Knowledge and learning performance · COOs

Lumina

Identify knowledge gaps.
Lift learning performance.

Lumina identifies knowledge operational gaps across time-to-retrieval, decision quality, and SOP currency; then uses AI-assisted workflows to lift COO, CHRO, and L&D team output. The system combines a knowledge graph, spaced-repetition, and concept-drift scoring inside Ordumo's 30 metrics in 6 areas.

Find retrieval gaps before they become repeated decisions
Keep SOPs current with concept-drift scoring
Ranked fix-list: which workflow, which document, which learning loop

Book a $15K Lumina audit How it works

Lumina/ Retrieval probe

Live · this morning

? "What was our March pricing decision and who signed off?"

Resolved · 0.4s 3 sources cited

Pricing memo · M. Chen Notion · Mar 12

Board notes · Q1 review Drive · Mar 18

RevOps update memo Slack · Mar 22

Accuracy · 97% Peer median · 68%

What Lumina measures

9 metrics. The knowledge slice of the 30-metric framework.

Every Ordumo product runs the same 30-metric framework. Lumina focuses on the 9 metrics that decide whether your knowledge work is compounding into institutional memory — or evaporating into chat logs.

01Decision quality

Are AI-assisted decisions sticking?

Decision-to-action latency
Recommendation outcome rate
Re-decision frequency
Forecast / call accuracy

02Retrieval & recall

How fast can people find the answer?

Time-to-retrieval
Retrieval accuracy
Knowledge half-life

03Production velocity

Is the analyst layer compounding?

Output per FTE
Approval & rework delay

04AI & data readiness

Are AI assistants citing the right docs?

Source coverage
Schema completeness

How Lumina works

Two weeks in.
The fix ships with evidence.

Same 3-step rhythm. Lumina just runs knowledge-work-specific probes — like asking your AI assistants real internal questions and scoring the answers.

01.

Connect your knowledge base

Read-only access to Notion / Confluence / SharePoint / Slack / Drive. Takes a day. We don't train on your data — pull-only.

Week 1 · 1 day

02.

Run the retrieval probes

We send 100+ real knowledge queries through Claude / GPT / Gemini against your stack, score the answers for accuracy and citation, and measure retrieval time end-to-end.

Week 1–2 · 10 days

03.

Get a ranked fix-list

Where the assistant confabulates, where retrieval fails, where decision quality dips. Each fix tied to a specific number.

Week 2 · 2-hour readout

04.

Ship & re-measure

Re-organise docs, install retrieval guardrails, swap models, retire underperforming AI workflows. Monthly re-benchmark.

Day 15–90 · then monthly

What you walk out with

A sample Lumina Benchmark deliverable.

Every engagement ends with the same artefact pack. So engagements compare like-for-like — and so the only variable is the work.

Lumina · Benchmark Report Sample · illustrative

"Where your AI assistants confabulate — and what to fix first."

The package

30-metric scorecard PDF
Domain probes report Live link
Peer cohort comparison Anonymised
Ranked fix-list · P0 → P3 Priorities
Evidence-backed deployment plan 1-pager
Operating-system spec Buildable
Evidence ledger seed Source-cited
Board-ready exec brief 1 page

Typical findings

AI assistants citing outdated runbooks
Retrieval accuracy below 60% on critical queries
Knowledge half-life under target threshold
Decision-to-action lag > 5 business days
Slack thread sprawl vs. living docs
Schema gaps in foundational sources
Output per FTE flat vs. peer cohort
Re-decision rate on the same issue

Connects to your stack

No re-platforming.
Plugs into your knowledge stack.

Read-only access. Pull only — never write. Connection takes a day; no IT marathon required.

Notion

Wiki · CMS

Confluence

Wiki

SharePoint

Docs

Slack

Comms

Google Drive

Docs

Box

Docs

Glean

Coda

Docs · ops

Linear

Engineering ops

Asana

Project ops

Loom

Video knowledge

Monday

Ops

Lumina · Pricing

Fixed prices.
No surprise invoices.

Same three engagement steps across all four Ordumo products. You see the price before you see the pitch.

Step 1 · Measure

Benchmark

$15K–$30K

2-week Lumina audit

Score across 30 metrics
100+ retrieval probes
Peer cohort comparison
Prioritised fix list

Step 2 · Build

Deploy

$10K–$30K

Build window

Operating-system spec ships
Internal or partner buildout
Each fix tied to a metric
Quarterly board brief

Step 3 · Compound

Operate

$5K–$15K/mo

Monthly rhythm

Monthly re-measurement
Outcome track record
Quarterly board brief
Cancel anytime

Common questions

Things ops leaders ask before booking.

How is Lumina different from Glean / Notion AI / Microsoft Copilot?

Glean, Notion AI and Copilot are retrieval tools. They answer questions. Lumina audits whether the answers are actually right — and whether the rest of your knowledge-work operating system is compounding around them.

What counts as "knowledge work"?

Analyst work, ops, research, strategy, customer-facing knowledge resolution. Anything where the team's job is to get to a good answer rather than ship a deliverable.

Will you ingest our private documents?

Read-only. Pull-only. We do not train on your data; the AI assistants you already use (Claude, GPT, Gemini) get scoped, revocable access for the retrieval probes.

How small does our team need to be?

If you have < 20 people doing knowledge work, you probably don't need Lumina. Above 50, the half-life and retrieval gaps start costing real money.

How do you measure "decision quality"?

We sample N decisions from the last quarter, score each on outcome (did it move the metric?), latency, and recall. Benchmarked against 30+ similar organisations' anonymised decision logs.

The other 3 products