From the SLED-wide archive edition of August 27, 2026
A large AI portfolio exposes inventory and workforce control gaps
U.S. Internal Revenue Service · Tax administration · United States
- Publisher
- IRS: Actions Needed to Address Skills Gaps, Information Quality, and Strategic Management
- Original publication
- March 24, 2026
- Source retrieved
- Not recorded in the historical archive
What happened
GAO reviewed how the IRS manages a fast-growing AI portfolio spanning operational tools, voice bots, chatbots, and development-stage use cases.
Why it matters
As portfolios grow, SLED leaders need an inventory connecting each use case to an owner, lifecycle status, data, risk tier, controls, and performance evidence.
Evidence and measured results
The IRS reported 126 use cases by June 2025, with about one-third operational. Eleven voice bots and two chatbots handled nearly 25 million sessions during the 2025 filing season; GAO also found inventory lag, mislabeling, and workforce gaps.
Limitations and uncertainty
The federal tax environment differs from SLED operations, and session volume does not demonstrate answer quality or citizen benefit.
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
- a growing AI inventory can lose contact with deployed systems and available workforce capacity.
- Stakeholders
- portfolio management, enterprise architecture, program owners, risk, privacy, and workforce planning.
- Discovery
- which records lack owners or correct lifecycle states; how is production reconciled to the inventory; and who measures service outcomes?
- Value hypothesis
- a maintained system of record may improve management decisions and expose unmet control or skill needs.
- Potential engagement
- inventory reconciliation and operating-model assessment before automation. GAO's findings of lag, mislabeling, and workforce gaps justify those questions.
- Unsupported claims
- the federal tax setting does not prove the same gaps locally, and nearly 25 million bot sessions do not establish accurate answers, citizen benefit, or cost savings.
Pre-sales engineering
Role takeaway
- Fit
- improve the existing portfolio record rather than assume a new AI platform is needed.
- Architecture and integration
- link each use case to architecture review, dependencies, environment, lifecycle state, monitoring, and retirement records.
- Prerequisites
- authoritative system lists, accountable owners, and definitions distinguishing experiments from operational services.
- Constraints
- inventory lag and inconsistent labels can undermine any automation built on them.
- Security
- record data classification, privacy assessment, authorization boundaries, model monitoring, and incident ownership without exposing sensitive system details broadly.
- Proposed validation
- trace a sample of inventory entries to deployed configurations and source records, and trace deployed AI systems back to entries; measure mismatches and verify corrections before relying on portfolio reporting.
Delivery
Role takeaway
- Work
- reconcile the inventory, assign missing owners, correct lifecycle states, assess skill gaps, and establish a recurring update process.
- Dependencies
- program cooperation, architecture and operations records, and workforce-planning capacity.
- Ownership
- a portfolio owner sets record-quality rules; service owners attest their entries; operations and privacy teams maintain control and incident details.
- Skills and adoption
- train contributors in status definitions and evidence requirements, using real misclassification examples.
- Governance checkpoints
- initial reconciliation, deployment/change approval, and periodic retirement review.
- Proposed acceptance
- sampled records match operational reality, discrepancies have dated remediation owners, and every operational sample identifies monitoring and incident responsibility. Risks include stale manual attestations, counting pilots as production, and reporting session volume without quality or outcome evidence.
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?
Make the use-case inventory a living system of record tied to architecture review, dependencies, production status, monitoring, and retirement.
Governance
Who approves, reviews and stays accountable for outcomes?
Assign portfolio ownership, inventory-quality controls, workforce planning, outcome reporting, and deadlines for correcting lifecycle records.
Security and privacy
What data, permissions and controls need testing?
Link each use case to its data classification, privacy impact, authorization boundary, model monitoring, and incident owner.
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: irs-ai-portfolio