The proof strategy at aswritten has evolved from external validation toward [recursive self-documentation](https://aswritten.ai/narrative#Narrative_PipelineSelfDocumentation_20260704), where the product now serves as its own primary case study. This trajectory addresses the persistent [value demonstration gap](https://aswritten.ai/narrative#Narrative_ValueDemonstrationGap), a hurdle framed as a [prerequisite for capital and enterprise growth](https://aswritten.ai/narrative#Position_ValueDemoPrerequisite). Early efforts focused on [quantifiable annotation metrics](https://aswritten.ai/narrative#Position_AnnotationResults80Claims) to verify claim support, while the current approach emphasizes [product-as-proof](https://aswritten.ai/narrative#Narrative_ProductAsProof) through high-fidelity, automated provenance. The system’s rhetorical architecture reinforces this through [antithetical framing](https://aswritten.ai/narrative#StyleObs_Antithesis) to define process semantics and [timestamped citation conventions](https://aswritten.ai/narrative#StyleObs_TimestampedQuotes) that mirror its provenance-first design. While internal development prioritizes [provenance over precision](https://aswritten.ai/narrative#Position_ContextArch_5), external feedback highlights [product stickiness](https://aswritten.ai/narrative#Obs_StickyProductQuote) and the potential for organic scaling, alongside operational benchmarks from [analogous collective models](https://aswritten.ai/narrative#Observation_ElephantCollective).
Witness phrases
verbatim The product documents itself: cutover decision saved through new production pipeline, citation offsets slice-exact on first try
verbatim Visible proof that aswritten changes how AI works is a prerequisite for raising bigger money and closing enterprise sales
verbatim I think you've got a sticky product. If you can get it into the hands of 20, 50 people... The word of mouth of that will just... that's really catapult.
verbatim The real risk is misattribution, not hallucination; mitigation is provenance in output, not query precision
verbatim Internal focus on technical provenance vs. external focus on user-driven 'stickiness' and word-of-mouth scaling.
verbatim The tension between 'evaluating' the system via annotation metrics and 'running' it as a self-documenting pipeline.
In this area
NarrativeNarrative_PipelineSelfDocumentation_20260704 The product documents itself: cutover decision saved through new production pipeline, citation offsets slice-exact on first try
NarrativeNarrative_ProductAsProof The YC application itself was generated by AI using collective memory—'one shot and usable'—demonstrating the product's capability to produce human-quality, contextualized output.
NarrativeNarrative_ValueDemonstrationGap a demo-call prospect's core feedback: the product needs to show how aswritten influences AI behavior. Agents should cite which memories shaped their recommendations. This is the 'value demonstration' problem (task-145). Without visible proo
ObservationObs_StickyProductQuote Tony's assessment of the product's stickiness and the word-of-mouth potential once it reaches 20-50 users.
ObservationObservation_ElephantCollective Bootstrapped from [amount]to [amount]/event in 6 months. ~20 artists-in-residence, 1,000–2,000 person events, 100–200 volunteer crew recruited via mass SMS.
StyleObservationStyleObs_Antithesis They solve memory for individuals. We solve memory for organizations. They create lock-in. We create freedom.
StyleObservationStyleObs_TimestampedQuotes Inline timestamp citations in [MM:SS] format appended to speaker-attributed quotes; functions as a lightweight call-recording reference convention.
3 further nodes are not published in the public projection. Each node links from the entity index.