Function cockpit

Research & Development

Agents run research operations. Scientists own hypotheses and conclusions.

Autonomy0% / 74%

Phase: Phase 2 — Supervised execution

Connect scientific programs to resources, evidence and decision gates: retrieval, scheduling, data integrity, simulation and reproducibility are agent work; hypothesis selection, safety and scientific conclusions remain human.

Trust score

Touchless

Human review

Override rate

Capacity released

20-35% research-operating capacity (no presumption that success rates improve)

Accountable human

Chief Scientific Officer

CoS agent: R&D Chief of Staff

Strategy, policy, observe, workflow, and AI GBS side by side

Five layers

What Research & Development actually runs

A function is not one thing. more

Strategy

18 agents

Agents that take positions, argue them, and get held to the outcome.

15 positions on the record

4 ratified, 3 in challenge

Policy

5 agents

The written rules every other agent is bound by, with version and owner.

34 binding rules

29 active, 5 superseded or retired

Observe

19 agents

Continuous sensing of the outside world and of the agents themselves.

19 watchers running

They watch the outside world and the agents themselves

Workflow

35 agents

Doing the work: lines, stations, and live units moving through them.

33 lines, 178 live units

11 held for a person right now

AI GBS

9 agents

Repeatable transaction processing run once for the whole enterprise.

2 shared towers

54 stations on this function run inside AI GBS

Strategy layer — positions on the record

Every position names its author, the agent paid to argue against it, and the conditions that would break it.

Research Direction Position

Superseded

Hold the ceiling at A3 for discovery research. more

author Hypothesis Generation Agentchallenged by Discovery Research Challenger Agentratifier Director, Discovery Researchconfidence 79%

Discovery Research Operating Position

Rejected

Keep the repeatable research thread volume in the spine and pull the exception tail back into the function. more

author Research Direction Agentchallenged by Discovery Research Challenger Agentratifier Director, Discovery Researchconfidence 66%

Portfolio Bets Position

Draft

Hold the ceiling at A2 for innovation portfolio. more

author Portfolio Bets Agentchallenged by Innovation Portfolio Challenger Agentratifier VP Innovationconfidence 83%

Innovation Portfolio Operating Position

Superseded

Keep the repeatable program volume in the spine and pull the exception tail back into the function. more

author Portfolio Bets Agentchallenged by Innovation Portfolio Challenger Agentratifier VP Innovationconfidence 66%

Experiment Design Operating Position

Ratified

Hold the ceiling at A3 for experiment design. more

author Experiment Design Orchestratorchallenged by Experiment Design Challenger Agentratifier Director, Research Methodsconfidence 64%outcome held

Lab Operations Operating Position

In challenge

Hold the ceiling at A4 for lab operations. more

author Lab Operations Orchestratorchallenged by Lab Operations Challenger Agentratifier Director, Laboratory Operationsconfidence 66%

15 positions across 11 sub-functions. Open a sub-function to see dissent, alternatives considered, and break conditions.

Policy layer — what binds the agents

Versioned rules with an owning agent, a human approver, and the agents they constrain.

Discovery Research Decision Envelope

vv2.7

Agents in this lane may act without a human when the research thread sits inside the stated value, risk and confidence envelope. more

owner Discovery Orchestratorapproved by Director, Discovery ResearchGlobal · EUbinds 3 agents

Discovery Research Evidence Standard

vv2.3

Every autonomous decision writes inputs, the rule version applied, the model and prompt version, the output and a reversal path. more

owner Discovery Orchestratorapproved by Director, Discovery ResearchLATAM · USbinds 3 agents

Discovery Research Escalation Rule

vv1.4

Escalation is mandatory when confidence falls below the floor, when two options sit inside the confidence band, or when a break condition on a ratified position fires. more

owner Discovery Orchestratorapproved by Director, Discovery ResearchUS · Canadabinds 3 agents

Innovation Portfolio Decision Envelope

vv5.1

Agents in this lane may act without a human when the program sits inside the stated value, risk and confidence envelope. more

owner Innovation Portfolio Orchestratorapproved by VP InnovationEU · UKbinds 4 agents

Innovation Portfolio Evidence Standard

vv2.2

Every autonomous decision writes inputs, the rule version applied, the model and prompt version, the output and a reversal path. more

owner Innovation Portfolio Orchestratorapproved by VP InnovationUS · Canadabinds 4 agents

Experiment Design Decision Envelope

vv1.6

Agents in this lane may act without a human when the protocol sits inside the stated value, risk and confidence envelope. more

owner Protocol Compliance Agentapproved by Director, Research MethodsEU · UKbinds 3 agents

34 rules in the register. An agent cannot act outside the version in force.

