Consarc

Consarc is a potential fit for finance teams that want a repeatable close task prepared continuously rather than rebuilt at month-end. Its public material describes Noa agents for reconciliations, accruals, prepaids, leases, borrowings, revenue validation, and intercompany work. Consarc also says accountants review decisions and its Forward Deployed Accountants configure and maintain the workflows. [S1]

The useful first question is narrower than “can it automate the close?”: can it prepare one known schedule from complete source data, under the buyer’s accounting policy, with a controller able to inspect the evidence and decide what posts? The public pages support specific preparation workflows, but they do not establish the buyer’s enabled integrations, accounting-policy configuration, identity permissions, write authority, error handling, security terms, or actual results. [S1][S2][S3][S4]

Best for: a mid-market or enterprise finance team with one recurring close task, stable data sources, documented policy, and a named controller who can run a controlled pilot.

Not for: a team seeking to remove accounting judgment or final-close responsibility; a company without reconciled source populations and account mappings; or a buyer unwilling to validate permissions, outputs, exceptions, and recovery before connecting a production ledger.

Prepare a recurring accrual schedule

• Trigger: A defined period cadence or a relevant source-data update.
• Inputs: Purchase orders, invoices, usage logs, spend data, approved timing and threshold rules, account mappings, and entity context.
• Output: A prepared accrual schedule, supporting documentation, and proposed journal-entry detail or an exception.
• Human checkpoint: A controller checks completeness, accounting policy, materiality, transaction support, proposed entry, and exception treatment before posting.
• Success measure: source-population reconciliation rate, policy-compliant schedule rate, material-error rate, exception age, reviewer time, and close-cycle impact.

Consarc says its Accruals Agent pulls updated source activity, applies buyer timing, threshold, and mapping rules, and prepares schedules and journal entries. It also describes transaction-level detail and examples such as missing-invoice, consumption, recurring-cost, multi-period, and multi-entity accruals. Those are vendor-described capabilities, not proof of a buyer’s accounting treatment. [S2]

Maintain prepaid amortization schedules

• Trigger: A new or changed invoice, contract, or period-end run.
• Inputs: Invoice dates, contract terms, prepaid classification policy, account mapping, prior schedule, and source documentation.
• Output: A prepared schedule showing prior amortization, current-period movement, remaining balance, and supporting details.
• Human checkpoint: An accountant confirms that the item is prepaid, verifies the service period and classification, reviews the roll-forward, and approves any downstream entry.
• Success measure: schedule-to-ledger tie-out rate, correct period allocation, carry-forward error rate, correction count, evidence completeness, and review time.

Consarc says the Prepaids Agent identifies prepaid items from invoices, creates schedules using invoice dates and contract terms, applies monthly amortization, and maintains roll-forwards. Its listed examples include insurance, maintenance, SaaS subscriptions, rent, multi-year payments, and retainers. [S3]

Prepare continuous reconciliation and variance review

• Trigger: A periodic close-control run or a detected source-data change.
• Inputs: Bank, AP, AR, intercompany, ledger, and transaction-detail data, plus buyer-defined materiality and reconciliation policy.
• Output: Prepared reconciliations, a variance explanation, evidence, and an exception list for the finance team.
• Human checkpoint: The finance owner verifies source completeness, cause attribution, accounting interpretation, materiality, and any proposed action.
• Success measure: reconciled-balance rate, false-exception rate, material issue detection, time to resolve, audit-support completeness, and manual rework.

Consarc describes continuous reconciliations, transaction-level variance explanations, and a supporting audit trail. It says these features surface items carrying judgment risk, but the public pages do not quantify accuracy on the buyer’s systems or show how a buyer configures thresholds and exceptions. [S1]

Pilot a controller-approved close step

• Trigger: A buyer selects one entity, schedule, and accounting owner for a limited pilot.
• Inputs: Historical close files, production-like source extracts, approved policy, target workflow boundary, and named reviewers.
• Output: A parallel prepared schedule and evidence pack compared with the manually completed close.
• Human checkpoint: The controller signs off on each comparison, policy decision, exception, correction, and any approval to expand scope.
• Success measure: agreement with reviewed baseline, source completeness, exception precision, reviewer effort, recovery time, audit-evidence usability, and cost per completed schedule.

Consarc’s CEO describes a multi-agent accrual flow in which agents pull trigger events and open purchase orders, follow up with owners, calculate against controller-approved logic, and draft the entry and support for review. Treat that account as a vendor description of the product model; validate it against the buyer’s own close. [S4]

The described model is: finance-source data enters a configured workflow → an agent prepares a schedule, reconciliation, explanation, or entry support from the buyer’s policies → the finance team reviews exceptions and decisions → the buyer determines whether and how a result affects the ledger. Consarc says embedded Forward Deployed Accountants configure and maintain the agents around the team’s ledger structure and policy. [S1][S4]

This is not evidence that every step is read-only, approval-gated, reversible, or safe for the buyer’s data. The public sources do not establish the exact data fields read, which systems receive writes, service-account permissions, tool or model behavior, segregation of duties, support access, error messages, rollback, retention, or export path. A demonstrated workflow needs to answer those questions before a close owner treats it as production-ready.

