LexisNexis Embedded Innovation · Customer Operations & Finance
What we're building, and how we partner to deliver it
An introduction to the Embedded Innovation portfolio across Finance and Customer Operations: what's in flight, where each initiative sits today, and where our work connects with the rest of Global Business Systems.
What Embedded Innovation is. Embedded Innovation exists to put AI to work inside LexisNexis's own operations: Finance and Customer Operations. We're a small team of product, engineering, and data science, and we take an idea from a real workflow to a validated, working solution. Some of what we build integrates into the systems teams already own; some is a new platform that leverages AI and serves its own end users, like Alfred+. We build in partnership with the teams whose systems and users it touches.
Not sure when to bring us in, or whether something you're working on could use AI? Ping me, and I'm happy to help brainstorm. Email Danielle Brown →
How the collaboration works
We begin
Discovery
We start inside real Finance and Customer Operations workflows, finding where people lose hours and where errors slip in, before a line of code is written.
→
We build
From data to a working solution
We do the data-science work and turn the concept into a working, SME-validated solution with an engineering-ready specification.
→
Then it scales
Integrate, or a platform of its own
Validated work goes to production: either drawing on and integrating with the systems teams already own (Oracle Fusion, NICE / Agent Web, product APIs), or standing up as a platform of its own, like Alfred+. Always in partnership with the teams it touches.
One initiative, end to end: where our work meets our partner teams
Legal Research Copilot is the clearest picture of the handoff. It's a customer-service assist capability: when a customer asks a legal-research question in chat, it suggests the right source and search for the service rep to review and send. Nothing auto-sends. Here's who owns what, left to right.
Embedded Innovation
Discovery & data science
We analyze real chat interactions, build the model that turns a question into a suggested source + search.
Embedded Innovation
The assist interface + API
We build the interface that takes our output and calls the internal product API to generate a usable search permalink.
Embedded Innovation
Pilot & refine
Service reps use it; early data tells us how to refine the output before it's worth integrating.
Partner teams
Integrate
Once refined, we pass it to LN's team to integrate into NICE or Agent Web, wherever it fits the rep's flow.
Embedded Innovation + Partner teams
Run at scale
It runs in the customer-service platform your teams operate, with us maintaining the model behind it.
In short: we lead the discovery and build of the AI solution through pilot. From there, one of two paths: we partner with your teams to embed it into the platforms you run (like NICE and Agent Web, as here), or we run it as a build of its own, like Alfred+.
Our portfolio by maturity: where each initiative sits today
We track this portfolio as a maturity pipeline: each initiative graduates stage by stage on evidence, along the lifecycle our teams use. Most of our work sits early because we're a new team. The right-hand stages are where it moves into scaled solutions.
Continuous discovery
Phase 1 · Define
Phase 2 · Validate
Phase 3 · Ship
Discovery
parallel track
Always running. Feeds every concept that follows.
Concept
Stages 1–2
Legal Research Copilot
Customer Ops
Concept + delivery plan defined; scope locked to chat-first CSR assist. Re-clustering & evaluation next.
Prototype
Stage 3 · Concept Build
Cash Management
Finance
Reconciliation exception prototype built & SME-validated. Real-data alpha path next.
SME-validated
PPM Workflow Hub
Finance
Stage 3 prototype built; Finance validated the core concept.
SME-validated
Build Alpha
Stage 4
CER · Chief Executive Report
Finance
Deployed; current alpha on the monthly executive readout (Net POS), in an authenticated review loop.
deployed alpha
Alpha → Validate
Stages 5–6
Maxwell
Customer Ops
Production customer-support chat agent on CounselLink/CourtLink. Built across teams; our current work is its email human-in-the-loop enhancement.
in production
▶ Enterprise-app embedding led by Business Systems
Beta
Stage 7
Where validated work reaches real, unsupervised use, and where features expand and grow more sophisticated.
Launch · Iterate
Stages 8–9
Go-to-market & ongoing iteration on live usage.
◀ Embedded Innovation leads here: discovery to validated, build-ready work
Increasingly our partner teams: integration, systems of record, and scale ▶
Stage names follow the Embedded Innovation product lifecycle. The seam is a soft gradient, not a hard line; each initiative hands off at a different point.
Initiatives in flight: progress, value, and what's next
Click any initiative for detail
Progress to date
Built and SME-validated a cash-reconciliation exception prototype; mapped reconciliation rules across wires, FX wires, checks, EFTs, direct debits, and sweeps.
