Agent OneAgent One

00How it works

Six agents. One continuous discovery pipeline.

From the first stakeholder interview through to a ranked, evidence-linked roadmap. The stages overlap by design — analysis starts before interviewing finishes.

01The pipeline

Five stages, running together.

  1. Setup

    Research

    Scaffolds the engagement from your documents and context.

  2. Interview

    Scout

    Talks to every stakeholder, on their schedule, on their channel.

  3. Analyse

    Lens

    Builds live service blueprints as responses arrive.

  4. Synthesise

    Arc

    Turns findings into a ranked, evidence-linked roadmap.

  5. Deliver

    Pulse + Forge

    Keep operators steering and deliverables sharp.

02Where the time goes

The bottleneck was never the interviewing.

Traditional discovery spends most of its calendar waiting for synthesis that happens out of sight. Running analysis alongside interviewing removes the queue, not the rigour.

Traditional discovery

first output ≈ week 12

  1. Schedule workshops

    weeks 1–3

  2. Run workshops

    weeks 3–6

  3. Write up notes

    weeks 6–8

  4. Synthesise offline

    nothing visible

  5. Deliver report

    week 12

Agentic Discovery

first output ≈ day 3

  1. Interviews open

    day 1, all stakeholders

  2. Analysis runs continuously

    no queue

  3. Findings land

    from day 3

  4. Roadmap on request

    any time after

Illustrative. Durations vary with scope and stakeholder availability.

03Each stage in detail

What actually happens, week by week.

The stages are listed in order, but they overlap: Lens is analysing while Scout is still interviewing, and Arc can be asked for a roadmap at any point.

  1. Stage 01

    Research

    Setup: load the engagement

    We identify stakeholders across the process being studied. Context can be entered directly or scaffolded from documents you already have — org charts, process maps, existing reports. For each stakeholder: role, department, topics to cover, and priority.

    • Company and industry context researched up front
    • Engagement scaffolded from uploaded documents
    • Blueprint structure and discovery goals defined
    • Coverage plan per stakeholder and function

    At this point you have

    A scoped engagement with goals, a coverage plan and loaded context — before anyone is interviewed.

  2. Stage 02

    Scout

    Interview: talk to everyone, on their schedule

    Scout reaches stakeholders through web voice, Microsoft Teams, phone and text chat. It opens with one or two questions, follows the natural shape of the conversation, tracks what has been covered, asks for relevant documents, and follows up over days or weeks. A stakeholder can pause mid-thought on Tuesday and resume on Thursday without losing the thread.

    • One or two questions at a time
    • Redirect controls let stakeholders steer
    • Per-stakeholder memory across sessions
    • Document upload and analysis
    • Commitment tracking for follow-ups
    • No scheduling, no workshops

    At this point you have

    An interview base that grows daily, with coverage you can see per stakeholder and per function.

  3. Stage 03

    Lens

    Analyse: in real time, not post-engagement

    As data arrives, each message is embedded into semantic memory and Lens is triggered when analysis would be useful. It builds and updates six-layer service blueprints, types each finding, scores confidence per cell, and generates guidance for Scout — so the next interview is sharper than the last.

    • Six layers: physical evidence, customer actions, frontstage, backstage, support processes, analytics
    • Cross-stakeholder pattern detection
    • Findings typed as strengths, observations, challenges, pain points
    • Goals close automatically once evidence passes threshold
    • Guidance fed back into Scout’s next questions
    • Confidence scored per cell

    At this point you have

    A live six-layer service blueprint and a findings register, both updating as interviews happen.

  4. Stage 04

    Arc

    Synthesise: proposals grounded in evidence

    Arc works over Lens’s structured output, consolidating and de-duplicating findings before proposing practical interventions. Each proposal is scored on impact, feasibility and dependencies, then placed on an effort-versus-value matrix. Every recommendation traces back to the findings that support it.

    • Impact, feasibility and dependency scoring
    • Effort-versus-value prioritisation matrix
    • Evidence traceability for every proposal
    • Category grouping across pain points and opportunities
    • Non-tooling proposals where those are the real answer

    At this point you have

    A ranked, evidence-linked solution roadmap.

  5. Support

    Pulse + Forge

    Steer and deliver

    Pulse gives the operating team a dashboard agent for asking questions, checking coverage and surfacing findings as they emerge. Forge assists with structured blueprint editing so deliverables can be shaped for a board without losing the evidence chain underneath.

