Phacet

What Phacet does

Phacet builds AI agents for finance teams, offering over 40 pre-built agents that structure, reconcile, and control finance data with full auditability. The platform allows users to build, structure, match, and analyze financial data, applying business rules to 100% of flows. It extracts data from various documents, performs semantic matching with confidence scores, and generates reports and dashboards. Phacet integrates with existing ERPs, banks, and other systems, providing a no-code solution for finance automation. The company emphasizes security with EU-hosted data, ISO 27001 certification, and GDPR compliance, ensuring data privacy and never using client data for training.

Products

• AI MatchSemantic reconciliation with exposed reasoning at every line, combining exact and semantic matching with confidence scores and plain-language justifications. Used for 3-way matching, bank reconciliation, account matching, and inter-system reconciliation.
• AI SourceShows exactly which parts of documents were used for data extraction, highlighting relevant passages directly on PDFs for verification and debugging.
• Agent co-builderAllows users to describe their process, and the agent builds it with them, asking questions, configuring, and iterating to create automated workflows and configured Tables.
• Full audit trailEvery action traced, timestamped, and consultable, providing a complete history of operations and decisions.
• AI TablesStructured datasets with rows and columns, where each column can run AI, code, or web-based transformations. Functions as a database with configurable AI columns for LLM extraction, categorization, standardization, and web enrichment.
• ProjectsContainers that group AI Tables for the same business workflow or operational domain, bundling everything needed to run a process end-to-end, including Tables, Workflows, Conversations, and Data sources.
• WorkflowsProcedures that agents execute, including extraction, transformation, reconciliation, control, and alerting, blending LLM and deterministic rules.
• Phacet AI No-Code ExtensionAdvanced AI for no-code workflows, specializing in data extraction, matching, and validation, integrating with tools like Zapier, Make, and Power Automate.

Key capabilities

• 40+ pre-built AI agents
• Full auditability at every step
• Semantic reconciliation with exposed reasoning
• AI Source for data origin traceability
• Agent co-builder for workflow creation
• Full audit trail for every action
• Connects to Sage, Pennylane, NetSuite, Xero, QuickBooks, banks, emails
• EU-hosted data
• ISO 27001 certified
• GDPR compliant
• Data never used for training
• No-code platform
• AI-powered data extraction from various document types (PDFs, spreadsheets, emails)
• Configurable AI columns for categorization, standardization, web enrichment
• Exact and semantic matching with confidence scores
• Anomaly reports, monthly summaries, dashboards
• On-demand dashboards, PDF and CSV exports, proactive alerts
• Predictable credit-based pricing
• Transparent pricing with zero margin on AI costs
• Real-time usage tracking
• Roles & permissions for team access (Studio Max plan)
• Multi-entity and multi-site management (Studio Max plan)
• Advanced multi-ERP integrations (Custom plan)
• Custom connectors and custom-built agents (Custom plan)
• Continuous learning for improved accuracy
• Human validation and supervision for results
• Plug & Play integration with no-code tools
• Data encryption in transit and at rest
• Role-based access controls

Use cases

• Automated invoice intake and cost controls
• Automated document processing and compliance
• Automating financial and catalog operations
• Automating accounts payable (AP)
• Automating procurement processes
• Automating sales administration
• Automating treasury operations
• Rebuilding revenue from invoices for board reporting
• Reconciling CRM, billing, and accounting systems
• Monitoring card expenses and detecting duplicate subscriptions
• Checking ERP / CRM / Billing consistency
• Calculating MRR/ARR and decomposing revenue movements
• Identifying forgotten subscriptions and unused licenses
• Reconciling bank transactions and detecting unmatched flows
• Extracting and controlling supplier invoices
• Comparing invoiced price vs. negotiated catalog price
• Matching PO, delivery, and invoice (3-way match)
• Controlling supplier price lists (food mercuriale)
• Controlling POS cash across all sites
• Automating accounting reconciliation (lettrage)
• Categorizing and standardizing accounting entries
• Extracting and structuring contracts or leases
• Cleaning and verifying supplier databases
• Automatically controlling expense reports
• Managing shared accounting email inboxes
• Rebuilding revenue from invoices
• Tracking budget vs. actual variances
• Automating document workflows and replacing manual spreadsheets
• Generating and categorizing product catalogs
• Routing and tagging emails in a shared inbox
• Verifying supplier invoices against negotiated prices
• Cross-checking order confirmations before shipment
• Reconciling delivery notes line by line
• Reconciling payments from stores, e-commerce, and marketplaces
• Detecting supplier overcharges and OTA commission errors
• Automating financial controls for high-volume purchasing and complex supply chains
• Automating financial controls for fast-growing companies with complex subscription stacks
• Automating financial controls for multi-location restaurants and hospitality groups
• Automating financial controls for food & beverage groups

AI approach

Phacet builds AI agents for finance teams that structure, reconcile, and control finance data. Their platform uses AI for data extraction from various document types (invoices, contracts, delivery notes), semantic matching with confidence scores, categorization, standardization, and web enrichment. They emphasize full auditability with AI Source showing data origins and a full audit trail of actions. They also offer an agent co-builder where users describe processes and the AI configures workflows. Phacet states that customer data is never used for training their AI models and that AI costs are included at cost, implying they leverage external models but build proprietary applications and workflows on top.

Tech named: LLM, AI models, vectorization

Industries served

• Artificial Intelligence
• Finance
• Food & Beverage
• Hospitality
• Retail
• BTP
• Tech
• Service
• Accounting firms
• Manufacturing
• Procurement
• Technology companies
• Distribution

What it says sets it apart

• 40+ pre-built AI agents for finance teams
• Full auditability and transparency with AI Source and audit trail
• No-code agent co-builder for custom workflows
• Semantic matching with confidence scores and reasoning
• EU-hosted, ISO 27001, and GDPR compliant with data privacy guarantee (no data used for training)
• Integrates with existing ERPs, banks, and systems without migration
• Predictable, action-based credit pricing with no token-level billing
• Focus on finance-specific AI agents and controls
• Rapid deployment and time-to-value (first agent in under two weeks)
• Handles complex data types (PDF, CSV, advanced matching) beyond simple structured fields
• Continuous learning and human validation for AI results
• Scales with business growth without proportional headcount increases in finance
• Contextual understanding of business rules, negotiated rates, and approval thresholds