Weekly IT health brief
Northwind Studio
Overall healthStable
92What changed
Backup coverage needs attention.
One scheduled backup finished outside its expected window. Core services remained available.
3 linked factsEvidence ↗
Offline evaluation · Summer 2026
Turn tool data into client-ready proof.Without sending it away.
Lighthouse helps solo MSPs and small IT consultancies turn read-only infrastructure snapshots into a concise, reviewable client brief—with every claim linked to inspectable evidence.
- Zero-egress evaluation
- Fixture-only start
- No automatic email
Jul 13–19
✓Claim verified
↗Source attached
01
Offline before integratedStart with fixtures and no network dependency.
02
Proof before promisesEvaluation results stay separate from market hypotheses.
03
Consent before live dataA named operator controls every step beyond evaluation.
The reporting gap
Your tools know what happened. Your client still needs the story.
Monitoring platforms are built for operators. Clients want a short, trustworthy explanation of health, risk, and completed work.
× Raw dashboards create questions instead of clarity.
× Manual reports consume the time they are meant to prove.
× Ungrounded AI summaries can turn a small miss into a big trust problem.
A boringly safe workflow
From infrastructure signal to reviewable client proof.
Collect read-only facts
Bring structured snapshots from the tools you already run. No remediation access and no mystery agent changing infrastructure.
Build the evidence trail
Every metric carries its source, freshness, and confidence. Secret-shaped values are removed before reporting.
Review the client brief
Get a plain-language summary with each statement tied back to inspectable evidence. Preview first; delivery stays human-approved.
Built to earn trust
AI writes the explanation. It does not invent the record.
The reporting engine accepts only structured, validated evidence. Each sentence must point back to facts with provenance, freshness, and confidence.
Default redactionFreshness warningsDeterministic metricsSingle-use approval
Client-facing statement
“Core services remained available throughout the reporting period.”
✓
Supported by 4 evidence recordsUptime snapshot · collected 11 minutes ago
PolicyUnsupported claims fail closed
Evidence status
What is demonstrated—and what still needs a real pilot.
Demonstrated in the evaluation
428 automated checks pass against synthetic fixtures.
- Unsupported claims fail closed
- Freshness and provenance remain inspectable
- Evaluation sends no email or source data to a model provider
Not yet proven
Commercial value requires consented field evidence.
- How much reporting time an operator saves
- Whether clients find the brief more useful
- Which pricing model earns willingness to pay
Gated pilot path
Five deliberate gates from sample data to a paid decision.
Nothing skips the gate in front of it. Outreach begins only after the named five-seat slate is reviewed.
Synthetic evaluation
Run the workflow offline with included fixtures. No client data, live credentials, email delivery, or model-provider traffic.
Reviewed five-seat slate
A named candidate list is reviewed before any invitation is sent. No scraped list and no surprise outreach.
Consented integration
Only a willing evaluator can authorize a narrow, read-only connection using data they are permitted to share.
Measured pilot
Test report usefulness, operator effort, and trust with explicit success criteria. Findings stay separate from assumptions.
Paid conversion
Offer a paid plan only if the pilot demonstrates value and the operator wants to continue.
Pricing research
Two hypotheses to test—not prices pretending to be settled.
Hypothesis A
$99 / MSP seat / month
Simple operator-based pricing for a small reporting team.Hypothesis B
$15 / managed client / month
Usage-aligned pricing that scales with the client book.No checkout is open. Pilot evidence will decide whether either model is useful, affordable, and viable.
Start offline
Inspect the workflow before you trust it with anything real.
The first step uses included synthetic fixtures on your own machine. It does not connect to client systems, call a model provider, collect credentials, or send email.
- Run the included fictional scenario
- Inspect every evidence-linked claim
- Decide whether a consented pilot is worth discussing
Good fit if you…
Manage 5–50 small-business environments.
01
Already collect useful signals across more than one tool.
02Spend time translating technical status into client language.
03Prefer read-only, self-hosted-first software over another agent.
Straight answers
What the pilot is—and what it is not.
Is this another RMM or monitoring dashboard?
No. TT Solutions IT Evidence Reports sits after your existing tools. It turns the read-only snapshots you choose to provide into a concise, client-facing draft with a visible evidence trail.
Does AI decide whether a system is healthy?
No. Metrics, thresholds, freshness, and validation are calculated by deterministic code. AI is asked to explain only facts that passed those checks, and the output remains reviewable.
Does it need write access to client systems?
No. The pilot is designed around read-only inputs and sanitized or non-production data. It does not remediate, restart, patch, or reconfigure infrastructure.
Can it email a client automatically?
No. The evaluation can create a local email preview, but it cannot send email. A later pilot may test a consented delivery integration only after a named operator approves the scope.
Does evaluation data go to an AI provider?
No. The current evaluation is offline and fixture-backed. It sends no source data to a model provider, makes no live infrastructure connection, and collects no production credentials.
Is the pricing final?
No. The two prices shown here are research hypotheses for pilot conversations, not an offer or a checkout. We will keep, change, or reject them based on consented pilot evidence.