Limited published evidence · Startup finance, customer success and HR
AI startup finance model dashboards: Alex Lieberman example
In a post on X, Alex Lieberman recounts an unnamed Series B startup finance leader using Claude and Claude Code to prepare a fundraising model, customer dashboards, an editable QBR deck and compensation scenarios. This is a secondhand account. Artifacts and results were not independently inspected, and it does not establish autonomous finance decisions.
Prepare a financial model, customer dashboards, QBR materials and compensation scenarios
Unnamed startup (reported by Alex Lieberman)
Secondhand account of an unnamed Series B startup finance leader's work. Artifacts, integrations and results were not independently inspected.
A published third-party example, not a Nuvepro customer case or evidence of a partnership.
The workflow at a glance
Finance models, account materials and compensation scenarios
Illustrative summary of the published account
Reported work components. Their arrangement does not imply a fixed execution order or simultaneous work.
- Financial model
ARR waterfall and expense budget
Alex recounts an unnamed finance leader using company context and iterative questions to prepare fundraising materials.
- Customer dashboard
Metrics and intervention alerts
The secondhand report describes account dashboards with internal customer-impact alerts.
- Customer materials
Editable QBR from metrics and charts
The account describes a branded deck prepared from customer metrics and charts.
- Compensation scenarios
Sheets macros, Rippling data and manager materials
The account links financial-model context to compensation scenarios and leveling-band flags.
What the source reports
Alex Lieberman recounts a finance leader using company context and iterative questions to prepare a fundraising model, including an ARR waterfall and expense budget. He also describes customer dashboards with internal intervention alerts, an editable QBR deck built from metrics and charts, and compensation scenarios using Claude Code-generated Google Sheets macros, Rippling data and a financial model. The compensation example includes manager materials and leveling-band flags. The company and operator are unnamed, and the account does not establish autonomous finance or compensation decisions.
Source accounts: @businessbarista
Reported technology: Claude and Claude Code; spreadsheet and business tools described in the X post. Scope and outcomes are source-reported.
What the enterprise environment would need
The following is a proposed project design for an enterprise doing similar work. Published details are marked; the remaining systems, data and scenarios would be agreed for your project.
Proposed practice environment
Check linked finance, QBR and compensation drafts
Illustrative project design to scope with your enterprise.
Checks designed, not executed. Domain owners must agree the reference results, permissions and acceptance tests before implementation.
- 1 · Set up the environment
Fictional company, customer and employee data
Version assumptions, customer metrics, leveling bands and approved spreadsheet/deck tools.
- 2 · Finance practitioner and agent
Prepare models, account materials and scenarios
A finance practitioner and agent prepare a model, one dashboard/QBR and compensation scenarios, keeping customer and manager views separate.
- 3 · Independent checks
Independent finance and access checks
Recalculate models; reconcile charts to customer/period; test manager and customer isolation.
- 4 · People decide acceptance
Finance, account and HR owners review
Finance, account and HR owners accept their respective practice evidence and hold unresolved assumptions; acceptance grants no live compensation or external-sharing authority.
A reviewable evidence package
A checked operating model, one customer dashboard/QBR and compensation scenarios, with assumptions and separate owner decisions.
If a check fails → withhold acceptance.
A QBR uses another customer’s data and a manager export exposes other staff. Block sharing and repair both outputs.
Failure or owner rejection returns to review
- Hold
- → 2 · Correct models and isolate account data
- → 3 · Rerun independent checks
- → 4 · Owner review
Operating boundary. No private employee data, compensation changes or external commitments without authorized review.
- Systems and tools
- Source-described: Claude, Claude Code, Google Sheets Apps Script and Rippling data. Proposed: use approved spreadsheet/deck tools and a synthetic HR export or test API, not private employee records.
- Data and state
- Source-described: fundraising context, ARR and expense assumptions, customer metrics and leveling bands. Proposed: provide linked fictional company, account and employee records with versioned assumptions.