Verified facts from primary sources

• Consarc’s current website describes Noa as a collection of close-task agents, including accruals, prepaids, leases, borrowings and interest, revenue validation, and intercompany work. [S1]
• The Accruals Agent page says it uses purchase orders, invoices, usage logs, and spend data; applies timing, threshold, and mapping rules; and prepares schedules and journal entries. [S2]
• The Prepaids Agent page says it uses invoices, invoice dates, contract terms, and source data to build schedules and calculate amortization across periods. [S3]
• Consarc’s public site says Forward Deployed Accountants configure and maintain the automated close. [S1]

Vendor statements to validate

• Consarc says accountants review every decision, agents surface judgment-risk items, and workflows include audit trails and role-based access controls. Confirm the proposed workflow’s actual roles, data, approvals, logs, and audit export with a tenant-specific demonstration. [S1]
• Consarc’s CEO says multiple agents can sequence through a close task and that accruals, prepaids, variance analysis, and leases already operate that way. Confirm which agents and handoffs are available for the proposed entity, system, jurisdiction, and accounting policy. [S4]
• Consarc says its system can connect to ERP, CRM, billing, and HR systems including NetSuite, SAP, and Oracle. Confirm connector availability, authentication, refresh behavior, scope, write actions, retry behavior, and support responsibility. [S1]

B2Bagents assessment

The product looks most plausible as a focused close-preparation service, not a blanket replacement for the finance function. The combination of discrete agent pages and embedded accounting implementation gives a buyer a practical pilot shape: choose one schedule, map every input, compare the prepared result to a reviewed baseline, and retain sign-off with the controller. A broader continuous-close promise should wait until the buyer has repeatable evidence for source completeness, exceptions, permissions, corrections, and audit retrieval. [S1][S2][S3][S4]

Material unknowns

• Public pricing, usage unit, contract minimums, pilot terms, cancellation rights, implementation scope, and ongoing support commitments.
• Exact connector list, data mappings, system-of-record boundaries, service accounts, permissions, write actions, approval requirements, error handling, retries, rollback, and data-export path.
• Applicable DPA, subprocessors, model providers, training use, residency, retention, deletion, encryption, SSO, RBAC configuration, audit-log retention, security reports, business continuity, and incident commitments.
• Outcome quality, coverage, accuracy, close-time improvement, implementation duration, and operational cost for the buyer’s entity structure, ledger, source systems, and exception profile.
• Accounting treatment across jurisdictions, policy changes, late adjustments, unsupported documents, intercompany complexity, and controller workload during exceptions.

Start with one entity and one recurring schedule—an accrual or prepaid schedule is a natural candidate—rather than the full close. Name a controller as accountable owner, an accounting-policy owner, a technical integration owner, and a vendor counterpart. Run the workflow in parallel with the existing process for a defined number of close cycles.

Before the pilot, record the baseline source-population completeness, schedule preparation time, reviewer time, close days, material adjustment count, post-close correction count, exception age, error rate, audit-support retrieval time, and fully loaded cost. Include normal activity plus incomplete inputs, late invoices, missing contracts or purchase orders, conflicting terms, policy changes, wrong entity or account mapping, material variances, integration failures, retry behavior, and correction or rollback tests.

Set success thresholds before connection: full source-population reconciliation, no unexplained material difference from the controller-approved baseline, complete evidence for every output, a maximum exception age, bounded reviewer effort, demonstrated access revocation, and a tested correction path. Stop the pilot after an unexplained accounting treatment, missing source support, incorrect entity or account, unauthorized or opaque write, access-control concern, missing audit evidence, data-handling issue, or a failed integration that cannot be contained and recovered.

• Maxima: compare a finance-graph-based accounting automation product with Consarc’s implementation-supported close service; test source-data lineage, entry controls, reviewer workload, and audit evidence.
• Truewind: compare a managed accounting close model with Consarc’s agent-plus-embedded-accountant operating model; focus on owner responsibilities, data access, delivery scope, and close outputs.
• Rillet: compare a finance system-of-record approach with a layer designed to work with existing finance systems; validate integration boundary, accounting workflow, and migration requirements.
• Numeric: compare close-management and task-control tooling with Consarc’s claimed execution model; test the amount of work actually prepared, the audit trail, and the controller’s approval role.

1. Which exact Noa agents, integrations, models, and workflow versions will operate for the proposed schedule, and what data does each read or write?
2. Can the vendor show the buyer’s actual control path from source record to prepared result, exception, controller decision, ledger handoff, log, correction, and recovery?
3. Who is accountable for accounting policy, source-data completeness, mapping changes, late adjustments, exceptions, and final sign-off at each stage?
4. What roles, service accounts, authentication methods, approval gates, audit exports, retention rules, deletion procedures, and emergency-revocation controls apply to each connection?
5. Which security, privacy, DPA, subprocessor, model-provider, residency, certification, support-access, continuity, and incident terms apply to this buyer’s tenancy and data?
6. What is the commercial unit and total cost for implementation, embedded accountant time, connections, entities, volume, support, pilot exit, renewal, export, and transition?