The Solution
What Narrative Does
To improve the baseline, Narrative transforms unstructured narrative time and matter data into structured, searchable, and comparable records.
The input: messy time entries, inconsistent phase codes, scattered matter information across multiple systems.
The output: a rich, queryable profile for every matter — complete with summaries, timelines, phase breakdowns, staffing analytics, and pricing-relevant data points.
This transformation enables firms to:
Find relevant matters instantly — Query by practice area, deal size, jurisdiction, outcome, or any combination of attributes
Benchmark new work accurately — See hours, fees, and timelines from comparable historical matters
Understand how work breaks down — Phase-by-phase analytics showing effort, cost, and team composition
Generate proposals faster — Complete with scopes, assumptions and phased analytics which can be adjusted to consider different AFA requirements or AI workflows
Manage the lifecycle — Track when work is likely going to overrun budgeted expectations drive learnings for future pricing and work allocation
Relevant KPIs
In addition to the above benefits, we're looking to drive these key KPIs:
Win/Loss Ratio → winning only one new deal nets a positive ROI. Increased transparency and flexibility can drive a 5-10% increase
Realisation Rate → having a better starting point and WIP tracking system should drive top-line revenue increases in the 2-4% range
Profit Margin → creating visibility on expected effort can help negotiation techniques that drive profitability
Time Efficiency → eliminating non-chargeable time per fee earner to increase effectiveness. 2-4hrs saved per senior fee earner translates to millions in opportunity cost
Client Satisfaction → Transparency, predictability and quality will make clients more satisfied with their advisors to drive more business
The Transformation Pipeline
Input: What We Ingest
Narrative starts with data the firm already has:
Data Lake / PMS
Time entries
Narratives, hours, fees, dates, timekeepers, matter IDs
Data Lake / PMS
Matter records
Names, clients, practice areas, open/close dates
Data Lake / PMS
Financial data
Rates, WIP, AR, write-offs, realization
DMS
Documents
Deal characteristics, outcomes, key terms (metadata, not content)
Other sources
Supplementary data
Deal size, jurisdiction, counterparties
We don't require complex integrations to start. A CSV export from your data lake is enough to begin transforming historical matters. We find this is sufficient to start proving initial value.
Process: The Transformation Journey
The transformation happens through multiple AI-powered passes. See our Data Transformation and Data Model pages for a detailed breakdown on the process we used to improve the historical data and create new comprehensive and structured matter profiles that are enriched with over 40 data points.
What Makes This Different
Before Narrative
Finding what matters were comparable and how they were relevant meant:
Emailing partners asking "have we done something like this?"
Searching billing system by keyword (hit or miss)
Scrolling through spreadsheet experience lists
Piecing together partial information from multiple sources
Hoping someone remembers the details
Result: Hours of work to produce an incomplete, inconsistent picture.
After Narrative
We move beyond simple search to an intelligent, agentic workflow:
RFP Ingestion & Clarification: Start by inputting the RFP. The system parses requirements and proactively asks clarifying questions to pinpoint key details.
High-Precision Retrieval: Narrative identifies the most relevant historical matters with 95%+ accuracy and consistency, filtering through tens of thousands of records instantly (<1 minute). It's equally important to identify how these matters are relevant which is core to the later stages.
Automated Proposal Generation: Once you are happy with the precedent set, the system generates a comprehensive RFP response. This includes:
Recommended pricing per phase
Optimal role mix and staffing plan
Detailed rationale explaining why the price is set at X, citing specific precedents and impact factors.
Human Input: Experts are then able to jump in and adjust variables like AI-tool usage, rates, caps or the substnative set of scopes and assumptions to tailor the proposal to the specific client in a way that they are expertly able to. This shifts the focus to high-impact and strategic inputs from partners and pricing professionals.
Result: A fully justified, data-backed pricing proposal that is competitive and profitable.
The Architecture of Memory: A Persistent Context Layer
Most tools are "stateless"—treating every request as a new event. Narrative is built on a Persistent Context Layer that "remembers" and evolves.
Client Profiles: "Client X always demands Fixed Fees for these matter types."
Partner Preferences: "Partner Y consistently discounts this phase by 10%."
Feedback Loops: When a user rejects a price, the system explicitly updates its model. It doesn't just execute; it learns.
Technical Foundation: We are investigating the Model Context Protocol (MCP) to standardize how our agents read and write to this memory layer, ensuring that insights gained in one workflow (e.g., pricing) immediately inform others (e.g., business development).
Security infrastructure: All of this is done in a way that complies with the highest security and privacy requirements. No data is used to train LLMs and all data is fully encrypted and siloed on multi-tenanted Azure environments deployed in the required regions. To learn more you can visit our trust centre at https://trust.inc/narrative.
The Flywheel Effect
Every matter processed makes the system smarter:
More data → Better pattern recognition
Better patterns → More accurate phase classification
Accurate classification → More reliable benchmarks
Reliable benchmarks → Better pricing decisions
Better pricing → More profitable matters
More matters → More data
Over time, Narrative learns your firm's specific patterns, terminology, and work styles. The transformation becomes more accurate and the insights more valuable.
Last updated