- People and handoffs
- Proposed: finance reviews model assumptions, customer success interprets alerts and QBRs, and authorized HR/finance owners decide compensation. Do not infer these approval controls were demonstrated by the post.
- Rules and authority
- Proposed: separate internal and customer-facing records, restrict manager folders to the right staff and require human approval before sharing investor materials or changing compensation.
- Normal and exception cases
- Source-described: compensation outside leveling bands is flagged. Proposed: add stale customer metrics, conflicting model assumptions, formula errors and cross-manager data exposure.
- Reset and replay
- Proposed: restore a fixed spreadsheet, customer dataset and fictional employee state; rerun the same scenarios and compare exported charts, slides and access assignments.
A proposed simulation
An illustrative practice design derived from this work, not an executed assessment.
Proposed: prepare a linked fictional operating model, one customer QBR and compensation scenarios, retaining assumptions and separate review decisions.
Review gate: Proposed: finance, account and HR owners verify their outputs, access boundaries and unresolved assumptions before approving any intended business use.
Challenge the result: Proposed: a QBR uses another customer's data while a compensation export includes another manager's staff. Both outputs must be blocked and corrected before sharing.
What needs checking
- Proposed: independently recalculate the ARR waterfall, expense model and compensation scenarios rather than grading presentation alone.
- Proposed: reconcile every QBR metric and chart to the correct customer and period; verify alert thresholds and unanswered evidence gaps.
- Proposed: test manager/customer access boundaries and verify that drafts cannot become compensation changes or external commitments without authorized review.
Review environment fidelity, human capability and agent performance separately. The enterprise’s accountable owners decide acceptance of the shared workflow and any operating use.
Questions about this example
- What does the finance leader use AI for?
- As Alex Lieberman recounts, the leader uses company context and iterative questions to prepare a fundraising model with an ARR waterfall and expense budget, customer dashboards with intervention alerts, an editable QBR deck, and compensation scenarios using Claude Code-generated Google Sheets macros, Rippling data and a financial model.
- Who checks the AI's finance and compensation work?
- Alex Lieberman's account does not establish autonomous finance or compensation decisions, and it does not demonstrate approval controls. He says compensation outside leveling bands is flagged. Nuvepro proposes that finance reviews assumptions and authorized HR and finance owners decide compensation, as a suggested check.
- How strong is the evidence for this example?
- Early-stage: Alex Lieberman's secondhand account of an unnamed operator, not a firsthand audited implementation. Dashboards, models, macros, HR integrations and permissions were not inspected, and the stated build-time comparison has no independently verified baseline. It is not evidence of production readiness.
- What would an enterprise need to build something similar?
- These are Nuvepro's proposed checks, not Alex Lieberman's. Use a synthetic HR export or test API and linked fictional company, account and employee records with versioned assumptions. Independently recalculate the ARR waterfall and compensation scenarios, reconcile QBR metrics, and test manager and customer access boundaries.
Shape this around your enterprise
Bring a similar workflow and the people responsible for it. Together, define the records and systems to use, the result your team will build or configure, and how its owners will check it. Nuvepro scopes the simulation, environment and preparation. Agree any existing-agent evaluation, implementation assistance and production responsibilities separately.
Sources and limitations
Curation review: . This is the date this source example was reviewed, separate from source publication dates.
- This is Alex's report of a conversation with an unnamed operator, not a firsthand audited implementation. The stated build-time comparison has no independently verified baseline.
- Actual dashboards, financial models, generated macros, HR integrations and permissions were not inspected. Human approval and exception tests above are proposed.
- The proposed environment, practice project and checks are designs for discussion, not executed Nuvepro work or evidence of production readiness.
- Finance leader show-and-tell: operating model, customer dashboards, QBRs and compensationSource publication: 2026-02-11
More finance and accounting examples
- AI month-end accruals: Campfire exampleCampfire
- AI purchase request review: Anthropic exampleAnthropic
- AI AP exception resolution: Varick pattern by Eyad KhraisVarick AP pattern (customer unnamed)
- AI billing discrepancy review: Telstra exampleTelstra