What's new in ResAI

Processing Pipeline

This is the centerpiece of v1.0.2. Version 1.0 ran each file type down its own fixed path and stopped at extracted fields. The new pipeline runs one intelligent flow that tracks every document through a visible lifecycle and turns supported inputs into searchable, interconnected knowledge, while still producing everything 1.0 did.

For every supported file type, the new pipeline adds:

• A per-document lifecycle. You can see exactly which stage each file is at, rather than a single status for the whole batch.
• Resumability. If a file fails partway through, it restarts from the stage that failed, not from the beginning.
• A review queue. Files that can't be processed automatically are routed to a queue for a person to look at, instead of failing silently.
• Semantic search. Find content by meaning, not just exact keywords.
• A knowledge graph. Entities like people, assets, and events are linked across documents, so the system understands how information in one file relates to another.

The pipeline stays backward compatible. It emits the same extracted-field output as 1.0, and a single flag routes uploads to the old or new pipeline, which lets us roll it out gradually and revert instantly if needed.

Audio Processing

ResAI now accepts audio as a first-class input. Upload an MP3 or WAV file and ResAI transcribes it, plans the work from the transcript, chunks it, and enriches it with entities, the same way it treats any other document. A quality gate based on transcription accuracy diverts poor transcriptions to the review queue so a person can check them before they're used.

Because the content of an audio file can't be known until it's been heard, audio takes a slightly different route through the pipeline: it's transcribed first, then the processing plan is built from what was actually said. Once processed, audio becomes searchable, connected knowledge like everything else, which opens up use cases such as quality assurance on recorded calls and reasoning recorded content against your Standard Operating Procedures.

Supported audio formats today are MP3 and WAV.

Data Sources

Datasets remain what makes ResAI different from a generic AI assistant: answers are grounded in your curated data, and you can trust where they came from. This release deepens that in a few ways.

Beyond extracting fields, supported files are now chunked and embedded for semantic search and linked into the knowledge graph, so a dataset is no longer just a set of values to query but a connected body of knowledge ResAI can reason across.

Human in the Loop

Where a decision carries downstream consequences, ResAI asks for confirmation before it proceeds rather than acting on an assumption. This keeps a person in control of the choices that shape later results.

Structured data is the clearest example. When a dataset arrives without field definitions, ResAI proposes a name for each column based on what it sees in the data and asks you to confirm or correct it before the dataset goes live, because an incorrect definition would affect every downstream answer. The same checkpoint applies across the pipeline, so automated steps do not commit to a low-confidence result on their own.

Automation

Automation expands in this release so that more of the work between your data and the people who need answers happens on its own.

Scheduled ingestion now includes a SharePoint connector alongside Azure Blob Storage. Point ResAI at a SharePoint library and it checks on the schedule you set, pulls in new files, runs them through the pipeline, and makes them available to Ask ResAI and reporting, with no manual uploads. If your team drops a new batch of files into SharePoint on a cadence, your dataset reflects it automatically.

Reports continue to run a saved query on a schedule and deliver the output. In this release, scheduled runs generate the report and send it by email automatically, so the right people get the right answer in their inbox without anyone running anything by hand. Additional output destinations, including SharePoint and Power BI, are planned for an upcoming release.

Ask ResAI

A set of improvements makes everyday work in Ask ResAI smoother, especially around datasets.

• You can now attach datasets directly within the Ask ResAI flow.
• Dataset statuses update on their own. Previously they didn't refresh until you reloaded, so you couldn't see where processing stood.
• Long tables now return in a preview window with virtualized scrolling, so large result sets stay responsive.
• You can export the complete result set to CSV, not just the rows currently on screen.

Authentication and Onboarding

Microsoft Entra remains our primary identity provider. The onboarding flow is now a clear path from trial to a fully managed tenant. Trial users can sign in with a one-time password and start working immediately, without waiting on an app registration. When an organization is ready, an initial user registers their tenant details to connect the organization's own Entra ID, after which the client adds and manages its own users through their existing Microsoft sign-in. This lets your operational team get value first and bring IT into the conversation once it's obvious.

