own organization feels behind the curve, this is worth sitting with. The curve is lower than the conference circuit suggests.
Maturity is not the same as autonomy
One of the more important details in the research is how it defines the score. Maturity reflects the ability to deploy and orchestrate agentic AI at scale, with appropriate data, integration, governance and human oversight. It is not a measure of autonomy alone.
That matters because autonomy is where the hype concentrates, and the data shows how far reality sits from it. When respondents describe how autonomous their sourcing is, 25% say they do not use AI at all, 23% describe it as assisted and 25% as guided. Another 20% use AI with human approval, 7% say it operates mostly independently and none describe sourcing as fully autonomous.
Fully autonomous sourcing remains near 0%. A high maturity score, in other words, is not about removing people. It is about building the foundations that let agents do more, safely, with people retaining authority over consequential decisions.
Deployment is real, but reach is thin
The deployment picture reinforces this. Among respondents, 3% have deployed agentic AI enterprise-wide and 17% broadly across sourcing. The largest group, 30%, is in limited production. Another 23% are piloting, 22% are exploring and 5% have no deployment at all.
Capability tells a similar story. Across nine sourcing activities, advanced support is the exception. In bid analysis and supplier selection, one of the strongest areas, just 23% of organizations deliver extensive or end-to-end agentic support. The more complex the activity, the thinner the coverage tends to be.
The research frames the next frontier clearly: governed agentic execution in more complex sourcing activities. Insight and analysis are where agents have gained traction first. Execution inside guardrails is where the work remains.
What a low median means for you
A median of 3.6 can be read two ways.
The pessimistic reading is that the market is far from where the marketing says it is. The optimistic reading is better: you are probably not as far behind as you fear, and the gap between you and the leaders is smaller and more closable than it looks.
The Index offers a practical way to think about it. Rather than chasing a headline label such as “autonomous,” it asks which stage you are in and what the next stage requires. The report’s maturity playbook sets out a consistent sequence:
- Nascent: prove it. Select one high-volume, low-risk activity and define trusted data, human approvals, guardrails and audit evidence.
- Developing: standardize it. Standardize two or three proven workflows, connect core applications and data, and retire off-system workarounds.
- Scaling: expand it. Extend proven flows across categories, instrument performance and data quality, and build outcome ownership.
- Highly evolved: optimize it. Operate an integrated agent ecosystem with governed autonomy and tie funding to verified outcomes.
Each stage also names the accountable roles and the measures that matter, which turns a vague ambition into a plan.
Where technology partners fit
Moving from stage to stage is as much about architecture as intent. Procurement teams in the developing stage often have several tools that do not talk to one another. Platforms built to connect intake, sourcing, contracts and analytics on shared data can shorten the path.
Zycus, the Index’s partner, describes its approach as moving enterprises from Intake to Outcomes on one platform. Its Merlin Agentic AI Platform includes agents such as Merlin Intake and the Autonomous Negotiation Agent, built on an integrated source-to-pay core. Teams that want to build their own agents can also look at its low-code environment, AppXtend. As always, evaluate any vendor against your own data, integration and governance needs, since the Index itself is vendor-neutral and The Hackett Group does not endorse any product depicted in its research.
Know your number
The most useful first step is also the simplest: find out where you actually stand. The report recommends completing the Autonomous Sourcing self-assessment and comparing your position with the 3.6 median and 4.7 top-quartile line, dimension by dimension. The output is a prioritized gap map and a single target index point to aim for.
That clarity helps with executive conversations too. “We need more AI” is a hard case to fund. “For example, if our hypothetical score were 3.1 against a median of 3.6, closing the gap in data and integration could be a concrete next-stage priority” is a case a CFO can engage with.
The full report goes well beyond the headline scores, including how capability varies across nine sourcing activities, how readiness breaks down by strategy, organization, process and skills, and what the most successful organizations do differently.
Download The Autonomous Sourcing Index and find out where you stand.
For a complementary view of broader procurement adoption patterns, explore The Hackett Group Agentic AI in Procurement Adoption Index. This is a separate study from the Autonomous Sourcing Index.
For a broader view of how procurement leaders can plan AI readiness, see the Gartner report on agentic AI readiness. This is a separate analyst resource, not a component of the Autonomous Sourcing Index.

