Your logs, answered.
Your work, automated.
ForgeOrbis is an on-prem intelligence and automation platform on one local model, entirely inside your perimeter. Log Intelligence answers questions about your machine logs with exact, cited evidence. Orbis Flow is sprint planning for managers and scrum-masters: a plain-language goal becomes epics, assigned tickets, and reports, with scope and acceptance criteria written in. Standalone, or fed by Log Intelligence. It never calls an external AI provider.
No data leaves your network. Air-gap capable.
The platform
One local model. Two products. Zero egress.
ForgeOrbis runs entirely inside your perimeter. Log Intelligence finds what is happening across your systems; Orbis Flow turns goals and findings into tracked, delivered work. Together they close the loop from detect to done.
Log Intelligence
Find what is happening.
Ask your machine logs in plain language. Get an exact count, or a grounded answer with the file, line, and match score behind every claim.
- Exact counts, computed not generated
- Cited, verifiable evidence on every claim
- Honest refusal when nothing matches
Orbis Flow
Deliver the work.
Plan a sprint from a plain-language goal. Orbis Flow drafts tickets with scope and acceptance criteria, assigns them, runs the ceremonies, and ships the report.
- Epics and tickets from a goal or brief
- Assigned, estimated, tracked to done
- Standups, retros, and sprint reports
Log Intelligence · the shift
Stop grepping through thousands of folders by hand.
Regulated plants keep months of machine logs in date-and-process folder trees. Answering one auditor question can mean hours of manual searching. ForgeOrbis turns that into a sentence.
Today
A manager opens a share, guesses the right date folder, and reads log files line by line hoping to reconstruct what happened.
$ cd //HOST07/app-logs$/printsvc/LabelRun/20260709
$ grep -ri "print queue" . | less
… 4,912 matches across term windows
With ForgeOrbis
One question, one cited answer, with the exact file, line, and match score, verified against the index and never invented.
Printing halted at 14:22 when the print queue connection reset, then recovered at 14:24.
How it works
Three steps, entirely inside your perimeter.
Ingest & structure
Point Orbis at your log folder path. It parses date and process branches, deduplicates templates, and builds a private index in place, inside your network.
4.1M events · 71% dedup
Ask in plain language
Type the question an auditor would ask. Orbis routes it: counting questions to a precise counting engine, explanatory questions to grounded, cited evidence.
42 candidates · ranked
Get cited, scored answers
Every answer carries its citations: file, line, timestamp, and a match score, so anyone can verify it. Nothing is generated without evidence.
match 0.94 · high
Two-engine routing
Exact when it can be exact. Grounded when it must explain.
Counting questions get an exact figure computed from the logs Orbis indexed. Explanatory questions get an answer built only from cited log evidence. Orbis picks the right engine for every question, so no one writes a query or greps a folder.
Tier 0 · exact
exact · verifiable“how many labels did STATION_A1 print on 2026-07-10?”
Counted directly from the logs Orbis indexed. Exact and repeatable, never generated.
Tier 1 · grounded
cited evidence“why did LabelRun stop printing around 14:20 on 2026-07-09?”
The printservice service lost its print queue connection at 14:22, retried three times, and recovered when the watchdog restarted it at 14:24. No labels were lost.
Every clause traces to a line you can open.
Orbis Flow · work automation
Describe the goal. Approve the plan. It ships to your board.
Orbis Flow is project management for managers and scrum-masters, automated. Turn a plain-language goal into a full sprint plan, review and edit it, and approve, then it creates the tickets in your real boards with owners, estimates, dates, and links. On the local model, inside your walls, and no log connection required.
Plan from a goal or a PRD
Describe the outcome or paste a brief. Orbis Flow drafts epics and stories with scope, acceptance criteria, estimates, risks, and dependencies, and reasons through the breakdown in the open.
Review and approve, you decide
Edit any description, acceptance criteria, assignee, points, or dates. Orbis balances the work across your team, but nothing is created until you approve.
Execute to your real boards
On approval it creates real epics, stories parented to them, assignees, points, dates, and dependency links, in every board you connect: Jira, GitHub, or Asana.
Run ceremonies and report
Standups, retrospectives, sprint reports, and burndown, written from live board state, so the numbers are computed, not guessed.
Credentials are entered in the app and encrypted at rest. Optional: a finding from Log Intelligence arrives here as a pre-filled, cited ticket, so an incident becomes tracked work in one step.
Connectors & alerting
When Orbis finds it, the right people already know.
An answer only matters if it reaches the person who can act on it. The moment Orbis detects an issue in your logs, it routes the grounded finding to the responsible owner, opens a ticket, and keeps the channel updated, with no one watching a dashboard.
- 01
Detect
Orbis spots the issue in your logs and grounds it in cited evidence: the file, the line, the match.
- 02
Route
Severity and the responsible owner come from your ownership map, so the right manager is chosen, not everyone.
- 03
Notify
The owner hears it in Slack, Teams, or email, and on-call is paged for a SEV-1.
- 04
Ticket
A Jira, ServiceNow, or Asana ticket opens automatically with the finding and a link to the evidence.
- 05
Update
Orbis posts updates to the channel and closes the ticket when the condition clears.
Connect & get notified
try it, toggle a toolConnect your tools
one click · revocableOutbound only, opt-in, and audited. Log content and inference stay sealed in your perimeter. Alerts leave only for destinations you allow-list, carry the finding and a link to the evidence (never raw log lines), and every dispatch is written to the tamper-evident audit log. Air-gapped sites route the same alerts to on-prem targets: Mattermost, Jira Data Center, an internal SMTP relay.
The cost of the status quo
What is manual log search costing you?
Two honest numbers about how your team answers log questions today. ForgeOrbis returns each of those answers in seconds, cited.
Opening shares, guessing the right date folder, reading files by hand. ForgeOrbis runs the same search in seconds and shows the citation behind every answer.
analyst time reclaimed / month
54 hrs
About 6.8 working days handed back, every month.
manual, today
108 searches
with ForgeOrbis
seconds each
An estimate from your own inputs, not a measured figure. The point is the order of magnitude.
Security
Sealed by design.
Security here is a feature, shown with quiet confidence. Everything an auditor or CISO needs to relax is built into the product, not bolted on after.
Zero data egress
No log, prompt, or answer leaves your network. Outbound is default-deny.
On-prem & air-gap
Runs fully offline. No license phone-home, no telemetry, no external calls.
Role-based access
Scope a role to only its folder prefixes. Auditors see everything; a line sees its line.
Immutable audit trail
Every question is hash-chained and tamper-evident, ready to export to your SIEM.
Your own weights
Orbis is fine-tuned to your logs and stored on your hardware. You hold the keys.
Attestation pack
Ship compliance a firewall export, SBOMs, and signed model hashes on demand.
A live outbound attempts: 0 counter and downloadable attestation pack ship with every deployment.
The model
Orbis: your model, your walls.
Orbis is a language model fine-tuned on your own logs and deployed on your own hardware. The weights never leave your perimeter, never sync to a vendor, and never call an external provider. You own the model that reads your data, completely.
Deploy anywhere
Wherever your perimeter is, Orbis runs inside it.
On-prem
Install on your own servers. Bare metal or your hypervisor of choice.
Private cloud (your VPC)
Run in your own account and network. No shared tenancy, no vendor plane.
Air-gapped
Fully offline installs for classified or isolated networks. Signed bundles.
Request a deployment
Put the whole platform inside your walls.
We scope your deployment, size the hardware, and stand up Log Intelligence and Orbis Flow in your environment on one local model. Nothing leaves your network, including this conversation.