Function cockpit
Research & Development
Agents run research operations. Scientists own hypotheses and conclusions.
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
0
Touchless
0%
Human review
0%
Override rate
0%
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
Eleven decomposed sub-functions, each with its own estate rating
Sub-functions
Research & Development, decomposed
Nobody automates a function. They automate the sub-functions inside it, at different speeds, with different ceilings. Each one carries its own orchestrator, its own human owner, and its own honest limit.
Autonomy ceiling range
64%→91%
The binding constraint here is Innovation Portfolio — capped at 64%.
Agents assemble evidence and even recommend killing a program. Stopping or funding research is a capital decision with careers attached, so a committee owns it.
Discovery Orchestrator · human owner Director, Discovery Research
estate rating
Owns research thread work end to end inside Research & Development. Runs 3 lines across 17 stations, drains the repeatable share into no shared tower, and holds 2 human gates.
Literature synthesis and hypothesis generation are exactly where agents outperform — they read everything. A scientist validates before anything becomes a program.
Units / wk
2,471
Agents
7
Lines
3
Cost / unit
$17
was $50
Innovation Portfolio Orchestrator · human owner VP Innovation
estate rating
Owns program work end to end inside Research & Development. Runs 3 lines across 16 stations, drains the repeatable share into no shared tower, and holds 4 human gates.
Agents assemble evidence and even recommend killing a program. Stopping or funding research is a capital decision with careers attached, so a committee owns it.
Units / wk
11,992
Agents
8
Lines
3
Cost / unit
$5.54
was $12
Experiment Design Orchestrator · human owner Director, Research Methods
estate rating
Owns protocol work end to end inside Research & Development. Runs 3 lines across 16 stations, drains the repeatable share into the C2L tower, and holds 4 human gates.
Protocol drafting and power analysis are agentic. Ethics clearance and scientific validity judgments are personally owned by a qualified researcher.
Units / wk
10,635
Agents
8
Lines
3
Cost / unit
$17
was $45
Lab Operations Orchestrator · human owner Director, Laboratory Operations
estate rating
Owns lab request work end to end inside Research & Development. Runs 3 lines across 15 stations, drains the repeatable share into the C2L tower, and holds 1 human gates.
Scheduling, replenishment and instrument monitoring are pure logistics and run touchless. Calibration certificates carry a technician signature.
Units / wk
9,721
Agents
8
Lines
3
Cost / unit
$2.96
was $13
Modeling Orchestrator · human owner Director, Data Science
estate rating
Owns model work end to end inside Research & Development. Runs 3 lines across 19 stations, drains the repeatable share into the DATA-ANALYTICS tower, and holds 4 human gates.
Training, validation and drift detection are fully agentic. Promoting a model to production requires a named reviewer, because a model in production is a decision-maker.
Units / wk
15,032
Agents
8
Lines
3
Cost / unit
$14
was $46
Prototype Orchestrator · human owner Director, Product Development
estate rating
Owns prototype work end to end inside Research & Development. Runs 3 lines across 17 stations, drains the repeatable share into the C2L tower, and holds 4 human gates.
Test execution, iteration tracking and feedback synthesis are agentic. Deciding to scale a pilot commits capital and operations capacity.
Units / wk
5,131
Agents
8
Lines
3
Cost / unit
$9.63
was $27
IP Capture Orchestrator · human owner Director, IP Capture
estate rating
Owns disclosure work end to end inside Research & Development. Runs 3 lines across 16 stations, drains the repeatable share into the C2L tower, and holds 3 human gates.
Detecting patentable work inside research output is something agents do continuously and people forget to do. Inventorship is a legal attribution a human confirms.
Units / wk
1,100
Agents
8
Lines
3
Cost / unit
$13
was $44
Regulatory Science Orchestrator · human owner VP Regulatory Science
estate rating
Owns submission work end to end inside Research & Development. Runs 3 lines across 19 stations, drains the repeatable share into the C2L tower, and holds 6 human gates.
Requirement mapping and evidence assembly are agentic at scale. Everything filed with a regulator is signed by a named officer with personal accountability.
Units / wk
22,214
Agents
8
Lines
3
Cost / unit
$15
was $40
Collaboration Orchestrator · human owner Director, External Innovation
estate rating
Owns collaboration work end to end inside Research & Development. Runs 3 lines across 19 stations, drains the repeatable share into no shared tower, and holds 6 human gates.
Scouting and milestone tracking are agentic. Anything that shares data or publishes results outside the company is cleared by counsel every time.
Units / wk
6,778
Agents
8
Lines
3
Cost / unit
$3.65
was $12
Technology Scouting Orchestrator · human owner Director, Technology Scouting
estate rating
Owns candidate work end to end inside Research & Development. Runs 3 lines across 16 stations, drains the repeatable share into no shared tower, and holds 3 human gates.
Landscape scanning and screening are continuous and touchless. Engaging a target company is a corporate development act with a named owner.
Units / wk
3,124
Agents
7
Lines
3
Cost / unit
$13
was $46
R&D Analytics Orchestrator · human owner Director, R&D Analytics
estate rating
Owns analysis work end to end inside Research & Development. Runs 3 lines across 14 stations, drains the repeatable share into the DATA-ANALYTICS tower, and holds 3 human gates.
Health scoring and spend attribution are fully agentic. The readout is reviewed by a leader before it influences funding conversations.
Units / wk
173
Agents
8
Lines
3
Cost / unit
$557
was $2K
How to read an estate rating
The rating is not a maturity badge. It scores seven control dimensions — control coverage, evidence quality, override discipline, data integrity, recovery readiness, cost transparency, and judgment readiness. Judgment readiness runs deliberately against autonomy: a sub-function that has pushed automation hardest usually scores worst on whether a human could still take the decision back.