The Agentic Turnaround Office

Abstract

A company can have a cash forecast, an improvement program and experienced advisers while still struggling to decide what to do next. Each function sees a different part of the situation. An overdue invoice may reflect missing acceptance documents, a quality dispute or customer distress. An order that looks profitable may require cash the business needs for payroll before the customer pays. Completing a project on schedule says little about whether it has restored margin or improved cash flow.

This paper proposes an Agentic Turnaround Office that connects operating evidence, competing explanations, liquidity timing, professional constraints and execution responsibility around specific recovery decisions. It draws on established turnaround, systems-engineering, transformation-office, organizational-change and AI-governance work. Its contribution is a recovery-specific design and evaluation approach: a decision remains connected to its evidence, authority, implementation conditions and observed result. A worked cash-and-operations example demonstrates how that design changes the sequence of action and exposes a fragile plan. The paper then specifies where agents can assist, how to compare them with manual work and what would count as failure. The architecture requires evaluation in actual practice; its usefulness depends on whether it improves decisions and execution after the full cost of coordination and oversight.

1. Recovery work begins with a decision that cannot wait

The operations leader wants components purchased so a late order can ship. The finance leader sees a cash cushion disappearing before the customer pays. Sales expects an overdue invoice to clear, but nobody has checked whether the customer accepted the work. Counsel knows that a commitment under discussion could constrain the next financing conversation. Each concern may be reasonable. The company needs a sequence it can carry out.

Management needs to decide which actions to authorize this week, what must happen first and when to change course. That requires facts from across the business and someone who follows the decision through procurement, production, customer acceptance and collection. The financial result depends on that entire sequence.

A Turnaround Office provides that temporary coordinating capacity around a recovery mission. Its remit is defined by the company’s actual situation: restoring delivery reliability, stabilizing liquidity, recovering contribution, resetting a failing program or resolving interacting constraints that management cannot presently control. It establishes the working connections between operating owners, qualified finance leadership, qualified legal or restructuring counsel and authorized executives.

Agents support the people doing this work. They can retrieve authorized records, reconcile identifiers, compare commitments with current evidence, calculate scenarios and prepare exceptions for review. The team must still verify its explanation of the problem and test whether the proposed intervention is viable. Agent support is useful when it improves the information and follow-through available to those responsible.

The design needs to answer how a recovery decision can remain defensible as conditions change and how the office can carry it into action without creating another reporting burden. Agent assistance also needs to justify its preparation, review, correction and technical costs.

The office manages consequential decisions and the conditions for carrying them out. An instruction to “expedite components” needs enough context to act safely: which order needs them, why it is late, what other constraints remain, when cash leaves, what makes the customer liable to pay, who can authorize the commitment and when the result should be visible. Recording those details is justified where they improve the decision or make execution more reliable.

2. What established work already provides

This is an integrative design paper, informed by a scoped review completed on 17 October 2026. The review covered turnaround and restructuring, crisis management, systems engineering, PMO and transformation offices, working capital, negotiation, organizational change, human-AI work, operating conditions and competing services. It used full papers and official materials where accessible, with access limits recorded separately. It is not an exhaustive systematic review or an evaluation of client outcomes.

Turnaround research already treats recovery as a combination of operational, managerial, portfolio and financial responses. Schweizer and Nienhaus emphasize the variety of contexts, processes and measures involved; no single intervention follows from the label “distress.” [1] Systems engineering contributes disciplined treatment of requirements, alternatives, assumptions, interfaces, trade-offs and verification. Applying NASA’s methods to corporate decisions requires adaptation to organizations whose objectives, authority and stakeholder preferences can change during the work. [2]

The PMO literature also extends beyond schedule administration. Crawford describes an enterprise PMO concerned with strategy execution and organizational results. [3] Duchek’s resilience framework connects anticipation, coping and adaptation, suggesting why a recovery engagement should leave the organization better able to notice and respond after the immediate mission ends. [4] These traditions provide much of the architecture’s foundation.

