AI models with one defined role each
A specialist has exactly one role, for instance reviewing the identification strategy of a draft. Its output contract names the artifact it must produce, and a run fails the check if the artifact is missing. Specialists do not execute code themselves; the orchestrator runs their scripts and checks the results.
E2ER specialists (26) · Theory lab: 7 core roles and 60 field specialists
idea_developer
Develop RQ into paper plan.
Design, analysis and writing
used in 1 template · 1 published study · Examples: 3 example studiesliterature_scanner
Survey prior work.
Design, analysis and writing
used in 1 template · 1 published study · Examples: 3 example studies- Must write
- literature_review.md
- Skills
- base/researchersynthesis/context-builder
- Used in
- empirical · initialdid-staggered-panel · initialrd-platform-policy · initialiv-shift-share · initialevent-study-finance · initial
data_architect
Write data_dictionary.json.
Design, analysis and writing
used in 1 template · 1 published study · Examples: 3 example studiesidentification_strategist
Propose ID strategy.
Design, analysis and writing
used in 1 template · 1 published study · Examples: 4 example studies- Must write
- identification_strategy.md
- Skills
- causal-inference/judge-designscausal-inference/natural-experimentsreasoning/identificationcausal-inference/identification-spec-schema
- Used in
- empirical · initialdid-staggered-panel · initialrd-platform-policy · initialiv-shift-share · initialevent-study-finance · initial
econometrics_specialist
Specify the econometric model.
Design, analysis and writing
used in 1 template · 1 published study · Examples: 4 example studies- Must write
- econometric_spec.md
- Skills
- data/query-dataeconometrics/iv-estimationeconometrics/dideconometrics/panel-dataeconometrics/event-studyeconometrics/estimation-results-schema
- Used in
- empirical · initialdid-staggered-panel · initialrd-platform-policy · initialiv-shift-share · initialevent-study-finance · initial
data_analyst
the specialist that actually queries data
Design, analysis and writing
used in 1 template · 1 published study · Examples: 3 example studies- Must write
- data_summary.md
- Skills
- data/query-datadata/cleaningdata/figure-spececonometrics/panel-datadata/allium-clidata/allium-developer-apidata/yfinancedata/freddata/summary-statistics-schema
- Used in
- empirical · initialdid-staggered-panel · initialrd-platform-policy · initialiv-shift-share · initialevent-study-finance · initial
theory_specialist
writes model_spec.md (formal model in LaTeX with assumptions, derivations, propositions)
Design, analysis and writing
not used in published research yet- Must write
- model_spec.md
- Skills
- base/economistmodeling/game-theorymodeling/asset-pricingmath/proof-strategiesreasoning/identification
- Used in
- dispatched by the strategist when needed
paper_drafter
Draft the paper body.
Design, analysis and writing
used in 1 template · 1 published study · Examples: 3 example studiessection_writer
used with a specific section:<name> reference on iterations 2 and later
Design, analysis and writing
not used in published research yet- Must write
- paper_draft.tex
- Skills
- writing/paper-structurewriting/personal-stylereasoning/anti-slopwriting/cite-numbers-by-sourcedata/table-spec
- Used in
- dispatched by the strategist when needed
abstract_writer
Write the abstract.
