From the SLED-wide archive edition of August 27, 2026
AI acquisition reviews rarely capture reusable lessons
U.S. Government Accountability Office · Public procurement · United States
- Publisher
- AI Acquisitions: Agencies Should Collect and Apply Lessons Learned
- Original publication
- April 13, 2026
- Source retrieved
- Not recorded in the historical archive
What happened
GAO examined 13 AI acquisitions at four federal agencies and found that procurement processes did not systematically preserve lessons for future buyers.
Why it matters
SLED buyers face similar risks when contracts omit evaluation data, performance thresholds, portability, audit access, or an exit path.
Evidence and measured results
Four agencies lacked systematic lessons-learned requirements, missing reusable learning on data rights, testing, regional model accuracy, and discontinued solutions. All four concurred with GAO’s recommendations.
Limitations and uncertainty
The sample is federal and limited to 13 acquisitions; local procurement statutes and market conditions vary.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source as summarized in the preserved archive. Enriched 2026-09-05; this does not change the original publication date. Labels below come from the analysis itself.
Sales
Role takeaway
- Customer problem
- AI buyers can repeat avoidable contract and testing mistakes when prior procurement lessons disappear.
- Stakeholders
- procurement, legal, program sponsors, data owners, enterprise architecture, and security.
- Discovery
- where are evaluation results and failed purchases recorded; which rights cover customer data; and how are regional accuracy and exit needs tested?
- Value hypothesis
- reusable acquisition evidence may improve requirements and reduce poorly understood acceptance risks.
- Potential engagement
- review selected past acquisitions and build the next solicitation's evaluation and exit criteria. The audit's 13 acquisitions identify concrete questions about testing, data rights, and discontinued solutions.
- Unsupported claims
- this limited federal sample does not establish failure rates or financial benefit for SLED buyers, whose statutes and markets differ.
Pre-sales engineering
Role takeaway
- Fit
- apply the findings at requirements and pre-award evaluation for a defined AI use case.
- Architecture and integration
- specify system boundaries, data exchange, monitoring, performance acceptance, portability, and an executable exit route.
- Prerequisites
- representative permitted test data, regional use conditions, ownership of expected answers, and procurement/legal agreement on enforceable evidence.
- Constraints
- a supplier demonstration may omit local data or integration conditions; discontinued components can defeat an undocumented exit plan.
- Security
- require audit access, incident duties, data-use/deletion terms, model-change notices, and subcontractor controls. Proposed proof: exercise representative customer scenarios, failure handling, data export and deletion, and record results against agreed requirements before selection or acceptance.
Delivery
Role takeaway
- Work
- capture acquisition lessons, convert them into contract and test requirements, and retain evidence through acceptance, renewal, and exit.
- Dependencies
- procurement schedules, legal review, vendor cooperation, and access to technical findings from prior projects.
- Ownership
- procurement maintains the shared record; technical and program owners validate performance; legal/data owners resolve rights and deletion obligations.
- Skills and adoption
- teach buyers to distinguish supplier assurances from demonstrated evidence and make lessons easy to reuse.
- Governance checkpoints
- solicitation approval, acceptance, renewal, and discontinuation review.
- Proposed acceptance
- each agreed contract threshold has test evidence, unresolved gaps have accountable remediation, and export/exit steps are documented and exercised where feasible. Risks include generic clauses, inaccessible lessons, and criteria that cannot be enforced under local procurement rules.
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
Require pre-award test plans, integration boundaries, portability, performance acceptance criteria, observability, and a documented exit architecture.
Governance
Who approves, reviews and stays accountable for outcomes?
Use AI-specific solicitation clauses and a shared lessons repository; assign procurement, legal, data, and technical owners to acceptance and renewal decisions.
Security and privacy
What data, permissions and controls need testing?
Contract for audit access, incident duties, data use and deletion, model-change notice, subcontractor controls, and security testing evidence.
The preserved archive analysis covered architecture, governance and security. Not assessed for this record: accessibility and workforce, procurement, operating model.
Publication history
- 2026-08-27SLED-wide archive · Issue 0110 resources
Stable resource ID: ai-acquisition-lessons