Blog
Functional AISolutions Blog
AI articles, product news and newsletter. Everything you need to know to implement AI in your business.
Articles
AI agentsRestart an interrupted AI workflow without repeating completed business actionsRecover interrupted ERP-connected workflows with confirmed results, protected retries and clear handoffs for uncertain actions—without reconstructing the entire run.
AI technologyTest reasoning settings to find a lower-cost way to run your existing AI automationCompare reasoning settings inside one existing workflow. Adopt a lower setting only when business-quality limits hold and the cost per accepted outcome improves—including retries, review and repairs.
AI TechnologyHow to test database performance for responsive reports and AI agent queriesTest reports and AI agent queries together, then use responsiveness, correctness, freshness, ERP impact and total cost to choose the infrastructure changes your business needs.
Workflow automationTurn factory vulnerability findings into maintenance requests your team can assess and scheduleConnect security evidence to equipment, production dependencies and maintenance constraints—then let qualified staff decide the work and approve the window.
Ontology & Business DataHow to preserve AI purchasing decision records for easier business reviewConnect purchasing evidence, approvals and ERP outcomes in a reusable decision record, making later business reviews easier without retaining every AI conversation.
Ontology & Business DataWhich purchase orders need an early supplier check? Test predictions on connected purchasing recordsConnect purchasing history, reconstruct what buyers knew at the time, and test whether a ranked queue finds more late deliveries within the same daily review capacity.
Ontology & Business DataFind reusable engineering designs with an ontology connecting requirements and evidence.Make existing designs easier to assess for reuse. Start with one product family, connect requirements to revision-specific evidence, and judge the pilot on search effort and engineer-confirmed relevance.
Workflow automationMatch Near-Expiry Ingredients to Suitable Production Orders—with Less Manual CheckingStart with one short-life ingredient: connect stock, quality records and formula versions, then give planners a source-linked review queue rather than another spreadsheet to reconcile.
Ontology & Business DataMake experienced colleagues easier to find with evidence from completed workConnect colleagues to recorded project responsibilities and service history so your team can find relevant experience—and inspect why a contact was suggested.
Ontology & business dataHow to automate quoting under framework agreements with less manual searchingConnect framework agreements to product and price records to prepare quote drafts with less searching, while keeping commercial exceptions with the responsible people.
Workflow automationMake the next shift’s handover from the work record—not another round of updatesYour teams already log jobs, incidents and temporary fixes. Turn those records into a source-linked handover draft that supervisors approve, so nobody rewrites the shift from memory.
Workflow automationGet contractors ready for their first shift: let AI assemble the onboarding worklistConnect each contractor assignment to its prerequisites, evidence and responsible people—so your team can resolve outstanding work before the first shift while retaining approval authority.
Workflow automation · Month-end closeReconcile subledgers to the general ledger more often, so month-end starts from exceptions your accountant has already reviewedFive operating rules let you reconcile daily on the ERP you already run, provided its ledgers can be queried. An agent proposes each fix, and your accountant decides.
Ontology & Business DataMake past quality investigations easier to find: test photo search with batch contextLink every defect photo to its product, revision, batch and process, and your quality team can reach the right past investigation faster. Here is how to test it on your own archive.
Ontology & Business DataA customer asks for your product’s carbon footprint. Build an answer you can reuse.Connect one product family’s material revisions, supplier evidence and approved emissions-factor mappings, so the next customer carbon request starts with reviewed records instead of an inbox search.
AI agents“Which of our customer contracts…?” Ask for an AI answer that shows it checked every oneA confident list is not a complete answer. Turn one contract portfolio into structured records, and your agent can return a tally you can check: what it read, what it couldn't read and what a person still has to decide.
Workflow automationKeep document automation useful when suppliers change their formsRoute every supplier certificate version to its own tested extractor so a revised form does not break the work that already runs, and quality release stays with your team.
Ontology & business dataA supplier changes the delivery date. Which customer promises need attention?Trace a supplier delay to the customer orders it touches, and check that orders marked “unaffected” are really covered, starting with one product family.
AI agentsOpenClaw vs branded AI agents: why your business should test the free option firstOpenClaw showed this year that AI can act, not just chat, and its creator kept it free. Before you pay for a polished, heavily marketed agent, test it against the free one on your own work.
Ontology & Business DataYour first factory ontology can map your data without moving itModel one production line's products, recipes, equipment and data sources by reference. Your team gets a shared map for tracing a batch across the ERP, MES and historian, and your data stays in the systems that already hold it.
AI technology · Field guideRAG or a business ontology? Count the lookups before you chooseA deployed aerospace case estimated the effort of RAG and a mapped ontology per component. Apply the same arithmetic to your own repeated cross-system check before you buy anything.
Ontology & business dataWhen a machine stops, can your newest technician find what your best one knows?Rockwell's Singapore plant built its maintenance copilot on fault knowledge from veteran engineers, organised by symptom, cause and fix. You can start the same foundation on one machine family this quarter.
Workflow automationCapture order-intake exceptions as business rules so AI agents move routine orders and route the rest to the right ownerHow to capture order-intake exceptions as business rules before AI agents take over routine orders: start with the colleague who handles them today, so each unusual case reaches the person who owns it.
Ontology & business dataBusiness ontology: faster decisions and less re-keying across ERP, CRM and AI agentsMap your customers, products, orders and permitted actions once. The automations and agents working across your ERP, CRM and spreadsheets can then reuse that map instead of each working out what your business means.
OperationsBefore expanding an AI prototype, connect one workflow to real business dataA prototype running on mocked data shows that an idea can be built. Before you fund wider automation, make one workflow pass against your real records, permissions and exceptions.
OperationsBefore AI suggests a replacement product, define what counts as equivalentA stockout shortlist should connect approved equivalence to the actual customer order and fulfillable stock. Start with one product family—and make “cannot verify” an explicit outcome.
OperationsWhen your CRM and ERP disagree, which fact should AI use?Define authority for the delivery decision—not a universal winner between systems. A bounded address pilot can make conflicts visible while keeping changes and customer commitments under human approval.
Procurement dataOne supplier, three names: fix the spend picture before AI recommends savingsTest one quarter of purchasing records: reconcile supplier identities and categories, preserve legal-entity distinctions, and keep every amount traceable before asking AI where to save.
OPERATIONSMonth-end close: let an agent chase the evidence, not approve the journalRivian’s accrual workflow suggests a practical starting point: automate the collection and organisation of supporting evidence, then test whether finance receives a better preparation pack—not merely a faster draft.
Operating existing agentsYour AI agent is online. Is the work getting done?Keep system health and quality scores. Add a completion record that checks the assigned outcome, makes unfinished work visible and names who owns recovery.
Enterprise accessGive your AI connectors an offboarding testCentralized authorization can simplify connector onboarding. Before relying on it, test whether changed access rules hold across fresh logins, existing sessions and issued tokens.