Recent work makes the overlap with agentic transformation explicit. McKinsey’s September 2026 article describes an agentic transformation office using connected information, performance intelligence and leadership support while humans retain stakeholder judgment. [5] BCG’s July 2026 article presents a future AI-powered office spanning program, finance and change work. [6] Agent-supported project management is also an active research subject. [7] Combining agents with an execution office is therefore an established direction of work.

The contribution proposed here is narrower and testable: organize recovery decisions so that operating causes, cash timing, professional constraints, authority and verified execution remain linked. A credible evaluation must compare this approach with a capable finance-and-operations team using ordinary project controls. An improvement against fragmented spreadsheets alone would not show that a new office or agents were necessary.

The evidence also argues for restraint in assigning tasks to AI. Dell’Acqua and colleagues found that assistance could improve work within some task boundaries while worsening correctness outside them. [8] Vaccaro and colleagues’ synthesis shows that human-AI combinations do not reliably outperform the stronger component in every setting. [9] Research on multi-agent failures identifies problems in specification, coordination and verification. [10] These findings support task-specific testing, explicit responsibility and recovery controls. They do not predict an ATO performance effect.

3. Design the office around connected decisions

The proposed office has three professional lanes beneath the client’s authorized executive or board authority. The systems and Turnaround Office lead integrates the mission, interfaces, decision cadence and execution. Qualified finance leadership owns the liquidity model, financial interpretation and finance recommendations. Qualified legal or restructuring counsel owns legal advice, legal process and the constraints attached to commitments. Mission specialists join for a defined need; their authority follows their actual appointment and engagement.

The architecture places client executive authority above those three lanes. They share decision and execution controls, while agents support approved evidence, analysis and monitoring tasks. Workstream owners remain responsible for delivery. An office title creates no officer appointment, lending authority or right to negotiate on another party’s behalf.

The common record for a material decision contains eight connected elements:

The office can maintain this structure in existing controlled records. It does not require a replacement enterprise platform. A small number of decision records, supported by the finance model and operating evidence, may be sufficient for an initial mission.

Freshness matters as much as completeness. A customer’s willingness to accept delivery can change. A forecast receipt can become disputed. A component may arrive while the qualified operator becomes unavailable. The office therefore attaches review timing to the consequence: daily or intraday for an immediate cash exposure, weekly for an operating recovery workstream, and at defined events for a contract-dependent action.

Disagreements remain visible until someone with the relevant responsibility resolves them. “Finance says paid” and “customer says not accepted” should not disappear into a blended status. They may refer to different invoices, dates or entities. Identity reconciliation is substantive work because the wrong join can produce an apparently coherent but false plan.

Bank cash may be restricted, pledged, offset or needed for obligations omitted from an operating extract. In Chapter 11, for example, the U.S. Courts’ guidance describes cash-collateral consent or court-approval requirements. [11] The responsible professional must determine whether the company can use that money for the proposed action.

4. A worked operating and cash decision

Illustrative case. The following manufacturer, orders, amounts and timings are constructed to make the decision reproducible. All cash tables use thousands of U.S. dollars. A six-week window is shown within a finance-owned rolling forecast; the decisions also require a daily view around tight cash dates.

The company opens week 1 with $500,000 of unrestricted cash. Management has set a $300,000 operating cushion. This is an internal planning threshold, not a stated covenant or legal requirement. The baseline reflects scheduled operating payments, payroll, tax, debt service and other known obligations for this example. It includes neither the disputed invoices nor receipts from an unfinished order considered below. Normal labor and already-paid inventory are included in the baseline or opening position. The comparison excludes newly commissioned office, professional or financing fees and assumes no new financing; an actual go/no-go decision must add their dated cash effects before judging the cushion.

Retaining the present plan breaches management’s cushion in week 4. The forecast exposes a timing problem but does not identify its cause. A 13-week cash forecast is an established tool for managing that visibility; A&M describes its use in short-term liquidity control. [12] The operating work must determine which receipts can actually change and what it takes to obtain them.

4.1 Separate a collection problem from an acceptance problem

The receivables ledger contains $120,000 of old invoices. An aggregate aging report invites a simple instruction: collect faster. The office examines each invoice against the order, shipment, acceptance requirements, customer correspondence and responsible account owner.

