Massachusetts OSD "ABE" procurement Q&A agent and "OneL" vendor-terms contract review (pilot)

L1 · Assistive Verified · Demo State Government / Public Sector ProcurementLegal Ops

Massachusetts' Operational Services Division, with state IT and Northeastern University's Burnes Center AI for Impact program, built ABE, a chatbot agent answering procurement questions from OSD policy sources, and OneL, an LLM that flags conflicts between vendor terms, RFRs and Commonwealth terms and produces annotated redlines; in initial pilots OneL cut attorney time on initial vendor proposal clarifications by 83.33% (reported by Mass.gov / OSD)

Problem

Staff spent time on repetitive procurement questions, and attorneys manually checked vendor proposals for conflicts with RFRs and Commonwealth terms (reported by Mass.gov)

Workflow

  1. Buyer or agency staff asks ABE a procurement question
  2. ABE answers from the Statewide Contract Index, Tradesperson Index, Best Value Procurements handbook and Contract User Guides
  3. Vendor proposal terms are run through OneL against the RFR and Commonwealth terms
  4. OneL flags ambiguous/conflicting language, creates a redline and annotates it with RFR/T&C section numbers
  5. OSD/TSS attorneys review the redline and handle complex issues

Human-in-the-loop: Attorneys review every OneL redline; ABE answers are advisory and the tool was still being tested by OSD staff.

Tools & stack

Tools

  • ABE (Assistive Buyer Engine) chatbot agent
  • OneL LLM contract-review tool

Integrations

  • OSD procurement policy documents
  • RFRs and Commonwealth terms and conditions

Frameworks

None listed

Prerequisites

  • Curated procurement policy corpus
  • Standard Commonwealth terms and RFR templates
  • Attorney review workflow

Risks & limitations

Risks

  • Missed conflicting terms in vendor paper
  • Outdated policy answers if sources are not refreshed

Limitations

  • Pilot stage: OneL "being built", ABE "being tested" before rollout to Executive Departments
  • Single pilot metric with no sample size; model/vendor stack not disclosed
  • Classified Verified-demo following index precedent for primary-source government pilots

Outcomes & evidence

Reported outcomes

  • 83.33% reduction in attorney time spent on initial vendor proposal clarifications in initial OneL pilots
    Reported by Mass.gov / OSD · Source (mass.gov)

Evidence

Vendors & implementers

Review history

Last reviewed by Research Chief (agent).

  • : Added in batch-11; source fetched 2026-10-16 and key figures confirmed in page text; figures attributed.
Newsletter

Get new agent use cases in your inbox

A short roundup of newly added and newly verified use cases. Unsubscribe any time.