All-in-one SaaS platform for measuring and managing social sustainability

Comprehensive platform powered by AI agents that helps financial institutions track, measure, and rebalance their asset portfolios towards better ESG scores

“Thank you for your first class input throughout the project and the speed of execution that made this project successful.”

Drishdey Caullychurn

CEO & Founder, TEXpert AI

Overview

Texpert wanted to give banks and fund managers a clear, numeric view of how the companies and funds they hold perform on social and wider ESG factors. The vision: pull facts from official reports, turn long narrative text into numbers you can trust, show scores at both portfolio and single asset level, let users try “what if” changes, and make every number traceable back to a source line.

They asked us to help turn a set of research notebooks into a real, secure, multi‑tenant product that could scale and win enterprise trust.

Challenge

ESG info is mostly long-form text: annual reports, sustainability reports, press releases, regulatory filings. Existing tools often output vague bands like “good / average / poor,” with little clarity on data sources or how a rating was decided.

Scores can’t be drilled into, methods are hidden, and users can’t test how swapping one holding for another changes overall impact. TEXpert’s data scientists had prototype Python notebooks that scraped and parsed documents for signals, but nothing production ready: no unified pipeline, no consistent scoring model, no user interface, no role-based access, and no way to handle large portfolio uploads.

They needed a clear, transparent system that turned messy narrative data into stable numeric scores (e.g. 7.2 / 10) with full traceability and fast scenario testing.

Outcome

We partnered with TEXpert’s data team to design and build the end-to-end platform. We converted exploratory notebooks into robust ingestion and parsing services that pull in annual and sustainability reports (PDF and HTML), apply text extraction and classification, and map findings to a defined ESG taxonomy.

Each metric gets a numeric score plus links back to exact source snippets for audit. We introduced a weighting and aggregation engine to produce per-dimension and composite 0–10 scores at asset level, then rolled these up to portfolio level. We added bulk portfolio import (CSV / API), on-the-fly “what if” rebalancing to test swaps and reallocations, and instant recompute of scores and projected improvements.

Multi-tenancy, user roles, secure document storage, and an audit log were built in from the start. An AI chat interface lets users ask, “Why did Company X get a low labour rights score?” and get an answer with cited source passages.

The result: a clear, transparent ESG lens with actionable numbers instead of vague labels. The strength of the working prototype and enterprise features helped Texpert secure a paid pilot with HSBC UK’s investments and wealth management group, positioning the product to be offered to fund managers who want to understand and improve the social impact of their portfolios.