The first $60,000 concerns completed work with missing acceptance documentation. The operating owner can supply the required evidence in week 1. For the modeled receipt to enter week 2, the customer must confirm that the documents resolve the issue and identify a payment date. Sending the documents alone is insufficient. Incremental administrative work costs $2,000 in week 1.

The next $40,000 involves a genuine quality defect. Collection calls cannot repair it. A qualified operating owner verifies the remedy, capacity and acceptance criteria. The scenario assumes $8,000 of repair cost in week 2 and customer acceptance followed by payment in week 5. The final $20,000 lacks evidence sufficient to forecast collection within the window. It remains outside the scenario receipts; finance determines its accounting treatment separately.

This distinguishes at least four explanations for late cash: a documentation failure, a delivery or quality failure, an incorrect invoice and a customer unable or unwilling to pay. Each requires different evidence and action. If customer liquidity is the cause, more internal automation will not produce the assumed payment. If the invoice is wrong, the appropriate result may be a corrected claim rather than collection of the original amount.

The receivables-repair option adds $100,000 of conditional receipts and $10,000 of payments across six weeks. Closing cash becomes $418,000, $390,000, $360,000, $330,000, $350,000 and $330,000. It stays above the modeled cushion and ends $90,000 above baseline. That is a cash difference over this horizon. It is not $90,000 of new sales or recurring profit.

4.2 A profitable-looking shipment can worsen the immediate constraint

The second opportunity is an unfinished order for 80 units at $2,000 each: a $160,000 customer payment after acceptance. Completing it requires $40,000 of components in week 1 and $7,000 of incremental overtime and freight in week 2. Existing inventory has already been paid for, and ordinary labor is in the baseline. The scenario assumes shipment and acceptance in week 2, followed by payment in week 4 under the agreed terms.

Completion alone ends week 6 with $353,000, or $113,000 more cash than baseline. Yet it reduces closing cash to $293,000 in week 2 and $263,000 in week 3. The attractive end position hides an earlier shortfall against management’s operating cushion.

Combining immediate completion with the receivables repair appears feasible: the lowest week-end balance is $313,000. The margin is thin. If the $60,000 documentation receipt moves from week 2 to week 3, week 2 falls to $283,000. A plan that depends on two separate events occurring in a particular order needs explicit conditions for commitment.

The operating hypothesis also needs challenge. Missing components may be the visible problem while the actual constraint is quality approval, skilled labor, machine capacity or customer willingness to receive the shipment. Before a purchase, the workstream owner verifies that components remove the limiting constraint and that all remaining acceptance conditions can be satisfied. The OCC’s lending guidance similarly emphasizes the quality of receivables and inventory and the relationship between cash needs and repayment capacity. [13] A ledger amount or a production milestone is insufficient evidence of available cash.

4.3 Change the sequence before increasing the commitment

The office develops a staged alternative for the authorized account owner to discuss with the customer. It seeks an $80,000 deposit in week 2, then buys components for $40,000 in that week. Overtime and freight of $7,000 occur in week 3, with shipment and acceptance in week 3 and the final $80,000 payment in week 5.

The deposit is a proposed contractual change requiring agreement. It is not treated as an existing promise. Counsel reviews relevant terms, including refund, security and acceptance implications; finance determines the cash and accounting treatment. Procurement receives authority to proceed only after the deposit is verified as available and the operating owner confirms capacity and quality conditions.

The staged option ends with the same $443,000 as the immediate combined plan. Its lowest modeled week-end balance is $363,000 because the deposit arrives before the component purchase. The revised sequence improves the cash position during the work without adding to the final cash benefit.

The six-week increase over baseline is $203,000: $100,000 of old-invoice receipts plus $160,000 of order receipts, less $10,000 of receivables-repair costs and $47,000 of order-completion costs. The calculation does not establish order margin because historical inventory cost, ordinary labor, overhead and revenue recognition require their own treatment. Nor should the $160,000 be counted again as “revenue recovered” in a combined benefit total. The model follows cash dates; the financial statements answer different questions.

