Every AP team has the same inbox problem. Invoices arrive from dozens or hundreds of vendors, scattered across shared mailboxes, personal inboxes, and forwarded chains. Attachments get downloaded one at a time. PDFs get opened, scanned visually, and keyed into spreadsheets or ERPs by hand. The work is predictable, repetitive, and consuming hours that could go toward anything else. At 50 invoices a month, someone handles it between meetings. At 5,000, you have a team doing nothing but opening emails and typing numbers into fields.
Lido is the best option for AP teams that need to automatically extract invoice data from email without templates, model training, or per-vendor configuration. It handles any invoice format, including scanned and handwritten documents, and connects directly to email inboxes for fully automated intake.
Lido assigns each data extractor its own dedicated email address for auto-forwarding. You set up inbox rules that route vendor invoices by sender, document type, or subject line, and Lido automatically extracts data from the attachments into structured output. Esprigas, a gas distribution company processing 27,000 documents per month across 20,000 invoices, 2,000 supplier statements, and 5,000 customer POs, uses auto-forwarding rules to route documents from their inboxes directly to Lido extractors without manual intervention.
The problem with email-based invoice processing is that email was never designed to be a document management system. It's a communication channel that happens to carry attachments. Every invoice that arrives in your inbox requires someone to open the email, identify the attachment, download it, determine the correct process for that vendor, and manually enter the data into a spreadsheet or ERP. Multiply that by hundreds of vendors and thousands of invoices, and you have a team whose entire job is email triage.
The failure modes are specific and predictable. Invoices get buried in inboxes during high-volume periods. Forwarded chains strip attachments or bury them in nested replies. Multiple team members process from the same shared inbox and either duplicate work or skip documents assuming someone else handled them. Vendors send combined PDFs with invoices and statements in the same file, and someone has to split them manually before processing.
A gas distribution company with 80% of suppliers using the same two billing systems still spent, in their words, "a ton of time retraining models" in their extraction tool. New suppliers, combined PDFs, and format changes created constant maintenance. Even with auto-forwarding rules already routing documents to their extraction system, the extraction itself was the bottleneck. Their approval process existed solely because they couldn't trust the accuracy of the extracted data.
Most AP teams centralize vendor invoices into a shared inbox. It's a reasonable starting point. All invoices go to one place, and anyone on the team can process them. But shared inboxes create their own set of problems as volume grows.
There's no reliable way to track which invoices have been processed and which haven't without an external system. Two people opening the same email creates duplicate entries. One person skipping an email because they assume it's handled creates a gap. Month-end becomes a scramble to reconcile what was processed versus what's sitting unread in a mailbox with 4,000 messages.
The manual download step is where the most time gets consumed. Opening an email, saving the attachment to a local folder or shared drive, then uploading it to an extraction tool or keying it into a spreadsheet. For a trucking company like Disney Trucking, which had six full-time employees doing nothing but data entry, the cost of this manual pipeline was measured in headcount. Six people, full time, processing 360,000 pages a year. Their primary concern was "the manual entry risk of data error." They switched to Lido and replaced all six.
Not every invoice comes through email. Many teams store documents in shared folders on OneDrive, Google Drive, or SharePoint. Vendors upload to portals. Internal teams scan physical documents and save them to network drives. The result is the same problem with a different address: documents sitting in a folder, waiting for someone to open them and key the data in.
Lido connects to Google Drive and OneDrive and can automatically check for new files every five minutes. When a new document appears in a connected folder, Lido extracts data from it using the same extractor configuration as email-based documents. The output goes to a spreadsheet, CSV, or pushes to downstream systems via API. Disney Trucking's workflow starts with scanning physical driver tickets to a OneDrive folder. Lido monitors that folder, extracts the data automatically, and exports the structured output back to OneDrive for their accounting team.
For teams that receive invoices through a mix of email attachments, cloud drive uploads, and manual scans, Lido handles all three intake paths through the same extractor. One set of extraction rules covers the same document type regardless of how it arrived.
The most effective way to automate email-based invoice intake is auto-forwarding. You create rules in your email client, Outlook, Gmail, or whatever you use, that automatically forward invoices to a dedicated extraction address based on sender, subject line, or other criteria.
Each data extractor in Lido has its own unique email address. You set up your inbox routing rules to forward vendor emails to the correct extractor address, and Lido processes the attachments automatically. You can configure it to extract from attachments only, email body only, or both.