Design, analysis and writing
1 published study- Must write
- abstract.tex
- Skills
- writing/abstractreasoning/anti-slopwriting/cite-numbers-by-source
- Used in
- dispatched by the strategist when needed
latex_formatter
Design, analysis and writing
not used in published research yet- Must write
- paper_draft.tex
- Skills
- latex/econ-modellatex/tables
- Used in
- dispatched by the strategist when needed
mechanism_reviewer
Review panel
used in 2 templates · 1 published study · Examples: 3 example studiestechnical_reviewer
Review panel
used in 2 templates · 1 published study · Examples: 3 example studiesliterature_reviewer
Review panel
used in 2 templates · 1 published study · Examples: 3 example studiesidentification_reviewer
Review panel
used in 2 templates · 1 published study · Examples: 3 example studies- Must write
- review_identification.md
- Skills
- causal-inference/sensitivityreview/technical-review
- Used in
- empirical · reviewpaper-review · reviewdid-staggered-panel · reviewrd-platform-policy · reviewiv-shift-share · reviewevent-study-finance · review
self_attacker
Adversarial specialist finds critical flaws (severity 1-10)
Design, analysis and writing
used in 1 template · 1 published study · Examples: 3 example studiespolish_numerics
Polish
used in 1 template · 1 published study · Examples: 3 example studies- Must write
- polish_numerics.md
- Skills
- data/cleaningreview/consistency-check
- Used in
- empirical · polishdid-staggered-panel · polishrd-platform-policy · polishiv-shift-share · polishevent-study-finance · polish
polish_institutions
Polish
used in 1 template · 1 published study · Examples: 3 example studies- Must write
- polish_institutions.md
- Skills
- base/economistdata/crypto-defi
- Used in
- empirical · polishdid-staggered-panel · polishrd-platform-policy · polishiv-shift-share · polishevent-study-finance · polish
polish_bibliography
Polish
used in 1 template · 1 published study · Examples: 3 example studies- Must write
- polish_bibliography.md
- Skills
- latex/bibtexsynthesis/context-builder
- Used in
- empirical · polishdid-staggered-panel · polishrd-platform-policy · polishiv-shift-share · polishevent-study-finance · polish
polish_equilibria
Polish
used in 1 template · 1 published study · Examples: 3 example studies- Must write
- polish_equilibria.md
- Skills
- modeling/game-theorymath/proof-strategies
- Used in
- empirical · polishdid-staggered-panel · polishrd-platform-policy · polishiv-shift-share · polishevent-study-finance · polish
revisor
Revisor specialist addresses feedback (if MAJOR_REVISION)
Design, analysis and writing
used in 1 template · 1 published study · Examples: 3 example studiespatch_revisor
Your job is to address a specific list of Finding objects by emitting a structured patch file: paper_draft.tex.edits.json.
Design, analysis and writing
used in 1 template · 1 published study · Examples: 3 example studiesreplication_packager
You are producing the replication materials that allow any researcher to reproduce every quantitative result in the paper from the raw data.
Design, analysis and writing
used in 1 template · 1 published study · Examples: 3 example studiesSpecialists for theory building
The E2ET theory lab brings specialists from 60 fields to a phenomenon. Each reasons from its field's canonical models and asks that field's diagnostic question. A few core roles run the process.
Researchers from these fields: edit your field's specialist. Its diagnostic question, canonical models and known biases should come from people who work in the field. Open a specialist and choose "Edit this specialist"; your name is credited on it.
Core roles
cartographer
Pre-round agent that maps the existing theory landscape around a refined phenomenon, identifying gaps and cross-disciplinary opportunities.
evaluator
Evaluates the synthesized theory against four criteria grounded in IS theory-building methodology: 1.
integrator
Post-convergence agent that transforms raw propositions, constructs, and mechanisms into a coherent, integrated theoretical framework with: 1.
persona_agent
A single parameterized agent that generates a transfer report for a given disciplinary persona.
phenomenologist
Pre-round agent that refines a raw phenomenon seed into a structured phenomenon analysis suitable for theory landscape mapping.
scout_director
Pre-round agent that selects core team and guest pool personas from the roster, based on the phenomenon and theory landscape analysis.
synthesizer
Integrates persona transfer reports into a coherent theoretical framework through a 3-batch, 7-micro-stage synthesis pipeline: Batch 1: Abduction → Articulation → Coherence Batch 2: Formalization → Severe Test Batch 3: Story Check → Type Check Each batch is a single Claude Code invocation.