4.4 Stress the events that can reverse the decision

Three tests reveal the staged plan’s limits. First, moving the $60,000 documentation receipt from week 2 to week 3 lowers week 2 cash to $370,000; the week-end cushion still holds. Second, moving both the $40,000 repaired-invoice receipt and the final $80,000 order receipt beyond week 6 leaves $323,000 at the end. Third, if the $60,000 documentation receipt is also outside the window, week 5 falls to $283,000 and week 6 to $263,000.

The third case requires a different response before the shortfall arrives. Finance must assess the broader forecast, financing availability and payment obligations; counsel must assess any relevant restrictions. Management may need to narrow commitments, obtain permitted financing, change terms or pursue a different recovery route. Those decisions cannot be inferred from this six-week extract.

If the deposit does not arrive, the order-completion purchase remains unauthorized under the staged plan. The office returns to the receivables-repair option and updates its assumptions. If the customer rejects the repair or documentation, the relevant receipt is removed until a new basis exists. If procurement lead time or production capacity changes, the entire linked shipment-and-payment sequence is recalculated.

Weekly closing cash is also an incomplete control. A payroll payment on Monday and a customer receipt on Friday can create a midweek need that a closing balance conceals. The finance lead checks actual dates, bank availability and obligations beyond the displayed horizon before authorizing the sequence.

4.5 Execute against observable acceptance

The work now has distinct owners and evidence. The account owner obtains the customer’s exact acceptance and payment conditions. Operations completes documentation or repair and verifies capacity. Finance confirms receipts and updates the forecast. Procurement follows the conditional authority. Counsel handles the relevant contractual questions. The Turnaround Office maintains the dependencies, exceptions and decision deadlines.

“Document sent,” “repair finished,” “invoice issued” and “cash received” are separate events. The office closes each with its own evidence. It checks that a receipt belongs to the correct invoice and is available for use. If an expected event fails, the record identifies which downstream commitments are affected and who must reconsider them.

The resulting sequence preserves cash for the next commitments and identifies the fallback if the deposit does not arrive. The team can carry it out with a spreadsheet, source documents and disciplined human coordination. Agent assistance should be tested against that approach before the team relies on it.

5. Assign agents work that can be checked

An initial agent-supported workflow might assemble the evidence for the receivables review. Its permitted inputs could include approved invoice exports, order records, shipment confirmations and selected correspondence. It could flag mismatched amounts, missing acceptance evidence, contradictory dates and stale payment assumptions. Its output would cite the specific records requiring human resolution.

It should not infer customer consent from silence, rewrite commercial terms, decide credit treatment or send collection messages because an invoice is old. A generated explanation is a candidate to check against the records. External communications and consequential actions follow the client’s actual approval structure.

NIST’s AI Risk Management Framework supplies a useful organizing discipline for governance, context, measurement and management. Its Generative AI Profile adds concrete attention to fabricated content, evaluation and monitoring. [14–15] For this workflow, those ideas become testable operating controls: approved sources, limited permissions, identity checks, timestamps, retained evidence, exception routing and a working manual fallback.

A shadow evaluation can use a fixed set of authorized cases containing known problems: duplicate invoices, partial receipts, revoked acceptance, conflicting entities, obsolete forecasts and missing documents. A reviewer records whether the system detects each material issue, whether it invents facts, how much correction it needs and whether its output would support the intended decision. Test cases should include ordinary clean records so that an excessive false-alarm rate is visible.

The cost comparison must include all the work. Consider a separate illustrative estimate for six weeks of evidence handling. A manual approach needs 12 setup hours plus three hours each week: 30 hours. At an assumed economic cost of $100 per hour, its cost is $3,000. An assisted approach needs 16 setup hours, 1.5 review hours per week and $600 of technology cost: $3,100. These are planning assumptions, not measured productivity results or operating costs included in the cash table.

With those assumptions, assisted work reaches equal cost only if total review and associated operating effort is eight hours across six weeks. Any preparation, monitoring, exception handling or correction omitted from that allowance must be added. Lower cost is useful only at the required quality. Better coverage or earlier detection might justify a higher cost, but that benefit must be demonstrated against the manual alternative.