Esprigas had this workflow already built out before finding Lido. They were using auto-forwarding rules to route invoices from different suppliers to different extraction models in their previous tool. The routing worked. The extraction didn't. Their nested tables failed. Handwritten propane invoices were unprocessable. Combined PDFs with invoices and statements in the same file required manual splitting. The forwarding rules stayed the same when they moved to Lido. Only the extraction endpoint changed.
The routing logic can be as simple or granular as your vendor landscape requires. Forward everything from a specific sender domain to one extractor. Route by subject line keywords to separate extractors for invoices versus purchase orders versus statements. Esprigas's operations lead mapped it out on the second call: receive the document, determine which rule set it belongs to, forward to that extractor's email address. When rules conflict across suppliers, you build conditional logic directly in plain-language instructions. "If supplier is Linde, use this pricing logic. If supplier is Airgas, use that pricing logic." All in one extractor, no code required.
The most common end state for extracted invoice data is a spreadsheet. Whether it's Excel, Google Sheets, or CSV, AP teams want structured rows and columns they can review, sort, filter, and import into their accounting system. The gap between an invoice PDF in an email attachment and a clean row in a spreadsheet is where all the manual labor lives.
Lido extracts directly into a spreadsheet format. Each extractor outputs structured data as rows in a table with columns you define: invoice number, vendor name, date, line items, amounts, PO number, whatever your downstream system needs. You can export to CSV or Excel on demand, or automate the export on a schedule. New data every five minutes, once a day, or on whatever cadence fits your process.
For teams that use Excel as the intermediary between extraction and their accounting system, the workflow becomes: invoices arrive by email, auto-forward to Lido, data extracts automatically, structured output exports to a shared drive as a CSV or Excel file. No one opens a PDF. No one types a number. Viking Trans, a trucking company processing 50 to 60 rate confirmations daily, was doing this manually. One person spending her time on data entry. As their operations lead put it, "it's stupid to waste her time" on manual processes when the work follows clear rules.
Extraction is only the first step. The structured data needs to reach your accounting system, ERP, or AP platform to become actionable. The question most teams ask is how the data gets from the extraction tool into the system where bills are paid.
Lido pushes structured data to downstream systems through API integrations, direct file exports to cloud drives, or automated email delivery. For teams using ERPs like Microsoft Business Central, Lido can push extracted data via API endpoints. Esprigas was evaluating exactly this path: replacing their current workflow of exporting XML from their extraction tool, routing through an AS2 connection, processing through EDI software, and handling manual approval gates. Lido would push directly to Business Central via API, one endpoint per document type.
Lido doesn't replace your accounting system or your AP approval workflow. It solves the extraction bottleneck that sits upstream of everything else. Once data is accurately extracted and structured, syncing with your accounting platform, whether through API, CSV import, or automated file delivery, becomes a straightforward integration rather than a manual re-keying exercise.
The end-to-end flow from email inbox to AP ledger has three phases: intake, extraction, and delivery. Most AP teams are manual on all three. Automating one without the others still leaves bottlenecks, but extraction is the phase where the most time and error concentrates.
The automated version looks like this. Vendor emails arrive in your shared inbox. Auto-forwarding rules route them by sender or type to the correct Lido extractor address. Lido processes the attachments and extracts structured data. Required fields, invoice number, vendor, amount, line items, PO reference, are captured without manual entry. If all required fields are present and accurate, the data exports automatically to your accounting system via API or file delivery. If something is missing, it routes to manual review.
Esprigas had built this exact approval logic in their previous tool. Documents with all required fields, invoice number, PO number, item, bill amount, supplier account number, auto-approved and exported. Documents missing a required field went to manual review. The problem was that their extraction tool's accuracy wasn't high enough to let the auto-approval threshold work. As their operations lead said:
"The approval is all about the accurate extraction of the data. It has nothing to do with the content."
They review every single extraction manually, not because of business rules, but because they can't trust the data. Accurate extraction is what unlocks the rest of the pipeline. Without it, every downstream workflow, approval routing, AP ledger posting, payment scheduling, requires a human to verify what should have been automated.
Speed in invoice processing isn't about how fast the tool runs. It's about how many manual steps you can remove between the email arriving and the bill being approved. Every manual step, downloading the attachment, opening the PDF, keying data, checking for errors, routing for approval, adds minutes per invoice that compound at volume.