Audit Logs and Settings

Audit logging and role-based permissions from 1.0 carry forward, now covering the new pipeline and automation activity. Every login, dataset change, ingestion run, Ask ResAI query, and automation run continues to be captured with the actor, action, timestamp, and outcome, and Platform Owners continue to manage role-based permissions across the organization from Settings.

Authentication

Most of our customers run on Microsoft, so Microsoft Entra is our primary identity provider. Once your Tenant ID, Client ID, and Client Secret are configured in Settings, your team signs in with the same credentials they already use for everything else. No new passwords to manage, no separate user list to keep in sync, and access follows whatever rules your IT team has already established in Entra.

We know that getting IT to register a new app during a trial is often where momentum dies, so we've added one time password access for trial users. Platform admins can onboard trial users directly and grant them 15 or 30 days of access without waiting on a tenant registration. Your operational team can start working with ResAI immediately and bring IT into the conversation once the value is obvious.

Data Sources

Creating and managing datasets is what makes ResAI different from a generic AI assistant. Instead of pasting documents into a chat window and hoping for the best, you build a curated dataset that ResAI works against, which means answers are grounded in your actual data and you can trust where they came from.

In this release you can upload PDFs, CSVs, and Excel files. Group related files together as a single dataset (for example, a year of monthly invoices, or every SOP for a given operation) or upload them individually. Datasets become available to Ask ResAI as soon as processing completes, and you can rename or delete them at any time as your needs evolve.

For structured data like CSVs and Excel files, we've added human in the loop validation. When field definitions aren't supplied, ResAI proposes them based on what it sees in the data and asks you to confirm or correct before the dataset goes live. You stay in control of how your data is interpreted, and ResAI gets smarter about your domain with every dataset you onboard.

Insights

Insights turns Ask ResAI into a dashboard. Anything you can ask in natural language, you can pin as a widget on a board and watch it update over time.

Promptable widgets let you build a chart by describing what you want to see, then choose how to visualize it. Pick from foundational chart types (bar, line, scatter), common variations (stacked, grouped, area, histogram), or specialist charts (pie, doughnut). Configure the time range, the dataset to query against, and how much space the widget takes on the board, from a quarter width all the way to full width. A guided five step flow walks you through type, configuration, preview, and naming so you can see what you're building before it lands on your board.

The result is a dashboard built by the people closest to the data, in minutes, without a BI team in the loop.

Automation

Two kinds of automations to start, both aimed at removing the manual steps that sit between your data and the people who need answers from it.

Ingest keeps your datasets fresh without anyone having to remember to upload anything. Connect your Azure Blob Storage and ResAI will check on a daily or weekly schedule for new files, adding them to the dataset automatically so they're ready for Ask ResAI and reporting the next time someone asks. If your team drops a new invoice batch or a new shift report into storage every Monday, your dataset reflects that by Tuesday morning without anyone lifting a finger.

Reports run a saved query on a schedule and deliver the output by email or PDF. Build the question once in Ask ResAI, save it, schedule it, and the right people get the right answer in their inbox on the cadence you choose. Weekly safety summaries, monthly utilization reports, quarterly board prep, all running on autopilot.

Audit Logs

The first batch of audit logs is live. Every login, dataset change, ingestion run, Ask ResAI query, and automation run is captured with the actor, the action, the timestamp, and whether it succeeded. You can see who created or deleted a dataset, who queried what, when an automation last ran, and whether processing completed successfully.

This matters for two reasons. Regulated industries need an audit trail for compliance, and any team operating on shared data needs to understand who changed what and when something broke. Coverage will expand over time until every meaningful action is tracked, but the foundation is here today.

Settings

Configure your organization profile from Settings, including organization name, primary time zone, and the Entra credentials for SSO. If you operate across multiple tenants, for example a parent company with several operating subsidiaries, you can switch between organizations from the same place without signing out.

Platform Owners can manage role based permissions across the org. Toggle capabilities like Manage Users, Permissions Management, Add and Delete Datasets, Create Connections, Create Automations, Use Ask ResAI, Saved Queries, and View Audit Logs for each user role. This means an Auditor can be given read access to logs without being able to change data, a Contributor can upload datasets without being able to delete them, and an Automation Operator can run scheduled jobs without touching permissions. You decide who can do what, and the audit log keeps everyone honest.