Three measures remain separate: accepted outputs per total human-hour, fully loaded cost per accepted output, and quality or missed-risk rates. Holding task mix and acceptance standards constant makes comparisons interpretable. Fast drafts that create more downstream repair can reduce actual productivity.

The workflow also needs failure recovery. Before retrying an uncertain write, it checks the destination to determine whether the prior action succeeded. It preserves completed work, identifies stale inputs and pauses the affected branch when authority or access expires. A technical owner controls runtime changes and deployment. The office can continue approved manual work while an automated path is unavailable.

6. Mobilize, execute and transfer without creating permanent dependency

A recovery engagement begins with a legitimate sponsor, an actual mission and the authority to obtain evidence and make decisions. Mobilization establishes immediate commitments, critical dates, professional coverage, the cash and constraint inputs, and the first decisions requiring action. It should produce a usable operating picture quickly while exposing unresolved information that could change the plan.

Recovery design develops alternatives, tests the causes of the problem and turns the selected approach into a detailed implementation plan. It identifies work that can begin immediately, the people responsible, the resources and costs involved, and the decisions or dependencies that determine the next steps. Dates follow the urgency and sequence of the actual work. Each action also needs a financial explanation: for example, how reducing rework should lower cost, how accepted deliveries should become cash, or how a capacity change could support profitable demand. Finance and operating owners establish the baseline and check whether those effects occur.

Immediate continuity and longer-term recovery can require different choices. Cost reduction can improve short-term cash while weakening the capability needed for recovery; continued investment can preserve capability while exhausting liquidity. Management needs both the operating consequences and the financial trade-off before committing resources.

Execution uses a cadence matched to the situation. A short daily exception review may be appropriate for immediate commitments. A weekly integrated review can address assumptions, forecast changes, accepted work and decisions that cross workstreams. Meetings earn their time by resolving something the responsible owners cannot settle through normal work. Repeated status transcription is a defect to remove.

Organizational readiness deserves equal attention. Weiner’s framework separates shared commitment from belief in collective capability. [16] For a recovery mission, agreement with the plan does not establish that an operating owner has time, information, skill or authority to execute it. The office checks those conditions before treating an assignment as accepted and adjusts workloads or scope when capacity is missing.

Stakeholder communication is part of execution. The team needs a common factual basis, an identified speaker and a clear account of what is known, uncertain and being done. Crisis-communication guidance provides useful discipline around timeliness, credibility and audience needs, adapted to the corporate setting. [17] Counsel and authorized leaders determine what may be disclosed and negotiated with lenders, customers, employees or other parties.

Negotiation preparation includes feasible alternatives and the conditions under which agreement would be worse than no agreement. Cohen and colleagues’ experimental work on the agreement trap illustrates the importance of recognizing an unattractive bargain. [18] The office can prepare commitments, trade-offs and evidence; responsible humans conduct negotiations and decide whether to bind the company.

Transfer is designed from the start. The receiving operating owner must be able to run the cadence, update the model, identify exceptions and obtain decisions without undocumented help from the temporary office. At closure, the team records remaining risks and commitments, assigns continuing work and settles who retains the records and access. A successful transfer may include continuing selected controls and retiring others.

The team can simplify how it carries out mandatory legal, safety and professional controls while retaining their requirements. It should also simplify discretionary controls when the burden exceeds the risk reduction they provide.

The commercial design follows these stages: a defined mobilization, renewable execution capacity and an accepted transfer. Specific scope, staffing, response expectations, fees and spending limits depend on the mission and professional arrangements. The architecture does not supply a universal turnaround price or imply that all missions need the same duration or technology.

7. Test the architecture against a capable alternative

The first evaluation should test the delivery method before making broad outcome claims. A complete synthetic exercise can check whether the team reconstructs source evidence, identifies conflicts, calculates scenarios, routes authority, executes simulated exceptions and transfers control. An authorized engagement can then test usability and observed results under actual constraints.

