Archive to asset case study

From Buried Interviews To A Living Research Engine

I used AI to help turn a messy research archive into a database, quote bank, claims register, and content/proof system. The lesson is simple: your existing data may already contain the next move.

Get Your Unlocked Diagnostic Book a Right-It Audit
research archive AI workflow evidence system next-move map

Scroll to explore

The buried archive

The value was already there.
It was just locked.

The archive came from a major African tech ecosystem research project. The hard part had already happened: the conversations, notes, context, and raw material existed. The problem was that the material was scattered, sensitive in places, and too hard to reuse quickly.

80+

Stakeholder interviews across the broader research effort.

45

Clean external transcript files in the verified corpus.

289,773

Words of transcript material processed into evidence infrastructure.

43

Validated people rows after first-pass duplicate consolidation.

42

Curated quote candidates, labeled before any public use.

20

Structured claims separated by evidence, interpretation, and confidence.

Public-safe note: this page uses aggregate stats, paraphrases, and broad themes. It does not publish raw internal employee quotes, sensitive shutdown claims, or unverified named quotes.

The mistake

A summary is not a strategy.

The default AI move is to upload a folder and ask for a summary. That can be useful, but it usually gives you one polished blob of text. I wanted something that could keep producing decisions.

AI summary

Useful for getting oriented.
Weak for repeated use.

  • No durable source paths
  • No public/private labels
  • No confidence register
  • No next-move map
Dead end
summary.txt
one-shot output · 10:41 AM
Decisions
don’t come
from blobs.
They come
from systems.

AI workflow

Useful because the archive becomes an operating system for insight.

  • Inventory before synthesis
  • Database before conclusions
  • Quote bank before publishing
  • Claims register before public claims
Inventory 365 items
Source paths

All mapped

Claims Register
  • C01 People want clarity Strong
  • C02 Trust drives adoption Strong
  • C03 Friction kills action Medium
  • Add claim
Quote Bank VERIFIED

People don’t just want information, they want clarity they can act on.

Participant 07
Next move:
Interview 07
→ Validate C02
Process pipeline

The pipeline changed the work.

Each stage made the next stage easier. That is the actual AI unlock: not magic output, but better structure around messy inputs.

ENTER EXIT
1

Raw Archive

Went in
Interviews, reports, notes, duplicate files, and unfinished ideas.
Changed
The mess was treated as a corpus, not a folder of guilt.
Came out
A clear source boundary and privacy-first working frame.
2

Corpus Inventory

Went in
External transcripts and file metadata.
Changed
Files were counted, checked, and duplicate decisions documented.
Came out
45 clean transcripts and 289,773 verified words.
3

Research Database

Went in
Transcript text, source paths, people metadata, and theme tags.
Changed
The archive became structured enough to query and export.
Came out
SQLite database, CSV exports, and validated people rows.
4

Theme + Persona Synthesis

Went in
Interview patterns by founder, investor, operator, ecosystem builder, and institution.
Changed
Patterns were mapped by pain point instead of flattened into generic themes.
Came out
Persona pain-point map and cross-cutting insight layer.
5

Quote Bank

Went in
Promising evidence candidates from the external corpus.
Changed
Evidence was labeled by public safety, confidence, and use case.
Came out
42 curated quote candidates ready for review, not raw publication.
6

Claims Register

Went in
The strongest recurring interpretations and hypotheses.
Changed
Claims were separated from evidence and given a validation status.
Came out
20 structured claims with confidence and public-use guidance.
7

Content / Proof Map

Went in
Insights, claims, anonymized patterns, and offer questions.
Changed
Research became public-safe proof for Unlocked, not just an internal memo.
Came out
Content lanes, teaching demos, case-study proof, and CTA paths.
8

Offers + Next Moves

Went in
The question every archive should answer: what do we do with this?
Changed
The archive was translated into diagnostics, audits, snapshots, and B2B opportunities.
Came out
A clearer next-move map instead of another document to forget.
What came out of it

Section 7 / 8

The output was not one memo.
It was a reusable system.

01

Research database

A structured evidence layer that can be queried, exported, and revisited instead of rediscovered.

02

Persona pain-point map

A clearer view of how founders, funders, operators, and institutions experience the ecosystem differently.

03

Quote bank

Evidence candidates labeled for sensitivity, confidence, and future use before anything becomes public.

04

Claims register

A guardrail that keeps evidence, interpretation, and hypotheses from getting blended into overconfident AI output.

05

Content lanes

Public-safe angles for Unlocked, founder support, AI leverage, and Right-It thinking.

06

Offer opportunities

Clear paths from research infrastructure into diagnostics, audits, snapshots, teaching demos, and B2B work.

Interactive Insight Explorer

Explore what came
out of the archive.

These are public-safe examples of what the system produced: insights, anonymized quote patterns, claims, content angles, and offer opportunities. The sensitive material stays private. The structure is the point.

Archive knowledge

24,317 insights

  • 9,812
  • 6,243
  • 5,672
  • 2,590
Persona
Creator
Theme
All
Output type
All
Customer use case
All
Content angle card / Creator / AI leverage

AI made building easier.
It did not make deciding easier.

A research pattern became a public content angle: the danger is not slow execution anymore. The danger is beautiful execution pointed at a weak assumption.

Content engine Wrong Thing Well content lane
Content angle card / Creator / Teaching asset

Audience notes can
become a course map.

Voice notes, comments, DMs, and old drafts can be structured into content pillars, objections, lesson modules, examples, and a better first offer.

Notes and comments Diagnostic, course outline, content lane

Privacy rule: these cards are paraphrased or aggregated. No raw internal employee quotes, sensitive claims, or named external quotes are exposed here.

Learn more
What this could look like for you

You may already be sitting on your next move.

Your version might be customer calls, founder applications, old consulting work, internal docs, or audience notes. The question is not whether AI can summarize them. The question is what system would turn them into decisions.

Founder

Customer calls

Turn calls into objections, buying triggers, pain patterns, positioning, and test ideas.

Operator

Internal docs

Turn SOPs and handoffs into workflow maps, decision rules, and automation opportunities.

Consultant

Old client work

Turn reports and workshop notes into reusable IP, lead magnets, diagnostics, and offers.

Researcher

Interviews

Turn transcripts into a quote bank, claims register, synthesis, and a public-safe report.

VC / studio

Founder data

Turn applications and updates into support gaps, founder journeys, and operating-system recommendations.

Creator

Notes and comments

Turn drafts, DMs, and audience feedback into content pillars, course modules, examples, and offer ideas.

Next move

You may already be sitting on buried signal.

If this made you think of your own messy folder, start with the research archive question.
If you already know what you want to unpack, book the audit.

Get Your Unlocked Diagnostic Book a Right-It Audit
Home / Unlocked / DIF Built from signal, not theory.
Before / After

From buried material to a reusable system

Before:
buried material

folder

Interview transcripts, old reports, notes, and half-remembered files.

problem

Good signal existed, but it was hard to retrieve, compare, or use safely.

risk

A one-time summary would create a neat document, then die in the chat.

need to
circle back

After:
reusable system

database

Structured corpus inventory with people, files, themes, and exports.

evidence

Quote bank, claims register, privacy labels, and confidence notes.

moves

Content lanes, offer opportunities, teaching demos, and decision paths.

Research System
  • Dashboard
  • Corpus
  • Evidence
  • Themes
  • Exports
  • Settings

Corpus Inventory

Total items 1,482 · Sources 73

Ready to Export

14 items