TfP Docs

Technology Commercialization

Imagine a Campaign attempting to map and continuously evaluate federally funded technologies available across universities and research institutions.

The goal is not to generate one static database.

The goal is to maintain an evolving body of structured, reviewed analysis that people and agents can continuously improve as technologies, markets, companies, and regulations change.

flowchart TD
    S[University / grant / patent / lab sources] --> D[Discover technologies]
    D --> T[Standing Problem for Technology X]
    T --> A1[Technical analysis]
    T --> A2[Market analysis]
    T --> A3[Technoeconomic analysis]
    T --> A4[Buyer analysis]
    A1 --> DS[Commercialization dossier]
    A2 --> DS
    A3 --> DS
    A4 --> DS
    DS --> R[Review]
    R --> V[Trusted dossier]
    V --> U[More Attempts can improve it]

1. The Campaign defines the standard, not TfP

The Campaign owner publishes one or more Skills.

For example:

@campaign/commercialization-dossier
@campaign/market-analysis
@campaign/technoeconomic-analysis
@campaign/buyer-analysis

The dossier Skill may contain:

SKILL.md
schemas/commercialization-dossier.schema.json
examples/
source-quality guidance/
market-analysis methods/
scripts/

The schema might expect sections such as:

technology identity

technical description

maturity

known applications

commercialization constraints

market segments

market-size estimates

technoeconomic assumptions

competitive landscape

candidate licensees / acquirers

evidence

open questions

last-assessed timestamps

TfP does not need a TechnologyDossier core object.

The dossier is an Artifact following a community-defined standard.

2. Discover technologies

A standing DISCOVERY Problem asks:

Find federally funded technologies that are not yet represented in this Campaign.

Agents may search:

An Attempt submits candidate technologies with provenance and deduplication evidence.

Accepted discoveries may generate Problems such as:

Maintain a commercialization dossier for Technology X.

3. The dossier is a standing Problem

This is important.

The Problem is not:

Create version 1 of this dossier.

It is:

Maintain the best supported commercialization dossier for Technology X.

As long as the instructions and verification standard stay the same, the Problem can remain on one Revision while arbitrary people and agents submit improvements.

Problem:
Maintain commercialization dossier for Technology X

same Problem Revision
│
├── Attempt A → dossier 1
├── Attempt B → dossier 2
├── Attempt C → dossier 3
└── Attempt D → dossier 4

An Attempt can begin for any reason.

No Campaign-owner trigger is required.

Someone familiar with the technology may simply notice:

The current dossier completely misses an aerospace use case.

They start an Attempt and submit a better dossier.

Another agent might notice that a new company looks like an obvious licensee.

Another might discover that a manufacturing process changed the economics.

4. Supporting analyses can be separate Problems

A good dossier may depend on substantial specialized work.

For Technology X:

Maintain commercialization dossier
│
├── Technical feasibility analysis
├── Technoeconomic analysis
├── Market sizing
├── Competitor landscape
├── Buyer / licensee search
└── Regulatory analysis

Those can be independent Problems with their own:

A synthesis Attempt can build on their trusted outputs.

Future generic lineage should make that explicit:

dossier v7
├── based on dossier v6
├── incorporates market analysis M4
├── incorporates TEA T2
├── incorporates buyer map B8
└── cites technical assessment F3

The dossier remains an immutable Artifact.

5. Review the dossier

Structural checks can run first:

sandbox
→ valid schema
→ required sections present
→ citations have valid structure
→ numeric units parse

Then structured review asks substantive questions.

An agent panel might review:

Does every major factual claim have evidence?

Are market-size assumptions internally consistent?

Does the technoeconomic analysis expose its assumptions?

Are candidate buyers connected to concrete strategic needs?

Are important competing technologies omitted?

Does the dossier distinguish facts from estimates?

Are stale claims clearly dated?

Material disagreement may escalate to a human reviewer with domain expertise.

The review should preserve rich outputs rather than a single score.

For example:

{
  "technical_quality": "strong",
  "market_analysis": "needs_revision",
  "buyer_analysis": "strong",
  "major_omissions": [
    "Aerospace thermal-management applications"
  ],
  "recommendation": "revise"
}

with rationale and citations.

6. A standing discovery Problem looks for reasons to revisit old dossiers

A second standing DISCOVERY Problem can ask:

Find material developments that make an existing commercialization dossier worth reevaluating.

This is different from automatically declaring dossiers stale.

Agents search for changes such as:

A candidate might say:

Technology X

Reason for reevaluation:
Company Y recently entered a market whose main technical bottleneck
closely matches Technology X.

Likely affected dossier sections:
- buyer landscape
- addressable market
- commercialization pathway

After triage, this can create a ProblemProposal or simply surface:

Technology X deserves another Attempt.

Importantly, anybody may start that Attempt even before the discovery result exists.

The discovery process improves routing; it does not gate participation.

7. Improvements do not require Problem Revisions

Suppose dossier v4 estimates:

TAM: $80M

Six months later, someone believes a new application changes the plausible market to:

TAM: $900M

That does not mean the Problem's instructions changed.

They submit another Attempt against the same Revision.

dossier v4
    ↓
new Attempt
    ↓
dossier v5

A new Problem Revision is appropriate only if the Campaign changes the task itself.

For example:

Old specification:
market sizing + buyers + technical maturity

New specification:
also require a formal technoeconomic sensitivity analysis

That is a specification change.

8. Competing analyses are useful

The platform should not assume there is always one obviously correct current dossier.

Two groups may disagree about market size or commercialization strategy.

Submission A
TAM estimate: $200M

Submission B
TAM estimate: $1.1B

Both can remain visible with:

A later synthesis Attempt can explain the disagreement.

This is often more scientifically useful than silently overwriting A with B.

9. Campaign-wide analysis

After agents have worked on thousands of technologies, the Campaign owner may want to ask:

Which technologies appear to have the highest commercialization potential?

Which institutions have the most promising undercommercialized portfolios?

Which analyses have changed most over the last six months?

What factors most often cause a technology's estimated TAM to increase?

Which companies repeatedly appear as plausible licensees?

What fraction of technologies have had independent reviews?

Because dossiers use a Campaign-defined structured schema, verified outputs can be projected into an analytical index.

canonical evidence:
R2 Artifacts

canonical provenance:
TfP object graph

derived analysis:
columnar projection of verified structured outputs

A Campaign export might produce:

commercialization_dossiers.parquet
technologies.parquet
buyer_candidates.parquet
verification_summary.parquet
artifact_manifest.ndjson

with references back to the canonical Artifacts and VerificationRuns.

This should be cheap relative to repeatedly invoking models or experts because parsing/indexing occurs once when trusted outputs are produced.

10. Swarm coordination

A large swarm does not need to edit one shared JSON file.

Instead:

flowchart TD
    D6[Dossier v6] --> M[Market update Attempt]
    D6 --> B[Buyer-search Attempt]
    D6 --> T[TEA Attempt]
    D6 --> C[Competitor scan]

    M --> S[Synthesis Attempt]
    B --> S
    T --> S
    C --> S

    S --> D7[Dossier v7]

Independent work stays independently attributable and reviewable.

Synthesis combines trusted outputs.

This makes both provenance and contributor credit much easier than collaborative mutation of one blob.

11. What this Campaign requires from TfP

Existing primitives

Stage-8 candidate features

Infrastructure

Explicitly not required