The comparison should preserve a capable baseline: the client’s existing finance, operations, legal and project-control practices. Where possible, comparable decisions can be handled with and without the additional structure, or introduced in a staged manner. Differences in urgency, team experience, data quality and case complexity must be recorded. A simple before-and-after improvement during a changing recovery cannot isolate the architecture’s effect.

Financial outcomes require separate definitions. Earlier collections can improve liquidity without increasing revenue. Reduced inventory can release cash while producing a loss or weakening service. Released staff time is capacity until an actual cost changes or useful output increases. New financing adds funding and obligations; it is not evidence of operating improvement. Opening cash, dated funding needs, financing sources and financing costs belong in distinct parts of the model.

Evaluation should report adverse outcomes as well as gains. The office fails if it slows urgent decisions, creates competing versions of truth, hides uncertainty behind polished summaries, consumes scarce operating capacity or encourages action outside authority. Agent assistance fails if review and repair erase its benefit, if material exceptions are missed or if it makes an unauthorized change. The proposed design should be narrowed, simplified or stopped when these failures persist.

Several limitations remain. A synthetic case demonstrates internal consistency, not effectiveness in a distressed company. The underlying literature spans different industries, scales and institutional settings. Conceptual frameworks do not establish causal impact. Consulting publications describe services, practitioner experience or visions with different evidentiary weight from independent evaluations. Outcomes in recovery also depend on market conditions, creditor choices, customer behavior and the quality of leadership. An operating office can improve coordination while a business still lacks a viable recovery path.

8. Fit, access and the first useful engagement

CXOJoe is initially developing the offer in Tampa, Lakeland/Polk and Orlando. This geographic focus is a market-development choice. Trusted finance, restructuring-counsel, lender, private-credit, PE operating-partner, CPA, banker, board and executive relationships are useful access routes because they may understand the situation and the relevant authority. Each relationship and introduction needs to be established on its own evidence.

National operating data help frame the research questions. The Federal Reserve’s 2026 employer-firm report describes cost, tariff, revenue and financing pressures among small employers. [19] The National Center for the Middle Market’s midyear survey concerns a different revenue-defined population. [20] Regional BLS series describe labor markets. [21] Determining whether a particular company needs recovery work requires evidence about that company. A transaction, expansion, hiring change or workforce notice is a reason to investigate before forming an operating hypothesis.

The entry route follows the problem. For one recurring operating problem with time for investigation, the Executive Systems Diagnostic provides two weeks of investigation and analysis for $5,800. CXOJoe tests the explanation against the evidence, compares practical solutions and develops a detailed implementation plan. The plan identifies the first actions, accountable owners, costs, dependencies and measures, so management can begin the work as soon as the necessary decisions are made. It explains how the proposed changes should improve financial health through their effects on profit, cash flow or the capacity to serve profitable demand.

A mission involving interacting recovery constraints may need Turnaround Office qualification and mobilization. An imminent liquidity, legal, safety or regulatory exposure needs the responsible executive and qualified professionals immediately. Operations Optimizer is commercially active through customer-specific implementation and can be considered where its verified capabilities fit the work. Existing systems, manual controls or another implementation path may also be appropriate. A recovery mission must remain executable when a proposed software capability is unavailable.

9. The standard of usefulness

The Agentic Turnaround Office proposed here brings the people responsible for operations, finance and legal matters into the decisions that determine whether a recovery plan can be carried out. They need current evidence, a defensible explanation of the problem and a feasible sequence of work. The office follows that work through acceptance and brings decisions back for review when conditions change.

In the worked example, those connections reveal why a strong end-of-period cash result can conceal an earlier constraint, why overdue invoices require different interventions and why a deposit-conditioned sequence changes the available room to act. Agents may make parts of that work faster or more reliable. The comparison must include their full operating cost and the quality of the resulting decisions.

Evaluation must also establish whether the company accepts the recovery work and whether its operating owners can continue controlling it after the office transfers responsibility.