Field specialists
Platform Economics
“Who are the sides? What are the cross-side externalities?”
not used in published research yet · Examples: 2 example studies Tier 0 · Information SystemsDigital Innovation
“What is being recombined? What generative capacity does the digital artifact have?”
not used in published research yet Tier 0 · Information SystemsIT Governance
“Who decides? What control mechanisms are in place?”
not used in published research yet Tier 0 · Information SystemsSociotechnical Systems
“How do social and technical elements co-constitute each other?”
not used in published research yet Tier 0 · Information SystemsTechnology Acceptance
“What drives adoption? Is it usefulness, ease of use, or social influence?”
not used in published research yet Tier 0 · Information SystemsKnowledge Management
“What knowledge is being created, transferred, or lost?”
not used in published research yet Tier 0 · Information SystemsDigital Transformation
“What organizational change is being driven by digital technology?”
not used in published research yet Tier 0 · Information SystemsIS Strategy
“How does IT create competitive advantage? Is the advantage sustainable?”
not used in published research yet Tier 0 · Information SystemsCrowdsourcing & Open Innovation
“Who is the crowd? What motivates participation? How is quality ensured?”
not used in published research yet Tier 0 · Information SystemsDigital Markets & E-Commerce
“How does digitization change search, matching, and trust?”
not used in published research yet Tier 1 · EconomicsGame Theory
“Who are the players? What are payoffs and information structure?”
not used in published research yet Tier 1 · EconomicsMechanism Design
“What is the designer's objective? What are agents' private information and incentive constraints?”
not used in published research yet Tier 1 · EconomicsBehavioral Economics
“Where do agents deviate from rationality? What heuristics and biases are at play?”
not used in published research yet Tier 1 · EconomicsIndustrial Organization
“What is the market structure? What are entry barriers and competitive dynamics?”
not used in published research yet Tier 1 · EconomicsLabor Economics
“Who works, for whom, under what contract, and with what frictions?”
not used in published research yet Tier 1 · EconomicsFinancial Economics
“What is being priced? What information is reflected? What frictions exist?”
not used in published research yet Tier 1 · EconomicsPublic Economics
“Is there a market failure? What is the public interest?”
not used in published research yet Tier 1 · EconomicsDevelopment Economics
“What structural barriers prevent improvement? What feedback loops maintain poverty?”
not used in published research yet Tier 1 · EconomicsAuction Theory
“What is the value model (private, common, affiliated)? What format is optimal?”
not used in published research yet Tier 1 · EconomicsContract Theory
“What can't be contracted upon? What are the information asymmetries?”
not used in published research yet Tier 1 · EconomicsNetwork Economics
“Are there network effects? Direct or indirect? How strong?”
not used in published research yet · Examples: 2 example studies Tier 1 · EconomicsEnvironmental Economics
“What is the externality? Who bears the cost? How can it be internalized?”
not used in published research yet Tier 2 · MathematicsCategory Theory
“What are the objects and morphisms? Is there a functor to a known structure?”
not used in published research yet Tier 2 · MathematicsGraph Theory
“What is the natural graph? What do structural properties tell us?”
not used in published research yet Tier 2 · MathematicsTopology
“What is the shape of the data? What topological features persist across scales?”
not used in published research yet Tier 2 · MathematicsDynamical Systems
“What are the state variables? Where are the fixed points and bifurcations?”
not used in published research yet Tier 2 · MathematicsProbability Theory
“What is the probability model? What are the tail risks?”
not used in published research yet Tier 2 · MathematicsOptimization Theory
“What is being optimized? What are the constraints? Is the problem convex?”
not used in published research yet Tier 2 · MathematicsInformation Theory
“What information is being transmitted? What is the channel? What is noise?”
not used in published research yet Tier 2 · MathematicsCombinatorics
“How many configurations are possible? What structural constraints limit them?”
not used in published research yet Tier 2 · MathematicsMeasure Theory
“What is the right measure? What sets have measure zero and why does it matter?”
not used in published research yet Tier 3 · PhysicsStatistical Mechanics
“What are the microstates? What macroscopic observables emerge from their statistics?”
not used in published research yet · Examples: 1 example study Tier 3 · PhysicsComplexity Science
“Is this emergent? What are the feedback loops? Are there tipping points?”
not used in published research yet · Examples: 1 example study Tier 3 · PhysicsQuantum Information
“Is there a superposition of states? Can entanglement-like correlations explain this?”
not used in published research yet Tier 3 · PhysicsThermodynamics
“What drives the system? Where does entropy increase? What work is extracted?”
not used in published research yet Tier 3 · PhysicsFluid Dynamics
“What flows? What are the viscosity and turbulence analogies?”
not used in published research yet Tier 3 · PhysicsNetwork Physics
“What is the network topology? How do dynamics on the network couple to dynamics of the network?”
