knowledge-synthesis

Knowledge Synthesis

The last mile of enterprise search. Takes raw results from multiple sources and produces a coherent, trustworthy answer.

The Goal

Transform this:

~~chat result: "Sarah said in #eng: 'let's go with REST, GraphQL is overkill for our use case'"
~~email result: "Subject: API Decision — Sarah's email confirming REST approach with rationale"
~~cloud storage result: "API Design Doc v3 — updated section 2 to reflect REST decision"
~~project tracker result: "Issue: Finalize API approach — marked complete by Sarah"

Into this:

The team decided to go with REST over GraphQL for the API redesign. Sarah made the
call, noting that GraphQL was overkill for the current use case. This was discussed
in #engineering on Tuesday, confirmed via email Wednesday, and the design doc has
been updated to reflect the decision. The related GitHub issue is marked complete.

Sources:
- ~~chat: #engineering thread (Jan 14)
- ~~email: "API Decision" from Sarah (Jan 15)
- ~~cloud storage: "API Design Doc v3" (updated Jan 15)
- ~~project tracker: "Finalize API approach" (completed Jan 15)

FashionUnited Example: Brand Lookup

Transform this:

~~CRM result: "Gucci - Account Status: Active Advertiser, Contract: Annual display package"
~~data warehouse/editorial: "47 articles mentioning Gucci in 2024, including Milan Fashion Week coverage"
~~data warehouse/marketplace: "1,247 Gucci products in marketplace catalog"
~~data warehouse/jobs: "12 Gucci job postings in last 6 months"
~~email result: "Latest correspondence: Invoice discussion with marketing team, Dec 2024"

Into this:

Gucci is an active advertiser with FashionUnited on an annual display package.

**Relationship summary:**
- Active advertising customer with annual contract
- 47 articles published mentioning Gucci in 2024 (including Milan Fashion Week coverage)
- 1,247 products in the marketplace catalog
- 12 job postings in the last 6 months

**Recent activity:**
- Latest correspondence in December 2024 regarding invoicing

Sources:
- ~~CRM: Account record (Vtiger)
- ~~data warehouse: Editorial dataset (47 articles), Marketplace dataset (1,247 products), Jobs dataset (12 postings)
- ~~email: December 2024 correspondence

Deduplication

Cross-Source Deduplication

The same information often appears in multiple places. Identify and merge duplicates:

Signals that results are about the same thing:

• Same or very similar text content
• Same author/sender
• Timestamps within a short window (same day or adjacent days)
• References to the same entity (project name, document, decision, brand name)
• One source references another ("as discussed in ~~chat", "per the email", "see the doc")

How to merge:

• Combine into a single narrative item
• Cite all sources where it appeared
• Use the most complete version as the primary text
• Add unique details from each source

Deduplication Priority

When the same information exists in multiple sources, prefer:

1. The most complete version (fullest context)
2. The most authoritative source (official doc > chat)
3. The most recent version (latest update wins for evolving info)

What NOT to Deduplicate

Keep as separate items when:

• The same topic is discussed but with different conclusions
• Different people express different viewpoints
• The information evolved meaningfully between sources (v1 vs v2 of a decision)
• Different time periods are represented

Citation and Source Attribution

Every claim in the synthesized answer must be attributable to a source.

Attribution Format

Inline for direct references:

Sarah confirmed the REST approach in her email on Wednesday.
The design doc was updated to reflect this (~~cloud storage: "API Design Doc v3").

Source list at the end for completeness:

Sources:
- ~~chat: #engineering discussion (Jan 14) — initial decision thread
- ~~email: "API Decision" from Sarah Chen (Jan 15) — formal confirmation
- ~~cloud storage: "API Design Doc v3" last modified Jan 15 — updated specification

Attribution Rules

• Always name the source type (Google Chat, Gmail, Google Drive, Vtiger CRM, BigQuery, etc.)
• Include the specific location (channel, folder, thread, dataset, repository)
• Include the date or relative time
• Include the author when relevant
• Include document/thread titles when available
• For Google Chat, note the space name
• For Gmail, note the subject line and sender
• For Google Drive, note the document title
• For Vtiger CRM, note the record type (Account, Contact, Opportunity)
• For BigQuery, note the dataset and query type
• For GitHub, note the repository and file/path

Confidence Levels

Not all results are equally trustworthy. Assess confidence based on:

Freshness

Recency	Confidence impact
Today / yesterday	High confidence for current state
This week	Good confidence
This month	Moderate — things may have changed
Older than a month	Lower confidence — flag as potentially outdated

For status queries, heavily weight freshness. For policy/factual queries, freshness matters less.