Value & impact
A ~200–400 hours/month opportunity once the variance journal-entry workflow is automated (200+ monthly occurrences × ~1–2 hrs each). Also narrows the duplicate-payment risk window and shortens the Cash↔Accounts-Payable cycle (cycle metric not yet measured).
In flight
Real-data alpha path; raw bank files + Oracle Fusion data access; Fusion write-back scoped with the owning team.
On the horizon
Routine matches Fusion misses are reconciled and written back automatically; manual queue shrinks toward <20% of transactions. Expansion: more AP processes, other RELX businesses, cash forecasting.
Progress to date
Built a Stage 3 prototype for resource managers, project watchers, and PPM operations. Finance validated the core concept: move allocation ownership back to managers and shift the central team from data entry to exception review.
Value & impact
Process touches 4,000+ resources and 11,361 active Fusion projects, plus monthly forecasting/reconciliation. Measured through reduced spreadsheet/form/email handling, fewer Fusion load errors, faster manager resolution, and potential audit (SOX / E&Y) evidence. Financial and time-savings impacts are still being quantified.
In flight
Time-phased allocation model; closed-period reclass flow; direct/indirect project bundle logic; budget / project-watcher discovery.
On the horizon
Closed-loop Workday/Fusion integration: project codes and allocation changes move through a governed workflow and write back to the systems of record, via the teams that own them.
Progress to date
Shifted the monthly executive financial readout from a flat slide deck toward a staged executive-reporting product. Current alpha covers analytical commentary on a core measure (Net POS), preparer/reader surfaces, and an authenticated executive review loop.
Value & impact
Executive decision-speed, not headcount savings: clearer commentary with action, owner, and follow-up. Builds the path toward a disciplined Issue / Action / Owner / Target / Timeline / Results readout.
In flight
Deployed. A roadmap from the Net POS alpha toward a broader continuous-improvement engine has been presented; alignment needed on prioritization.
On the horizon
A continuous tracking-and-improvement engine: leaders enter status directly, AI summarizes it, areas not improving are flagged, and actions/owners/outcomes are tracked over time.
Progress to date
Built the concept definition and delivery plan for the Customer Operations legal-research assist priority. Scope clarified: chat-first, service-rep assist, human-in-the-loop, search formulation and product help only.
Value & impact
~$1.5M estimated value from service-rep time saved. Customer-service baseline to beat: legal-research chat at parity or better than human output while holding customer satisfaction (~78%) and first-contact resolution (71%). Grounded in a corpus of 6,057 interactions and 1,446 question-to-search examples.
In flight
Specification; re-clustering the recent corpus; offline evaluation design; testing off-the-shelf models against a possible custom-trained one.
On the horizon
Suggest-only rep alpha → live measurement against satisfaction / resolution → possible customer-facing capability once accuracy and integrations are proven.
Progress to date
Contributing to Maxwell, a production customer-support chat agent on CounselLink and CourtLink (built across teams). The model work predates higher strategic decisions around incorporating it into Alfred+; current focus is its email human-in-the-loop enhancement, drafting replies for reps to review.
Value & impact
~$1.2M value-add (Customer Ops estimate). Time savings on email handling for customer-ops reps: an AI-drafted reply the rep edits and sends rather than composing from scratch. Value sits with the contact-center teams running it; our piece is the email enhancement.
In flight
Email human-in-the-loop: an interim copy-paste drafting step is live; tighter NICE / AgentWeb integration for in-line drafts is in progress, led by Business Systems.
On the horizon
Incorporation into Alfred+ as the Customer Ops vertical: Maxwell folds in as that platform consolidates, and the standalone brand goes away. Expansion to more products as knowledge-base content is cleaned up; auto-draft when the rep opens an email once NICE-side changes land.
Value figures are point-in-time estimates and opportunities, not booked savings. Where impact is decision-speed or quality rather than hours, it's labeled as such.
What unlocks the value: where we partner
The value above is real but conditional. Most of it unlocks through data and integration the rest of Global Business Systems owns. Two things matter most:
System & data access
Oracle Fusion / Workday access for the Finance work, which turns the cash and PPM prototypes from read-only views into closed-loop solutions.
Customer Ops data path
A NICE / live-chat data path for Legal Research and Maxwell, the live conversation feed the assist solutions depend on.
As we ramp up
As we ramp up across these solution areas, we'll continue reaching out to partner with your teams. We're genuinely appreciative of the support.