The fastest path removes all of them except exception review. Email arrives, forwards automatically, data extracts automatically, required field validation happens automatically, and the bill routes to approval with structured data already attached. At Soldier Field, this reduced processing from 20 hours of manual work per week to 30 seconds per invoice after switching to Lido.
Lido doesn't handle the approval or payment steps. It handles the extraction that makes fast approval possible. When the data entering your approval workflow is accurate and complete, approval becomes a confirmation step rather than an error-catching exercise. The time from invoice email to approved bill drops from days to minutes, not because approval got faster, but because extraction stopped being manual.
At high volume, the fundamental challenge of emailed invoices isn't reading them. It's the combinatorial explosion of formats, senders, document types, and exceptions that overwhelm any manual process. A team processing 500 invoices a month from 30 vendors can develop institutional knowledge about each vendor's format. A team processing 20,000 invoices from hundreds of vendors cannot.
Template-based extraction tools attempt to solve this by building a parsing configuration for each vendor format. Model-trained tools attempt to solve it by feeding sample documents for each format and training the system to recognize them. Both approaches create maintenance overhead that scales with vendor count. Esprigas built two separate models in their previous tool, one with 50 pages of intentional training, one without. They still spent "a ton of time retraining the models" every time a supplier changed formats or a new supplier came onboard.
Lido takes a different approach. There are no templates to build per vendor and no models to train per format. You describe what to extract in plain language, and the tool handles format variance automatically. When Esprigas's propane suppliers send handwritten invoices, the same extractor that handles their digital invoices processes those too. When a new supplier comes onboard with a format Lido has never seen, nothing breaks and no one reconfigures anything.
For AP teams managing thousands of emailed invoices monthly, the question isn't whether AI can help. It's whether the AI approach creates its own maintenance burden. Template-per-vendor AI does. Model-trained AI does. Layout-agnostic AI extraction, where the tool understands any document without prior exposure, does not.
Many organizations receive invoices across multiple email accounts. Different divisions, locations, or departments each have their own AP inbox. Regional offices receive vendor invoices locally. Subsidiary companies maintain separate email domains. The result is invoice data fragmented across five, ten, or twenty email accounts with no unified view.
Centralization doesn't require consolidating all those inboxes into one. It requires routing all of them to the same extraction system. Each email account sets up auto-forwarding rules that route invoices to the appropriate Lido extractor. Multiple email accounts can forward to the same extractor if the document types are the same, or to different extractors if the extraction rules differ. The output from all extractors feeds into the same structured format.
Disney Trucking operates from two locations. Viking Trans has 70 to 80 trucks generating documents across their operations. Esprigas has onshore and offshore teams processing documents across their supplier network. In each case, the documents originate from different sources, but the extracted data needs to end up in one system in one format. The extraction layer is what makes centralization possible without changing how vendors send invoices or how different parts of the organization receive them.
The full cycle from email to payment touches multiple systems: email for intake, extraction for data capture, accounting for AP ledger entry, approval workflows for authorization, and payment platforms for disbursement. No single tool handles all of it, and any vendor claiming otherwise is oversimplifying a process that varies by organization.
What Lido automates is the extraction step, which is the manual bottleneck that slows down everything downstream. When extraction is manual, the entire pipeline moves at the speed of data entry. When extraction is automated and accurate, the pipeline moves at the speed of your approval and payment processes, which are typically much faster.
The practical architecture looks like this. Email auto-forwarding handles intake. Lido handles extraction. API integrations or file exports deliver structured data to your ERP or accounting system. Your existing approval workflow handles authorization. Your payment platform handles disbursement. Each layer does what it's designed for. The extraction layer, which is where AP teams spend 80% of their time, is the one that benefits most from automation.
Lido uses a custom blend of AI vision models, OCR, and LLMs to extract data from any invoice format without templates or model training. Each data extractor gets its own dedicated email address for auto-forwarding, and can also connect to Google Drive, OneDrive, or accept direct uploads.
Esprigas processes 27,000 documents per month and routes them via auto-forwarding rules to Lido extractors. Disney Trucking replaced 6 full-time data entry employees processing 360,000 pages per year. Soldier Field went from 20 hours of manual processing per week to 30 seconds per invoice.
If your AP team is spending more time opening emails and typing numbers than reviewing data and managing exceptions, the intake path isn't the problem. The extraction step is. Automate that, and the rest of the pipeline accelerates.
%20(1).svg)
