Question 1

When do you need a custom integration instead of an off-the-shelf tool like Zapier or Workato?

Accepted Answer

No-code tools are the right call at low volume and simple, linear workflows, and cost little to set up. They start breaking down once you have conditional logic across more than two or three systems, need sub-second latency, or hit their per-task pricing at real volume — that's usually the point a Zapier or Make chain built for 50 records a day starts silently dropping or duplicating records at 5,000. A custom integration costs more upfront but is built and tested against your actual failure modes and volume, not a generic connector's defaults.

Question 2

What happens when the other company changes their API and our integration breaks?

Accepted Answer

That's exactly the situation we plan for. We monitor for it, and because we stay the point of contact, you don't have to figure out who's responsible. You call us.

Question 3

How do you handle webhooks so events aren't lost or duplicated?

Accepted Answer

Every webhook provider retries failed deliveries, which means duplicates are guaranteed, not an edge case — Stripe retries for up to 72 hours, for example. We deduplicate on the event's unique ID before processing, so a retried delivery gets acknowledged but not reprocessed, and we return a success response for duplicates rather than an error, since an error just schedules another retry. Ordering gets handled separately, since webhooks aren't guaranteed to arrive in the order they were sent.

Question 4

How do you handle a third-party API that has rate limits or goes down?

Accepted Answer

Rate limits get handled with backoff and queuing, so a burst of activity on your side doesn't get you throttled or banned by the other API. An outage on the other side gets a defined fallback — queue and retry, degrade gracefully, or alert a human — decided before launch, not improvised during the incident. That failure-mode plan is part of the architecture phase, not something we figure out after the first outage.

Question 5

How do you handle API versioning when the other side ships breaking changes?

Accepted Answer

We monitor for deprecation notices and version changes on the APIs we integrate with, and pin to a specific API version where the provider supports it, so a change on their end doesn't silently break your integration the day it ships. When a breaking change is unavoidable, it goes through the same monitoring window and point of contact as any other integration issue — you find out from us, not from a support ticket.

Question 6

How do you test an integration before it goes live?

Accepted Answer

Against failure cases on purpose, not just the happy path: rate limits, timeouts, malformed or unexpected responses from the other system, and load testing at the volume you actually expect, not the volume the demo used. Most integration failures we get called in to fix trace back to a case that was never tested, because it worked fine in every demo and only showed up at production volume.

Question 7

How long does an integration take?

Accepted Answer

It depends on what's being connected. A single payment gateway hookup can be done in days. A full CRM-to-ERP integration with custom data matching takes weeks. We give you a real number after a short call, not a guess.

Question 8

What systems do you connect?

Accepted Answer

Payment providers like Stripe and Razorpay, CRMs like Salesforce, HubSpot, and Zoho, ERPs like SAP and Odoo, and sign-in and access systems using OAuth 2.0 or Auth0. If it has an API, we can likely connect it.

Question 9

Do you build the AI-facing side of an API?

Accepted Answer

This service covers the connection itself: the data, the authentication, the sync logic. AI agent access is out of scope here; we'll point you to a partner who specializes in it.

Question 10

Custom API integration vs an iPaaS platform like Zapier, Make, or Workato — which is right for us?

Accepted Answer

iPaaS platforms fit simple, low-volume workflows between common apps and get you running in a day with no code. A custom integration fits when you need conditional logic across more than two systems, real-time sync instead of polling, sub-second latency, or volume that would put you into per-task pricing that no longer makes sense. Several of our engagements start as a fragile Zapier or Make chain that worked at low volume and is now dropping records — we replace the connector, not the entire architecture.

Question 11

Who owns the integration once it's live, if the original API changes?

Accepted Answer

We do, for the monitoring window in your scope. If the other side changes their API, we catch it and fix it as part of that window. Beyond it, you can renew monitoring or bring it in-house, since you own the code and configuration either way.

Question 12

Do you sign an SLA for the integration's uptime?

Accepted Answer

We scope response-time commitments before launch, specific to what's being connected. A payment gateway integration and an internal reporting sync don't carry the same urgency, so the SLA is matched to that, not a generic template.

Question 13

How is scope defined before you start, since you don't publish fixed prices?

Accepted Answer

Scope is defined after a short call: what's being connected, what's excluded, and who owns the integration once it's live, before any work starts.

Question 14

What actually drives the cost of an integration?

Accepted Answer

Mainly four things: how many systems are involved, how much data moves and how often, whether it's real-time or batch, and whether you want ongoing monitoring after launch or a one-time build. We walk through all four on the scoping call before quoting anything.

Situation	Likely Solution
An integration broke and both sides blame each other while the customer waits	Deploy & Support — one point of contact and ownership defined before launch
No-code automations (Zapier/Make) that worked at low volume start dropping or duplicating records	Engineering — a purpose-built API layer with retry logic and rate-limit handling
Two systems are silently drifting out of sync until a customer notices first	Architecture — data mapping and sync direction defined up front
A vendor changed their API and the first signal was a support ticket, not an alert	Monitoring & alerting on the connection, built in before launch

Before	After
Manual CSV exports and imports bridging two systems	An automated, monitored API sync between them
No-code automations that quietly drop records past a certain volume	A purpose-built integration tested against real failure modes and load
Integration failures discovered by a customer complaint	Failures caught by monitoring and alerting before customers notice
No agreed answer for who owns the integration when it breaks	One point of contact, with a response-time commitment matched to what's connected

Factor	No-code tools at scale	Generic API dev shop	DharmOps
Ownership after launch	Usually nobody's, once the consultant leaves	Varies by contract	One named point of contact, defined before launch
Handles real failure modes	Breaks silently past a volume threshold	Depends on the shop	Built and tested against rate limits, timeouts, malformed responses
Data-consistency expertise	Not built for it	Variable	Where the data lands and how it's stored is core to what we do
Pricing model	Per-task/monthly SaaS fee	Often fixed quote regardless of complexity	Scoped after a short call, matched to what's actually being connected

Custom API Development & Integrations for Growing Businesses

What Is Custom API Development & Integrations?

What Problems Trigger This Engagement?

An integration broke and nobody knew whose job it was

No-code glue is breaking under real volume

Two systems are silently drifting out of sync

A vendor changed their API and production broke

Match Your Situation to the Fix

What DharmOps Builds

Specific Use Cases

How the Engagement Works

1. Architecture

2. Engineering

3. QA

4. Deploy & Support

Technology & Platforms We Connect

Payments

CRM

ERP

Sign-In & Access

Before → After

How This Differs From Alternatives

Frequently Asked Questions

Data Infrastructure Engineering for 13 Industries

Tell Us What You Want Connected

Connect Your Systems With One Point of Contact, Not Three Vendors