Custom Software · Blog
Introduction
Hiring a custom software company in India is not hard because talent is scarce. It is hard because many proposals look the same on paper: a polished deck, a long technology list, and a promise of a “dedicated squad” you will never meet. Founders and ops leads then discover, three months in, that the people who sold the work are not the people writing the code, repositories live in someone else’s account, and the first demo still cannot prove a single business metric.
Buyers searching for a custom software development company or a partner near Gurugram usually want one outcome: a team that can discover, ship, and hand over software you can operate. That sounds obvious. In practice it means written milestones, visible demos, access you control, and honesty about what will not fit a boutique squad. Brochure vendors optimize for winning the signature. A staffable product squad optimizes for surviving week twelve with your stakeholders still in the room.
AVYRION builds custom software, SaaS MVPs, and integrations for startups and enterprises from Gurugram. This article is the checklist we wish every buyer used before they wire an advance — not a pitch for every module under the sun. If you already know you need discovery plus a fixed first slice, skip to what to ask. If you are still comparing ten PDFs that all claim “full stack + AI + cloud,” start with the problem.
We will not invent certifications, headcount theater, or prices we have not published. We will name the questions that separate delivery from theatre, and the mistakes that burn a quarter even when the slide deck was excellent.
The problem
Most failed partner hires share a pattern. The vendor sold a catalogue: mobile, web, AI, Shopify, DevOps, and “digital transformation.” The buyer bought reassurance. Nobody locked a first journey, a success metric, or who would own the repositories on day one. Kickoff felt productive. By month two the standups were status theatre, acceptance criteria were verbal, and the only durable artifact was a Figma file nobody could ship from.
Brochure vendors are easy to spot once you know the tells. They lead with logos they cannot name in a contract. They offer a “dedicated squad” without naming the tech lead who will still be on the account in ninety days. They treat IAM and environment access as an afterthought for “phase two.” They ask for a large advance without a milestone map a non-engineer can demo. When pressed on risk, they add more buzzwords instead of narrowing scope.
A staffable product squad behaves differently. They ask which operational metric would prove the MVP worked. They write roles before screens. They put source code, pipelines, and cloud accounts where you can revoke access. They say no when the brief is really an EHR replacement, a full bank core, or an AI product with no data ownership. That refusal is not a sales failure. It is how you avoid paying for a project that should have been a different RFP.
India’s market makes the noise louder, not the decision clearer. There are excellent engineering teams across Gurugram, Bengaluru, Hyderabad, and remote-first companies. There are also agencies that staff a proposal with senior resumes and deliver with a rotating bench. Your job as a buyer is not to find “the best company in India.” It is to find a partner whose operating model matches the risk you can afford in the next two quarters.
The solution
Treat hiring as product diligence, not vendor shopping. Start with a one-page brief: care setting or business context, systems you already run, the metric that would prove success, and constraints on data, hosting, and go-live date. If you cannot write that page, you are not ready to compare proposals — you are still clarifying the problem. A good partner will help you sharpen that brief in discovery; a bad one will fill the silence with a forty-slide capability matrix.
Ask for a delivery shape you can inspect. Prefer short discovery, then fixed milestones with demos and acceptance criteria, or a named squad with a tech lead, shared backlog, and transparent velocity. Both models work. What fails is an open-ended “team augmentation” promise with no definition of done, or a fixed price for a vague catalogue. For many India buyers, discovery plus a first production slice is the cash offer that de-risks the relationship before a larger roadmap.
Lock control of the work products before money moves. Repositories should live in an organization you own (GitHub, GitLab, Bitbucket). Cloud and SaaS admin for production should be in accounts you control, with the vendor as a collaborator you can remove. IAM should be least-privilege from week one, not a shared root password in a WhatsApp chat. If a vendor insists their private Git is “faster,” treat that as a transfer risk, not a convenience.
Interrogate the advance against the milestone map. Asking for money up front is normal. Asking for a large share of the budget before any demoable slice is a yellow flag unless the scope is tiny and the materials cost is real. Tie payments to outcomes a medical director, founder, or ops lead can see: journey map and role matrix; clickable UX for the first role; working flow against a stubbed or live API; one integration; UAT with the people who do the job. If the payment schedule only matches their cashflow, not your risk, renegotiate.
Check people, not only process. Ask who the tech lead is, how many hours they spend on your account, and what happens when they go on leave. Ask for a sample of how they write tickets, PRs, and release notes. Ask how they handle production incidents after launch. A partner that cannot describe handover — repos, environments, access reviews, named internal owner — is selling a dependency, not a product.