    • Daily engagement reports
    • Stakeholder coverage tracking
    • Friction signals surfaced early
    • Anonymisation for client-facing output
    • Themed finding grouping for presentation

    At this point you have

    A live console for steering, and deliverables ready to present.

service blueprint · six layers × five phases

schematic

Intake

Triage

Assess

Resolve

Close

Physical evidence

Customer actions

Frontstage

Backstage

Support processes

Analytics

  • Complete
  • In progress
  • Not started
What Lens produces: a phase-by-swimlane grid where every cell carries its own goals, evidence and confidence. Coverage accrues per cell as interviews arrive, so a gap is visible while there is still time to close it.

04The agent team

Three run the pipeline. Three support it.

Six specialised agents, each with one job, coordinated by an orchestrator. This is the full roster — you will see the same six everywhere on this site.

Scout

01 · Core

The Interviewer

“Talks to everyone; forgets nothing.”

External · stakeholders

Scout conducts semi-structured interviews through web voice, Microsoft Teams, phone and text chat. It manages the whole interview lifecycle: opening conversations, following up over days or weeks, tracking what has been covered, and requesting the documents it needs.

Key capabilities

  • Multi-channel: web voice, Teams, phone, text chat
  • One or two questions at a time, never overwhelming
  • Redirect controls, so stakeholders can steer
  • Per-stakeholder memory across sessions
  • Commitment tracking for follow-ups
  • Document upload and analysis

Lens

02 · Core

The Analyst

“Sees the pattern while the interview is still running.”

Internal · background

As interview data arrives, Lens analyses it against the existing blueprint, builds structured six-layer service blueprints, identifies findings, and generates guidance telling Scout what to ask next — so each interview is sharper than the last.

Key capabilities

  • Six-layer service blueprints
  • Cross-stakeholder pattern detection
  • Findings typed as strengths, observations, challenges and pain points
  • Automatic goal completion once evidence passes threshold
  • Guidance generation for Scout follow-ups
  • Confidence scoring per cell

Arc

03 · Core

The Strategist

“Turns evidence into the roadmap.”

Internal · on demand

Arc works over the structured output Lens produces, consolidating and de-duplicating findings before generating a ranked solution roadmap. Every proposal is grounded in the specific findings that support it.

Key capabilities

  • Impact, feasibility and dependency scoring
  • Effort-versus-value prioritisation matrix
  • Every proposal linked to its supporting findings
  • Category grouping across pain points and opportunities
  • Non-tooling proposals: governance, role clarity, ownership

Research

04 · Support

The Scholar

“Reads everything before Scout says hello.”

Internal · setup

Research supports engagement setup by investigating company context, industry background and relevant frameworks. It scaffolds new engagements from uploaded documents and pre-populates stakeholder context.

Key capabilities

  • Company and industry research
  • Document scaffold flows
  • Stakeholder context pre-population
  • Blueprint template selection

Pulse

05 · Support

The Navigator

“Keeps operators on the pulse of the engagement.”

Internal · dashboard

Pulse gives the operating team a conversational interface for monitoring progress, asking questions about the engagement, and surfacing findings as they emerge.

Key capabilities

  • Engagement-scoped access — it sees only what it should
  • Daily engagement reports
  • Stakeholder coverage tracking
  • Friction signal identification
  • Cross-engagement analytics for partners

Forge

06 · Support

The Craftsman

“Shapes evidence into deliverables.”

Internal · editing

Forge assists with structured blueprint editing, so the team can shape the narrative before it reaches a client — without losing the evidence chain underneath it.

Key capabilities

  • Cell creation and editing
  • Content structuring for delivery mode
  • Anonymisation for client-facing outputs
  • Themed finding grouping for presentation

05Under the hood

Built on enterprise infrastructure.

Four decisions carry the platform: which models run, where the memory lives, which channels stakeholders use, and which cloud the cell sits in.

Models

Frontier models for analysis and synthesis, and a native real-time audio model for voice. Every call is routed through Google Cloud Vertex AI, and client data is never used for training.

Data

PostgreSQL with pgvector for semantic memory, held inside the client’s isolated cell. No other engagement can reach it.

Channels

Microsoft Teams, web voice, phone and text chat — so stakeholders use what they already have open.

Platform

Google Cloud Platform, cell-based deployment, data stored in Australian regions and locked there by organisation policy.

The multi-agent orchestration, real-time analysis pipeline and structured discovery output system are the subject of a pending patent (USPTO Application 19/661,143). The full security architecture.

See the agents on your problem.

Thirty minutes, a live engagement on screen, and a map of Agentic Discovery onto your engagement.

What Agentic Discovery is