Confidential Proposal

ForgeWorks for LaSalle

AI workflow infrastructure for staffing, starting where req quality and compliance intersect.

Phase 1: Req intake, qualification, and compliance

Prepared for LaSalle Network on March 11, 2026.

The Vault

LaSalle Needs Better Decisions Before More Activity

The call made the priority clear: quality recs determine recruiter focus, margin, and downstream outcomes more than raw speed alone.

At the same time, posting compliance adds immediate legal risk, while disconnected AI experiments create long-term operational drag.

This should read as a platform proposal, not an AI pilot: one hosted layer, one first module, clear room to expand.
Operating Metric
Fill / Wash / Loss
Quality is the real north star because better reqs are more likely to go to work.
Risk Exposure
Posting Compliance
Illinois and California requirements need a repeatable gate before a job is posted.
Platform Need
One Hosting Layer
Centralized orchestration on AWS / Bedrock beats a growing pile of isolated bots.
Current Friction

Judgment Is Trapped in Notes and Handoffs

The current path works, but the highest-value decision points are manual, uneven, and hard to scale.

1
Capture
Customer Call
Salesperson gathers requirements in conversation.
2
System
Bullhorn Notes
Notes are entered, but the structure and completeness vary.
3
Judgment
Manager Review
Jeff and the team decide if the req is worth recruiter time.
4
Control
Compliance Gate
Public-posting rules must be checked and documented.
5
Launch
Recruiter Handoff
Only qualified reqs should trigger recruiter effort.
Missing information surfaces late
The team often finds the real gaps only after leadership review begins.
The quality bar lives in people
Jeff's judgment is valuable, but it is not yet a reusable operating model.
Compliance runs as a parallel risk loop
Posting language still needs its own repeatable check and approval trail.
Proposed Architecture

ForgeWorks Connects Bullhorn, Policy Inputs, and Approval Decisions

Do not replace Bullhorn. Start by wrapping ForgeWorks around the workflow LaSalle already runs.

It ingests notes, applies policy and rubric logic, routes approval, and returns auditable outputs from one hosted tenant.

Not a Bullhorn replacement
Hosted on AWS
Approval + audit trail
Module-based expansion
Inputs
Bullhorn
Account-manager notes, req records, historical signals.
Context
Policies + Rubrics
Compliance language, Jeff's criteria, approval rules.
ForgeWorks Core
Workflow Orchestration
Model routing, structured extraction, qualification logic, approval workflow, memory, and audit trail.
Outputs
Approved Rec
Recruiter-ready packet, gap report, compliant posting draft.
Expansion
More Modules
Candidate quality, decision memory, additional workflow automations.
Bullhorn connector
Approval routing
Audit history
Observability cadence
Phase 1: Requisition Intake, Qualification, and Compliance

Start here because it combines the two live pains from the call: req quality and public-posting compliance.

The output is an approved, recruiter-ready req, plus a compliant public-posting draft when LaSalle needs one.

  • Recruiter-ready req packet
  • Gap report for the account manager
  • Leadership approval decision trail
  • Marketing-ready public-posting draft
Capture
Bullhorn Note Ingest
Pull the raw notes LaSalle already requires from account managers.
Structure
Field Extraction
Convert unstructured notes into a structured req profile with evidence.
Quality
Missing-Info Detection
Flag what is absent, weak, or contradictory before recruiter time is spent.
Judgment
Req Scoring
Apply Jeff / Zach criteria so leadership reviews a structured recommendation, not a blank slate.
Governance
Compliance Controls
Check and append required Illinois / California posting language when relevant.
Workflow
Approval Gate
Route the req for leadership approval before recruiter or marketing handoff.
Phase 1 Workflow

This Workflow Lives Inside the ForgeWorks Platform

Req intake, compliance, and approval are not disconnected tasks. They become modules on one governed operating layer.

1
Ingest
Pull notes into a governed intake point.
2
Score + Check
Score the req and apply quality and compliance controls.
3
Approve + Release
Approve once, then route the req onward.
ForgeWorks Platform View
Concept illustration of the ForgeWorks platform as a modular workflow operating layer
A modular platform, not a point solution
Req intake, compliance, and approval routing sit on one platform instead of becoming separate tools.
The first LaSalle workflow becomes the foundation for later modules, not a dead-end automation.
Success Measures

Success Is Measured in Quality, Not Just Speed

Quality comes first because better reqs drive margin. Speed matters only after LaSalle knows the req is worth the team's time.

Better fill / wash / loss signal
Tie req quality decisions to downstream outcomes.
Fewer recruiter dead ends
Protect recruiter attention by filtering weak reqs earlier.
Faster approved handoff
Compress the path from notes to approved recruiter-ready req.
Traceable compliance review
Capture what was checked, applied, and approved.
Launch Milestone
Software live in 30 days
Hosted ForgeWorks instance, initial rubric, and approval path in place.
Operating Review
Check the quality signal
Review recruiter effort, req quality, and compliance trace with real usage data.
Expansion Path
Add the next module when ready
Use the same platform foundation for candidate quality, decision memory, or another workflow.
Module Roadmap

Module Roadmap After Phase 1

Show deliberate expansion, not a promise to solve all 15 use cases at once.

Phase 1
Req Intake + Compliance
Turn raw notes into an approved, compliant, recruiter-ready req.
Phase 2
Candidate Submission Quality
Check fit, missing evidence, screening completeness, and submission readiness.
Phase 3
Decision Memory
Capture leadership reasoning so judgment becomes reusable and trainable.
Phase 4
Additional Workflow Modules
Add more LaSalle use cases onto the same tenant, governance model, and orchestration layer.
Technical Appendix

Technical Fit for LaSalle's Stack

Answer hosting, model choice, security, and observability without turning the whole proposal into an infrastructure pitch.

AWS / Bedrock Alignment
Hosted in a governed tenant
Use AWS-hosted services and Bedrock-compatible model access instead of unmanaged chat interfaces.
Model Strategy
Pick the right model for the job
Use smaller task-fit models for extraction and classification, reserve larger reasoning models for higher-order analysis.
Data Ownership
LaSalle owns its software and data
The platform uses APIs and hosted infrastructure, not consumer chat interfaces training on operational data.
Observability
Tune on a cadence, not by drift panic
Use regular check-ins, user feedback, and outcome review to improve modules safely over time.

Important boundary: this is workflow infrastructure around existing systems, not a project to replace Bullhorn, Salesforce, and email wholesale.

How It Starts

4-5 Meetings. 30 Days to Software.

We align on the workflow quickly, lock scope, and move directly into build for the first module.

Working Sessions + Scope Package
[$ Align with Brendan]
Four or five working sessions to lock workflow, module scope, and the 30-day build plan.
Phase 1 Platform Engagement
[$ Align with Brendan] / month
Hosted ForgeWorks instance, Phase 1 module, support, and tuning cadence.
Usage + Expansion Terms
Scope-dependent
Usage above baseline, extra integrations, and future modules are scoped separately.
Pricing, Terms, and Next Step
Approve commercial review and scope confirmation.

Next step: Brendan alignment and the final Phase 1 scope package.

Schedule Proposal Review
Braydon McCormick
braydon@lightforgeworks.com lightforgeworks.com/contact
Prepared for LaSalle Network. Proposal valid through April 10, 2026. Pricing assumes Phase 1 only; added modules, non-baseline usage, and final terms are scoped separately and remain subject to Brendan alignment plus executed agreements.