Agent-conducted
AI agents interview stakeholders directly — not humans using AI tools to write up their notes faster.
00Defining a new category
Agentic Discovery
The use of coordinated AI agent teams to conduct stakeholder-driven organisational discovery: interviewing people, analysing their input in real time, building structured process intelligence, and delivering recommendations traceable to evidence.
Organisations arrive at this looking for process mapping, service design or service blueprinting, a current-state assessment, or the discovery phase of a digital or AI transformation. Agentic Discovery is how that work gets done, not a different piece of work — and it is organisational discovery, not agentic search over documents.
01Why it is a category
A category that did not exist until now.
Every enterprise transformation starts with discovery. Consultants interview stakeholders, map processes and synthesise findings. This has been done the same way for decades: manually, slowly, and at a scale limited by the number of humans you can deploy.
Agentic Discovery replaces the manual front end of that process. Instead of the fieldwork moving at the speed of a shared calendar, you deploy a team of AI agents that interview every relevant stakeholder on their own schedule, analyse every response as it arrives, build structured process maps in real time, and deliver recommendations grounded in traceable evidence.
The shift is in the format, not in the calibre of the people running it. Discovery has been rate-limited by how many conversations a human team can hold at once, and everything downstream inherits that limit. Agentic Discovery removes it, and the human role moves up: from interviewer and note-taker to strategic director and decision-maker.
02What defines it
Six properties, and AgentOne holds all six.
Miss any one of them and you are describing something else — a chatbot, an assessment, or process mining with a friendlier interface.
Stakeholder-driven
The primary input is human knowledge, not system logs or documents. Undocumented work is the point, not a gap.
Real-time analysis
Every piece of input is analysed as it arrives, not after fieldwork closes.
Structured output
Service blueprints with evidence chains, not transcripts or summary notes.
Continuous and asynchronous
Stakeholders engage on their own schedule, over days or weeks. Not a workshop.
Multi-agent coordination
Different agents handle interviewing, analysis, synthesis and steering, coordinated by an orchestrator.
03The landscape
Where Agentic Discovery sits.
Two axes define the field: what data drives the discovery, and who conducts it. Neither axis is about who is better at it.
Agent-conductedHuman-conducted
AgentOne
Agentic Discovery
1QLabs / AgentOne
AI agents interview stakeholders and analyse as they listen.
Agentic process management
UiPath / ServiceNow / Automation Anywhere
AI agents act on system data at speed. The knowledge in people’s heads stays there.
Workshop-based discovery
On-site workshops / interview notes / synthesis decks
People interview people, at human speed and human scale. The format sets the ceiling, not the practitioner.
Process mining
Celonis / SAP Signavio / ARIS
Analysts configure log analysis. Undocumented work is invisible to it.
Driven by human knowledgeDriven by system logs
The axes describe method, not quality, and the named vendors are examples of each category rather than a ranking. Workshop-based discovery is where the best discovery in the market is done today — it is also where AgentOne’s partners come from. The vertical axis is the one that moves: the firms in that quadrant deploy AgentOne inside their own projects to conduct the fieldwork agentically while keeping the judgment, the client and the recommendation.
04Side by side
Three approaches to understanding an organisation.
Workshop-based discovery, process mining and Agentic Discovery answer the same question by different means.
| Dimension | Workshop-based discovery | Process mining | Agentic Discovery |
|---|---|---|---|
| Evidence source | Workshop notes | System event logs | Direct stakeholder testimony |
| Who conducts it | A human team, sequentially | Software plus analysts | A coordinated AI agent team |
| Coverage | Whoever fits the calendar | Whatever the systems log | Every relevant stakeholder |
| Analysis timing | After fieldwork ends | After log extraction | Continuous, from day one |
| Time to first finding | Months | Weeks | Days |
| Traceability | Consultant synthesis | Log-level only | Every finding linked to source |
| Captures undocumented work | Partially, if someone asks | No | Yes — it is the whole point |
| Scalability | Linear with headcount | Dependent on data volume | Stakeholders in parallel |
05Boundaries
What Agentic Discovery is not.
The category is narrow on purpose. These are the five things it gets mistaken for.
Process mining with a chat interface
Process mining reads system logs. Agentic Discovery interviews people, which is where undocumented work actually lives.
A chatbot that asks questions
Scout conducts multi-session, multi-day interviews with redirect controls, commitment tracking and cross-stakeholder awareness.
AI-assisted consulting
An assistant helps a consultant write up the fieldwork faster. Agentic Discovery conducts the fieldwork itself, and hands the synthesis and the recommendation back to the consultant.
An assessment tool
Assessments score against a rubric. Agentic Discovery maps processes, finds patterns and proposes interventions.
Agentic search
Agentic search retrieves what an organisation has already written down. Agentic Discovery interviews people to establish what was never written down in the first place.
Automation discovery
Finding where to apply automation is a different problem. Agentic Discovery is the discovery work itself.
06Our point of view
A weak foundation produces weak outcomes.
Organisations make large transformation decisions based on discovery outputs they cannot verify, from processes they cannot trace, synthesised by people who cannot scale. The discovery phase is the foundation of everything that follows.
The bottleneck is not skill. It is the format. Human interviewers cannot hold hundreds of data points in memory across thirty stakeholders. They cannot analyse in real time. They cannot follow up asynchronously without burning social capital.
1QLabs built AgentOne to pioneer Agentic Discovery. Every organisation deserves discovery as rigorous as the decisions it informs.
Patent pending
The multi-agent orchestration, real-time analysis pipeline and structured discovery output system are the subject of a pending patent.
See Agentic Discovery in action.
AgentOne is running live projects today. Talk to us about what it could do for your organisation or your practice.