Chat context
Analyzing 0 grants across the portfolio
SR
Portfolio Insights
Ask qualitative questions about your portfolio, or quantitative ones — the assistant routes between vector search over grant documents and the structured grant CSV.
Social Return AdvisoryPortfolio Insights Demo
Behind the chatbot
Underlying Data
The chatbot draws every answer from the source documents and structured records below. Nothing is invented — when an answer cites a grant, it is referencing what is shown here.
All 0 grants × 4 documents = 0 source files ingested into the chatbot
Indexed in Pinecone across three namespaces · last refreshed when you re-ran npm run analyze
Documents per grant
Each of the 0 grants has these 4 documents — 0 files total
Application
0 files
Initial grant proposal — what the grantee planned to do, who they would serve, and what success looked like.
{grant_id}_application.md
Midpoint Transcript
0 files
Mid-grant check-in conversation — how the program is tracking against its plan halfway through the period.
{grant_id}_midpoint_transcript.md
Impact Survey
0 files
Participant-level survey results — direct beneficiary outcomes, satisfaction, and reported income changes.
{grant_id}_impact_survey.md
Results Transcript
0 files
Final results conversation — what actually happened, what the grantee learned, what they would do differently.
{grant_id}_results_transcript.md
By Country
By Strategy
By Status
By Performance Tier
Tier reflects outcome strength relative to plan, drawn from interim assessments for active grants and final results for completed grants. Tiers for active grants may shift as evidence matures.
How it gets to the chatbot
The offline pipeline that turns these files into queryable answers
Step 1
Source documents
Demo
0 markdown files
4 docs × 0 grants
Client
API connection to your document store
Google Drive, SharePoint, or grant repository — refreshed every 24 hours
Step 2
Structured CSV
Demo
0 grant rows
country, strategy, ROI, tier, status
Client
API connection to your GMS
Fluxx, SmartSimple, Salesforce, or proprietary — refreshed every 24 hours
Step 3
Pre-generated case files
Demo
0 case files
tagged challenges, outcomes, evidence
Client
Same pipeline, your taxonomy
case file schema configured to your M&E framework
Step 4
Cross-grant insights
Demo
~25–30 insights
patterns appearing in ≥3 grants
Client
Auto-regenerated on data refresh
new insights surface as new grants close
Data shown is fictional — the "Social Return Foundation" is a demo portfolio of 0 grants across USA, Uganda, and Ecuador, designed to illustrate the architecture without exposing any real foundation's grantmaking.