Capabilities · 02 of 04

The integration works. Until it quietly does not.

Systems talk to each other until they do not, and nothing announces it. An order does not reach the ledger. A commission calculates against the wrong entity. A payment run drops three invoices. Finance finds out at close, four weeks after it started.

We build the connections between your systems, and the visibility to know they are actually working.

Book a 20-minute review How an engagement works

The problem

Silence is not the same as success

Integration reliability fails when transactions are not reconciled from source to ledger, retries are weak and errors remain invisible until close. TechFarben maps the flow, rebuilds the weak points and makes failure visible.

Integration failure is rarely dramatic. Systems do not stop; they drift. A field changes upstream and stops mapping. A rate limit throttles a batch and half of it never retries. An exchange rate applies a day late. A rule fires against a record type nobody remembers creating.

None of that raises an alarm, because nobody built one. The error surfaces weeks later as a reconciliation difference, and by then it has propagated through recognition, reporting and possibly a payment run.

The second problem is worse and less discussed. When the connection breaks, nobody in finance can see why. The logic sits in scripts written by a supplier who has moved on, or in a middleware layer nobody has documented.

Diagnosing it means raising a ticket and waiting, which means the finance team is dependent on an outside party to understand its own numbers.

An integration you cannot observe is not an asset. It is a dependency.

What we do

Three ways in

Integration Reliability Review

A diagnostic on the connections you already have running.

Every integration in the estate, mapped: what it moves, when, under whose logic, and what happens when it fails.

We look for the four things that break mid-market finance data. Transactions that fail without raising an error. Volume that exceeds what the connection was designed to carry.

Logic that produces the wrong answer under conditions nobody tested for. And gaps where no reconciliation exists between what one system says and what the other recorded.

The output is a reliability register; every connection scored, every failure mode named, and a view of which ones can wait and which are one busy quarter away from a material misstatement.

Where an integration is sound, we say so. Most estates have two or three that need attention and a dozen that are fine.

Recent workCharles Monat Associates: a structured change-request programme against a live CRM-to-ledger integration, scoped as itemised days, covering premium adjustments, cancellation and surrender handling, co-broking arrangements and skipping rules; each change specified, delivered and confirmed.

Transaction & Revenue Engineering

High-volume plumbing, built to survive the volume. Delivered as assets you own.

This is the engineering half of the firm, and it is the work that most consultancies subcontract.

We build the connections that carry money: commerce platform to ledger, CRM to ledger, bank to ledger, policy or booking system to ledger. Order to cash, procure to pay, and the revenue logic in between; commission chains, revenue splits across entities, referral arrangements, clawbacks and reversals.

Volume is the part that gets underestimated. A connection that works at a thousand transactions a month behaves differently at fifty thousand. Platform APIs impose rate limits. Batch windows collide with close. We design for the peak rather than the average, because the peak is when it matters.

What you get

Source code and documentation, handed over. No black box, no licence, no dependency on us to keep it running. We would rather you stayed because the work is good than because leaving is difficult.

Recent worka global social commerce platform: a rate-limit-aware invoicing scheduler absorbing transaction spikes without breaching platform API limits, avoiding a licence-tier upgrade. Built on top of an order-to-cash integration between their internal platform and the ledger. Source code handed over.

AlsoCharles Monat Associates: commission and referral revenue triggered automatically from paid cases across ten entities, with revenue splits, co-broking, tiered referral arrangements and clawback reversals, and payables automated across every entity through a single bank integration. Manual effort down 90%.

Observability & Production Foundations

The layer that tells you it is working. Deployed inside your own perimeter.

Most mid-market finance estates have no monitoring at all. The alerting is a person noticing.

We deploy the infrastructure that makes automated flows visible: error tracking with the context needed to diagnose rather than just the fact of a failure, distributed tracing across systems, job and schedule monitoring so a batch that didn't run raises an alarm, performance profiling, and alert routing to the people who can act, with escalation when they do not.

Inside your network, not ours

This matters more than it sounds. Finance data, customer records and transaction detail stay within your environment. Nothing is shipped to a third-party platform. Personal data is scrubbed before it is ever logged. Access is role-based with single sign-on, and every action is auditable.

For any business handling regulated or client-confidential data, that distinction is usually the difference between a project that gets approved and one that dies in security review.

Built to be moved

The infrastructure is defined as code and portable across cloud providers or on-premise. Changes go through review before they apply. If you move cloud, or change your mind about hosting, the platform moves with you.

Recent workCharles Monat Associates: a self-hosted observability platform running entirely inside the client's network, with error monitoring, distributed tracing, release health, job monitoring, PII scrubbing and full audit logging; infrastructure defined as code and portable across providers, with review gates on every change.

How we engage

Three stages. You can stop after any of them.

01 · Diagnose

Fixed fee. The reliability register, scored and prioritised.

02 · Build

Scoped programme, billed on milestones. Code and documentation handed over.

03 · Run

Monitoring and monthly reporting, with committed response when something breaks.

Most estates need two or three connections rebuilt, not all of them. The Review tells you which, and it is the cheapest thing we sell.

Proof

What this looks like in practice

No tier upgrade needed

A global social commerce platform: transaction growth absorbed within existing rate limits by redesigning how and when records are submitted.

90% less manual effort

Charles Monat Associates: commission calculation and payables across ten entities running end to end without manual intervention.

Nothing leaves the building

A self-hosted monitoring platform inside the client's network perimeter, with PII redaction and full audit logging.

See documented client work →

Questions

Our integration was built by someone else. Will you take it on?

Yes, and most of ours are. we will review it first rather than assume it needs replacing; often it needs three fixes and a monitoring layer, not a rebuild.

Do we own what you build?

Yes. Source code and documentation are handed over. No licence, no lock-in, no ongoing fee to keep it running.

We use several different systems, not one platform. Does that matter?

No. Most of our integration work connects systems from different vendors; commerce, CRM, banking, payroll and ledger. Platform-agnostic is the normal case, not the exception.

Where does the monitoring run?

Inside your environment, if you want it there. That's the default we recommend for finance and client data, and it is usually what gets a security team comfortable.

How do you handle high transaction volumes?

By designing for the peak. Rate limits, batch windows and retry logic are addressed at design stage rather than discovered in production.

Can you just review, without building anything?

Yes. The Reliability Review is a standalone fixed-fee engagement with no obligation attached.

Start with a conversation, not a proposal

Twenty minutes. Tell us what is breaking; or what you suspect is breaking and cannot prove, and we will tell you whether it is the kind of thing we fix.

Book a 20-minute review