Power Automate & Copilot Workflows

Job ID: 40180407

Budget: £250 – £750 GBP

Build a document review assistant that processes up to 50 documents per day and produces a quality score plus short feedback for each document based on a scoring rubric. Output should be a simple report (one row per document). Rubric should be easy to update.

Project brief: Document Review Assistant (feature build)

Summary
Build a lightweight “document review assistant” that takes a batch of documents each day (up to ~50), reviews the quality using a defined scoring rubric, and produces a score and short feedback for each document.

This is intended for operational use: run it daily, get a simple output report.

---

What it should do

1) Batch input

Accept a set of documents in one go (up to ~50 per run).
Support a daily run and a manual run option.

2) Quality review + scoring

Score each document using a rubric we provide (with categories and weights).
Keep scoring focused on document quality (structure, clarity, completeness, consistency, presentation).
Provide:

an overall score (e.g., out of 100)
category scores (based on the rubric)
3–6 bullets of strengths and improvement suggestions
If a document can’t be processed properly, return a clear status and reason.

3) Output/reporting

Produce a simple report with one row per document containing:

document name / identifier
overall score
category breakdown
short rationale / key notes
suggested improvements
Output should be easy to share and compare across the batch.

4) Consistency + configurability

The rubric, weights, and scoring rules must be easy for us to edit later.
The assistant should produce consistent results run-to-run (as much as reasonably possible for GenAI).

Deliverables

A working document review assistant that runs end-to-end on a batch of documents.
A scoring rubric implementation (we provide the rubric; you implement it).
A clear output report format.
Short setup/hand-over notes: how to run it, how to update the rubric, and how to troubleshoot.

---

Acceptance criteria

Processes up to ~50 documents per run.
Produces a score + category breakdown + short feedback for every document (or a clear failure reason).
Output report is consistent, readable, and usable for daily workflows.
Scoring remains strictly about document quality , not personal attributes.

---

What to include in your proposal

Examples of similar “review / scoring / summarisation” assistants you’ve built.
How you’d ensure consistent scoring and structured outputs.
Estimated timeline and cost.
Any assumptions or dependencies you foresee.

Optional add-ons (nice to have)
A simple dashboard summary (e.g., average score, top issues across the batch).
A “change log” of rubric updates and versioning of scoring rules.
Basic audit trail for what was processed each day.