Workflow layer — the lines doing the work

Each line is a workflow. more

120 in time47 late11 late and material

Line	Sub-function	Kind	Units / wk	Touchless	Timetable	Actual	Stations
RD1A Discovery Sprint	Discovery Research	analytical	40	61%	82.2h	80.0h	7
RD1B Research Direction Position	Discovery Research	judgment	5	40%	334.7h	293.1h	6
RD1C Prior Work Sweep	Discovery Research	transactional	2,426	74%	21.0h	23.8h	4
RD2A Stage Gate Review	Innovation Portfolio	transactional	5,106	67%	12.0h	11.8h	6
RD2B Kill Signal Escalation	Innovation Portfolio	transactional	6,881	58%	4.5h	4.5h	5
RD2C Portfolio Bets Position	Innovation Portfolio	judgment	5	28%	198.6h	248.6h	5
RD3A Protocol Development	Experiment Design	transactional	430	70%	10.2h	12.0h	7
RD3B Protocol Deviation	Experiment Design	transactional	7,292	67%	22.2h	19.7h	5
RD3C Design Standard Refresh	Experiment Design	transactional	2,913	68%	24.3h	19.6h	4
RD4A Instrument Booking	Lab Operations	transactional	5,522	93%	7.1h	6.5h	5
RD4B Consumables Replenishment	Lab Operations	transactional	1,317	87%	7.5h	6.3h	4
RD4C Calibration Cycle	Lab Operations	transactional	2,882	88%	6.4h	5.5h	6
RD5A Model Development	Data Science & Modeling	transactional	2,326	76%	16.1h	13.6h	8
RD5B Drift Response	Data Science & Modeling	transactional	7,707	82%	17.4h	14.6h	6
RD5C Model Retirement	Data Science & Modeling	transactional	4,999	86%	7.2h	8.3h	5
RD6A Prototype Cycle	Prototype & Pilot	transactional	4,644	77%	16.9h	17.2h	7
RD6B Pilot Deployment	Prototype & Pilot	transactional	475	77%	12.0h	11.6h	6
RD6C Scale Readiness Assessment	Prototype & Pilot	analytical	12	49%	49.0h	50.9h	4
RD7A Invention Capture	IP Creation & Capture	transactional	634	73%	15.6h	19.1h	6
RD7B Trade Secret Registration	IP Creation & Capture	transactional	462	73%	13.6h	15.2h	5
RD7C IP Strategy Position	IP Creation & Capture	judgment	4	48%	455.0h	552.4h	5
RD8A Regulatory Submission	Regulatory Science	transactional	7,459	64%	2.6h	3.2h	8
RD8B Standards Change Response	Regulatory Science	transactional	7,276	63%	13.1h	13.4h	6
RD8C Compliance Testing	Regulatory Science	transactional	7,479	64%	11.0h	10.6h	5
RD9A Research Collaboration Setup	External Collaboration	transactional	3,828	65%	22.8h	18.1h	7
RD9B Data Sharing Request	External Collaboration	transactional	2,022	70%	6.1h	7.3h	6
RD9C Publication Clearance	External Collaboration	transactional	928	66%	14.4h	16.4h	6
RD10A Scouting Cycle	Technology Scouting	analytical	25	67%	151.9h	174.8h	6
RD10B Build-Buy-Partner Position	Technology Scouting	judgment	2	39%	436.0h	471.7h	5
RD10C Technical Diligence	Technology Scouting	transactional	3,097	80%	10.4h	13.4h	5
RD11A Quarterly Portfolio Read	Portfolio Analytics	analytical	64	66%	82.2h	75.2h	6
RD11B Success Rate Analysis	Portfolio Analytics	analytical	76	73%	130.9h	122.3h	4
RD11C Benchmark Study	Portfolio Analytics	analytical	33	62%	186.4h	198.9h	4

Observe layer — the watchers

Sensing agents and the agents that supervise other agents.

Discovery Orchestrator

Owns the discovery research lane end to end. Sequences the other agents, holds the timetable, and decides what surfaces to a human.

A396%

Discovery Research Challenger Agent

Adversarial reviewer for discovery research. Argues the opposite case on every position and flags where the evidence does not carry the claim.

A297%

Innovation Portfolio Orchestrator

Owns the innovation portfolio lane end to end. Sequences the other agents, holds the timetable, and decides what surfaces to a human.

A296%

Kill Signal Agent

Watches innovation portfolio continuously. Detects drift, quantifies it, and routes what matters without waiting for a reporting cycle.

A290%

Innovation Portfolio Challenger Agent

Adversarial reviewer for innovation portfolio. Argues the opposite case on every position and flags where the evidence does not carry the claim.

A287%

Experiment Design Orchestrator

Owns the experiment design lane end to end. Sequences the other agents, holds the timetable, and decides what surfaces to a human.

A387%

Bias Detection Agent

Watches experiment design continuously. Detects drift, quantifies it, and routes what matters without waiting for a reporting cycle.

A388%

Experiment Design Challenger Agent

Adversarial reviewer for experiment design. Argues the opposite case on every position and flags where the evidence does not carry the claim.

A288%

AI GBS layer — the shared spine

Repeatable transaction processing does not belong to this function. more

Open towers

c2l48,801 units / wk routed

Experiment Design · Lab Operations · Prototype & Pilot · IP Creation & Capture · Regulatory Science

data-analytics15,205 units / wk routed

Data Science & Modeling · Portfolio Analytics

Stations running inside AI GBS

54 of 184 stations(29% of the function's stations)

Read this page as the honest answer to “what would an 80% agentic Research & Development look like?” The strategy and policy layers are where the argument happens. The workflow and AI GBS layers are where the volume goes. The observe layer is the only reason the other four can be trusted.

Sub-function breakdown