not used in published research yet Tier 4 · Life SciencesEvolutionary Biology
“What is being selected? What varies? What is the fitness landscape?”
not used in published research yet Tier 4 · Life SciencesEcology
“What is the ecosystem? What are the niches, competitors, and interdependencies?”
not used in published research yet · Examples: 1 example study Tier 4 · Life SciencesEpidemiology
“What spreads? What is the transmission mechanism? What is R0?”
not used in published research yet Tier 4 · Life SciencesNeuroscience
“How do agents process information? What are the cognitive constraints?”
not used in published research yet Tier 5 · Computer ScienceDistributed Systems
“What consistency guarantees are needed? What failures must be tolerated?”
not used in published research yet · Examples: 1 example study Tier 5 · Computer ScienceCryptography
“What needs to be hidden? What needs to be proven? Who do we not trust?”
not used in published research yet Tier 5 · Computer ScienceComputational Complexity
“How hard is this problem? What computational resources are needed?”
not used in published research yet Tier 5 · Computer ScienceMachine Learning Theory
“What is being learned? What is the sample complexity? What is the generalization bound?”
not used in published research yet Tier 5 · Computer ScienceFormal Verification
“Can this property be formally specified and verified? What are the invariants?”
not used in published research yet Tier 6 · Social Sciences & HumanitiesSociology
“What social structures shape this? Who benefits, who is disadvantaged?”
not used in published research yet Tier 6 · Social Sciences & HumanitiesAnthropology
“What cultural meanings are embedded? How do participants understand their own practices?”
not used in published research yet Tier 6 · Social Sciences & HumanitiesPolitical Science
“Who governs? How is power distributed? Whose interests are served?”
not used in published research yet Tier 6 · Social Sciences & HumanitiesPsychology
“What motivates individual behavior? What cognitive processes are at play?”
not used in published research yet Tier 6 · Social Sciences & HumanitiesPhilosophy of Science
“Is this theory falsifiable? What is the paradigm? What would count as evidence?”
not used in published research yet Tier 6 · Social Sciences & HumanitiesLinguistics
“What is being communicated? How does meaning emerge from structure?”
not used in published research yet Tier 6 · Social Sciences & HumanitiesOrganizational Theory
“How does the organization structure affect behavior and outcomes?”
not used in published research yet Tier 6 · Social Sciences & HumanitiesScience & Technology Studies (STS)
“How are human and non-human actors assembled? What translations occur?”
not used in published research yet Tier 6 · Social Sciences & HumanitiesMedia Studies
“How does the medium shape the message? What logics of mediation are at play?”
not used in published research yet Tier 6 · Social Sciences & HumanitiesRhetoric
“How is this argument constructed? What persuasive strategies are at work?”
not used in published research yet Tier 6 · Social Sciences & HumanitiesDesign Theory
“What is the design space? What are the constraints and affordances?”
not used in published research yet Tier 7 · Law & RegulationLaw & Economics
“What are the legal rules? How do they shape incentives and outcomes?”
not used in published research yet Tier 7 · Law & RegulationRegulatory Theory
“Who regulates? What compliance mechanisms exist? Is code substituting for law?”
not used in published research yet · Examples: 1 example study Tier 7 · Law & RegulationIntellectual Property
“What knowledge is being protected? How does IP shape incentives to create?”
not used in published research yetCompose a panel
Example of the scout director's rulesThe scout director picks a core team of three or four field specialists that stays fixed across rounds. A guest pool of two or three rotates between rounds. No specialist can be in both.
Hold Ctrl or ⌘ to select several.
Example specialists for other theory templates
ExampleInvented specialists for the Gioia grounded-theory and disciplined-imagination templates.
first_order_coder
Codes transcripts in the informants' own terms.
second_order_theorist
Groups concepts into themes and dimensions and relates them into a model.
conjecture_generator
Generates conjectures from the perspective of one field.
selection_judge
Applies declared selection criteria and records its reasons.
What a specialist's description will state
Under draft 0.1 every specialist declares its role, inputs, outputs and the artifact it must produce, so a researcher can judge whether it fits before running it.
Describe a specialist