Authority

Source type	Authority level
Official wiki / knowledge base	Highest — curated, maintained
Shared documents (final versions)	High — intentionally published
Email announcements	High — formal communication
CRM records	High — official business data
BigQuery datasets	High — structured, validated data
Meeting notes	Moderate-high — may be incomplete
Chat messages (thread conclusions)	Moderate — informal but real-time
Chat messages (mid-thread)	Lower — may not reflect final position
Draft documents	Low — not finalized
GitHub issue comments	Contextual — depends on commenter

FashionUnited-Specific Authority

Query Type	Authority Hierarchy
Brand/Account info	CRM > Email > Editorial > Chat
Job market data	BigQuery jobs > CRM > Integrations > Email
Editorial content	BigQuery editorial > Drive > Chat
Product/Catalog	BigQuery marketplace > product-database > CRM
Technical/API	GitHub (api repo) > Google Chat #engineering > Email
Company policy	GitHub (about repo) > Drive > Email > Chat

Expressing Confidence

When confidence is high (multiple fresh, authoritative sources agree):

Gucci is an active advertiser with an annual display contract. [direct statement]

When confidence is moderate (single source or somewhat dated):

Based on CRM records, Gucci's last contract renewal was in Q3 2024.
The current status may have changed — recommend checking with the sales team.

When confidence is low (old data, informal source, or conflicting signals):

I found references to Gucci discussions from six months ago in Google Chat,
but couldn't find current account data in the CRM. The information
may be outdated. You might want to check with the sales team for current status.

Conflicting Information

When sources disagree:

I found conflicting information about the API approach:
- The ~~chat discussion on Jan 10 suggested GraphQL
- But Sarah's email on Jan 15 confirmed REST
- The design doc (updated Jan 15) reflects REST

The most recent sources indicate REST was the final decision,
but the earlier ~~chat discussion explored GraphQL first.

Always surface conflicts rather than silently picking one version.

Summarization Strategies

For Small Result Sets (1-5 results)

Present each result with context. No summarization needed — give the user everything:

[Direct answer synthesized from results]

[Detail from source 1]
[Detail from source 2]

Sources: [full attribution]

For Medium Result Sets (5-15 results)

Group by theme and summarize each group:

[Overall answer]

Theme 1: [summary of related results]
Theme 2: [summary of related results]

Key sources: [top 3-5 most relevant sources]
Full results: [count] items found across [sources]

For Large Result Sets (15+ results)

Provide a high-level synthesis with the option to drill down:

[Overall answer based on most relevant results]

Summary:
- [Key finding 1] (supported by N sources)
- [Key finding 2] (supported by N sources)
- [Key finding 3] (supported by N sources)

Top sources:
- [Most authoritative/relevant source]
- [Second most relevant]
- [Third most relevant]

Found [total count] results across [source list].
Want me to dig deeper into any specific aspect?

FashionUnited: Summarizing Brand Results

For brand lookup queries with multiple data sources:

## [Brand Name] Overview

**Relationship Status:** [Active/Inactive] [Customer Type]
**Contract:** [Contract details from CRM]

### Editorial Coverage
- [X] articles in [year]
- Key topics: [list main themes]
- Recent: [most recent article]

### Marketplace Presence
- [X] products in catalog
- Categories: [product categories]

### Job Activity
- [X] job postings in last [period]
- Roles: [common job types]

### Recent Communications
- [Summary of email/chat activity]

Sources:
- ~~CRM: [Account record]
- ~~data warehouse: Editorial ([X] articles), Marketplace ([X] products), Jobs ([X] postings)
- ~~email: [Recent correspondence]
- ~~chat: [Relevant discussions]

Summarization Rules

• Lead with the answer, not the search process
• Do not list raw results — synthesize them into narrative
• Group related items from different sources together
• Preserve important nuance and caveats
• Include enough detail that the user can decide whether to dig deeper
• Always offer to provide more detail if the result set was large

Synthesis Workflow

[Raw results from all sources]
          ↓
[1. Deduplicate — merge same info from different sources]
          ↓
[2. Cluster — group related results by theme/topic]
          ↓
[3. Rank — order clusters and items by relevance to query]
          ↓
[4. Assess confidence — freshness × authority × agreement]
          ↓
[5. Synthesize — produce narrative answer with attribution]
          ↓
[6. Format — choose appropriate detail level for result count]
          ↓
[Coherent answer with sources]

Anti-Patterns

Do not:

• List results source by source ("From ~~chat: ... From ~~email: ... From ~~cloud storage: ...")
• Include irrelevant results just because they matched a keyword
• Bury the answer under methodology explanation
• Present conflicting info without flagging the conflict
• Omit source attribution
• Present uncertain information with the same confidence as well-supported facts
• Summarize so aggressively that useful detail is lost

Do:

• Lead with the answer
• Group by topic, not by source
• Flag confidence levels when appropriate
• Surface conflicts explicitly
• Attribute all claims to sources
• Offer to go deeper when result sets are large