Approaches Compared
Two ways to approach integration — and what separates them
Most AI implementation follows a standard pattern. Orb Lab Mesh's approach differs at several points that matter — particularly for organisations where the existing system cannot be replaced.
Back to homeUnderstanding differences in approach helps clarify which engagement type fits your situation.
Why the comparison matters
The standard pattern and its assumptions
Most AI integration engagements begin from the assumption that implementation is the goal. The assessment phase, where it exists, is typically conducted by the same party selling the implementation — which limits what it can honestly conclude.
For organisations with well-established core systems, this creates a gap: the cost of a failed connection is high, but the evaluation of risk is not independent.
What an independent assessment adds
When the assessment is separated from the implementation, the conclusion can include a recommendation not to proceed. That is not a failure of the engagement — it is the assessment working as intended.
Orb Lab Mesh's engagements are structured so that the written output is useful regardless of what the client decides next. It belongs to them, not to a subsequent engagement.
Side by side, the structural differences become clearer.
Direct comparison
| Dimension | Standard approach | Orb Lab Mesh's approach |
|---|---|---|
| Assessment independence | Assessment is typically conducted by the implementing party, whose interest in proceeding shapes what is examined. | Assessment is conducted independently of implementation. Recommending against proceeding is an accepted outcome. |
| Written output | Output is often a proposal for implementation rather than a neutral record of findings. | Output is a documented record — architecture views, assessment notes, recalculated figures — that belongs to the client. |
| Existing system treatment | Existing systems are treated as the context for adding new tools. Their constraints may not be fully surfaced. | Existing systems are the starting point. The assessment examines how a new connection would affect them and whether they can bear the additional load. |
| Fragility assessment | Risk is often addressed at a high level. Points of failure may not be documented in plain terms. | Fragility is a primary area of examination. The written output names where connections would introduce fragility, without softening the conclusion. |
| Pilot evaluation | Pilot results are often reviewed by the pilot's sponsor. Recalculation at operational scale is uncommon. | Orb Lab Mesh evaluates pilots independently, recalculates figures at full operational scale, and may conclude the pilot should not proceed to deployment. |
| Staff involvement | Staff are often consulted during implementation rather than during the decision phase. | Technical and operational staff are involved during the assessment itself, where their knowledge of existing systems is most relevant. |
| Engagement length | Varies considerably. Initial phases may be compressed to reach implementation quickly. | Three to seven weeks depending on service. The period is fixed and matched to the scope of the assessment. |
Distinctive elements arise from the choice to treat the join as the subject of the work.
What sets Orb Lab Mesh apart
Elements not found in the standard pattern
The join diagram
Each engagement includes a static diagram showing the systems on either side of a proposed connection, the information passing between them, and the point at which a person reviews it. This is not a marketing illustration — it is a working document.
Data handling panel
Our written output includes a plain statement of where information is processed, how long it is retained, and how deletion is confirmed. This applies to client data handled during the engagement and to any system being assessed.
Fixed engagement periods
Each service has a defined duration — three, five, or seven weeks. The scope is matched to the period, not extended indefinitely. Clients know what they are entering before they begin.
Results that hold under ordinary conditions are more useful than those measured under pilot conditions.
Effectiveness
What makes an assessment reliable
Measurement method examined
In pilot evaluations, we review how results were measured before treating them as valid. A favourable number produced by a flawed method is not evidence of favourable performance.
Pilot conditions compared to ordinary operation
Pilot conditions frequently differ from ordinary operation in ways that affect the result. We examine whether the pilot resembled the environment where the system would actually run.
Figures recalculated at scale
A system that performs well at pilot volume may cost significantly more at full operational scale. We recalculate and present the figures without adjustment.
Fragility named, not softened
Where a connection would introduce fragility into an existing system, our documentation names the point. We do not present fragility as a challenge to be managed during implementation.
Maintenance cost included
Every connection requires ongoing maintenance. In architecture assessments, we include an account of what each proposed connection would require to maintain — not just to install.
Human review point specified
In correspondence handling and similar automations, the point at which a human reviews output before it is sent is documented and built into the process — it is not left as an optional step.
Cost-benefit is not simply a matter of price — it includes the cost of a connection that fails.
Cost and benefit
What the investment covers and what it avoids
¥29,000 – ¥40,000 JPY
Fixed engagement fee
Each service is priced at a fixed amount. There are no variable fees, no add-ons during the engagement, and no further charge for the written output.
What it replaces
The cost of proceeding without assessment
For organisations with established core systems, a failed integration can be significantly more disruptive than the cost of not integrating at all. An independent assessment provides the information needed to decide which risk is larger.
Long-term perspective
The output lasts beyond the engagement
A documented architecture view or pilot evaluation remains useful after the engagement closes — for internal decisions, for discussions with other parties, and for reference if circumstances change.
Some commonly held ideas about AI integration are worth examining more carefully.
Common misconceptions
Ideas worth reconsidering
"A favourable pilot result means the system is ready to deploy."
Pilot conditions are often chosen or adjusted in ways that produce a result. Before treating a favourable pilot as evidence of readiness, it is worth examining whether the conditions resembled ordinary operation and whether the figures hold at full scale. They may — but the question deserves a direct answer.
"If we don't integrate now, a competitor will."
Competitor activity is a relevant consideration, but it does not change whether a proposed integration would hold under your conditions. A connection that fails in an established system creates its own disruption. Speed of adoption is not, by itself, a sufficient reason to skip the assessment phase.
"The vendor's assessment covers what we need to know."
A vendor's assessment is conducted from the perspective of making the integration work. The questions it is likely to examine are those that lead toward implementation. Questions that might lead toward a recommendation to stop are not equally likely to receive attention.
"Automation means removing the human step."
In correspondence handling and similar processes, a human review step before output is sent is not an obstacle to automation — it is a feature of a process that can be trusted. Removing review entirely changes what the organisation is responsible for when errors occur.
Sustainability in integration means the connection holds over time, not just at launch.
Sustainability and long-term results
What holds beyond the launch period
Integrations are often evaluated at the point of launch. Maintenance requirements, connection drift as either system is updated, and the ongoing cost of the human review step are less frequently examined at the outset — and less frequently documented.
Orb Lab Mesh's architecture assessments include an account of the maintenance each proposed connection would require over time. This does not prevent organisations from proceeding — but it means the full picture is available before the decision is made.
Consider an assessment