Finally, match the partner to the work. A boutique custom software squad is often right for an MVP, a portal, an ops dashboard, or a focused integration. It is often wrong for a multi-year core-system replacement with a fifty-person program office. Honest scoping includes “you should hire differently for that.” Use case studies as pattern proof, not as borrowed prestige: anonymized delivery with metrics and constraints beats a logo wall you cannot verify.
Best practices
Write the success metric in one sentence before you issue an RFP. “Fewer no-shows,” “time-to-first-dashboard under two weeks,” or “checkout completion on mobile recovers X” — whatever is real. If vendors cannot restate that metric in their proposal, they did not read you.
Require a milestone plan a non-engineer can demo. Each milestone needs an artifact: clickable flow, staging URL, integration proof, or UAT checklist signed by an operator. Slide decks are not milestones.
Put repositories and cloud under your org from day one. Vendor-owned Git with a “we will export later” promise is how companies lose leverage at the worst moment.
Ask for IAM and environment design in discovery, not after go-live. Separate production data from shared staging. Name who can export, who can deploy, and how secrets are stored.
Prefer partners who document refusal. “We will not claim certifications we do not hold” and “this brief is an EHR replacement; start elsewhere” are signs of adults in the room.
Interview the people who will write code, not only the account manager. A thirty-minute conversation with the proposed tech lead reveals more than a capabilities brochure.
Keep commercial models explicit: fixed milestones when scope is clear; dedicated squad when the roadmap will move; retainer after launch for care. Mixing all three without labels creates disputes.
Plan handover in the same SOW as build. Repos, runbooks, access reviews, and a named internal owner belong next to go-live, not in a hopeful appendix.
Examples from real delivery
When we scope a B2B SaaS MVP, the useful first conversation is rarely “which framework.” It is which buyer workflow must work end-to-end for the next fundraising or sales quarter, and which integrations can wait. The anonymized SaaS MVP pattern on our site follows that rule: a shippable core, not a catalogue of every module on the pitch deck. Buyers who insist on every module in v1 usually hire the wrong partner for the wrong reason — speed theatre instead of learning.
On a modern banking-style platform engagement (anonymized in our case study), the buying lesson was operating model plus cloud cost discipline, not a feature laundry list. Stakeholders who asked for IaC, environments, and dashboard paths they could trust got a partner conversation about reliability next to cost. Stakeholders who only asked for “AI and mobile” would have selected a brochure. The published pattern — including roughly 35% cloud cost reduction and faster dashboard paths reported in that case — came from sequencing work, not from buying a bigger vendor logo.
Healthcare and clinic buyers often arrive with an RFP that asks for HIPAA language as if it were a plugin. The partners worth shortlisting rewrite the first slice as appointments, identity, or a trusted ops view, and put privacy into access control and logging. The ones that paste certification claims without hosting or BAA clarity are selling comfort. Our healthcare writing and anonymized analytics case exist to make that distinction boring and explicit.
Across these patterns the hiring signal is the same. The partner who can show a milestone a floor operator or product manager can click will outlast the partner who can only show a slide of technologies. Use that as your filter when five Indian vendors look identical on LinkedIn.
Common mistakes
Buying a brochure of every service line when you needed one journey and one metric. Scope inflation is how discovery turns into a year.
Accepting a “dedicated squad” without named people, time commitment, or a backup plan when the lead rotates off your account.
Paying a large advance before repositories, environments, and a demoable milestone exist. Cashflow for the vendor is not risk transfer for you.
Believing website badges for ISO, SOC2, HIPAA, or NABH without asking who holds the certificate, which entity, and what it covers. If they will not answer plainly, walk.
Letting source code and production admin live only in the vendor’s accounts “until hypercare.” Hypercare is when you most need the keys.
Skipping operator UAT. If only IT clicks through, you will ship software the desk abandons on the first busy Tuesday.
Hiring for AI slogans before identity, appointments, or order truth exist. Automation on top of chaos creates faster chaos.
Treating the cheapest fixed bid as the safest. Underpriced scope is usually unpaid risk that returns as change orders or silent quality cuts.
Conclusion
Hiring a custom software company in India is a diligence problem: brochure versus staffable squad, catalogue versus first slice, vendor-owned Git versus keys you hold. Ask for milestones, repos, IAM, and a payment shape that matches demos you can see. Refuse fake certifications and unnamed teams. Prefer partners who will narrow the brief until a non-engineer can recognize success.
If you want AVYRION to help scope that first slice — discovery, milestones, and a squad that writes in repositories you own — share the business context, systems you already run, and the metric that would prove the MVP worked. Start from the contact form or book a call on the site. We will reply with clarifying questions and an outline, and we will tell you if the work needs a different kind of partner than a boutique product squad.