References

1. Schweizer, L., and Nienhaus, A. (2017). Corporate distress and turnaround: integrating the literature and directing future research. Business Research. https://link.springer.com/article/10.1007/s40685-016-0041-8
2. NASA (2016). NASA Systems Engineering Handbook, Revision 2. https://www.nasa.gov/wp-content/uploads/2018/09/nasa_systems_engineering_handbook_0.pdf
3. Crawford, J. K. (2010). The enterprise PMO as strategy execution office. PMI Global Congress paper. https://www.pmi.org/learning/library/strategic-project-management-office-execution-6737
4. Duchek, S. (2020). Organizational resilience: a capability-based conceptualization. Business Research. https://link.springer.com/article/10.1007/s40685-019-0085-7
5. McKinsey & Company (16 September 2026). The agentic transformation office: Redefining the economics of change. https://www.mckinsey.com/capabilities/transformation/our-insights/the-agentic-transformation-office-redefining-the-economics-of-change
6. Boston Consulting Group (27 July 2026). The AI-Powered Transformation Office. https://www.bcg.com/publications/2026/ai-powered-transformation-office
7. Assalaarachchi, L. I., Masood, Z., Hoda, R., and Grundy, J. (2026). Toward Agentic Software Project Management: A Vision and Roadmap. Preprint, version 1. https://arxiv.org/html/2601.16392v1
8. Dell’Acqua, F., et al. (2026). Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality. Organization Science. https://pubsonline.informs.org/doi/10.1287/orsc.2025.21838
9. Vaccaro, M., Almaatouq, A., and Malone, T. (2024). When combinations of humans and AI are useful: A systematic review and meta-analysis. Nature Human Behaviour. https://doi.org/10.1038/s41562-024-02024-1. Open manuscript consulted: https://arxiv.org/html/2405.06087v1
10. Cemri, M., et al. (2025). Why Do Multi-Agent LLM Systems Fail? Open manuscript, version 1. https://arxiv.org/html/2503.13657v1
11. U.S. Courts. Chapter 11—Bankruptcy Basics. Consulted October 2026. https://www.uscourts.gov/court-programs/bankruptcy/bankruptcy-basics/chapter-11-bankruptcy-basics
12. Alvarez & Marsal (March 2026). The 13-week cash flow. https://www.alvarezandmarsal.com/thought-leadership/the-13-week-cash-flow
13. Office of the Comptroller of the Currency. Comptroller’s Handbook: Commercial Loans. Version modified March 2025; original narrative 1990. https://www.occ.treas.gov/publications-and-resources/publications/comptrollers-handbook/files/commercial-loans/pub-ch-commercial-loans.pdf
14. NIST (2023). Artificial Intelligence Risk Management Framework (AI RMF 1.0). NIST AI 100-1. https://nvlpubs.nist.gov/nistpubs/ai/nist.ai.100-1.pdf
15. NIST (2024). Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile. NIST AI 600-1. https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf
16. Weiner, B. J. (2009). A theory of organizational readiness for change. Implementation Science. https://link.springer.com/article/10.1186/1748-5908-4-67
17. Centers for Disease Control and Prevention (2018). Crisis and Emergency Risk Communication: Introduction. https://www.cdc.gov/cerc/media/pdfs/CERC_Introduction.pdf
18. Cohen, T. R., Leonardelli, G. J., and Thompson, L. (2014). Avoiding the Agreement Trap: Teams Facilitate Impasse in Negotiations with Negative Bargaining Zones. Negotiation and Conflict Management Research. https://ncmr.lps.library.cmu.edu/article/229/galley/232/download/
19. Federal Reserve Banks (March 2026). 2026 Report on Employer Firms: Findings from the 2025 Small Business Credit Survey. https://www.fedsmallbusiness.org/reports/survey/2026/2026-report-on-employer-firms
20. National Center for the Middle Market (2026). Middle Market Indicator, Mid-Year 2026. https://www.middlemarketcenter.org/wp-content/uploads/2026/08/NCMM_MMI_MID-YEAR_2026.pdf
21. U.S. Bureau of Labor Statistics. Economy at a Glance: Tampa, Orlando and Lakeland metropolitan areas. Consulted October 2026. https://www.bls.gov/eag/eag.fl_tampa_msa.htm; https://www.bls.gov/eag/eag.fl_orlando_msa.htm; https://www.bls.gov/eag/eag.fl_lakeland_msa.htm