AI integration · LLM features for existing software

Add AI to the software you already run, without betting the product on it.

Most companies don’t need a new AI product. They need the system they already have to summarize, classify, draft, search or extract. CS2’s AI integration services add LLM features to an existing product or internal system for businesses in Canada and the US, behind a provider abstraction, with privacy boundaries, cost and latency budgets, and a rollout you can switch off in a minute.

Live AI products
3
Live REST endpoints
130+
Years on a Tier-1 telecom backend
10
When you need this

When the right move is adding AI, not starting over

Your product works and customers want AI in it

Summaries, smart search, drafted replies. The feature list is clear. The question is how to add it without destabilizing the thing that already sells.

An internal process is drowning in documents

Invoices, claims, applications, support tickets. Someone reads each one and types it into a system. An LLM can do the first pass, if the output is structured and checked.

Your first AI feature is locked to one vendor

The provider changed prices, or compliance now requires a different one. You need an abstraction layer and a safe way to switch.

Latency or cost killed the pilot

The feature worked, but it took too long or cost more than it saved. Running it for real needs a budget for both, set per feature.

Your data has to stay inside a boundary

Health, financial or personal data can only reach providers covered by the right agreement. The integration has to enforce that in code, not only in a policy document.

What we do

AI integration services built to be measured and reversed

Provider abstraction and switching

One interface for the model layer, so AWS Bedrock, OpenAI, Anthropic or a hosted open model can be swapped by configuration. We moved Notermed entirely from OpenAI to Bedrock this way.

Data privacy boundaries

Rules for which data may reach which provider, enforced in code. Where compliance requires it, the app refuses to start with an unapproved provider configured.

Latency and cost budgets

Each feature gets a response-time target and a monthly spend budget. We pick the smallest model that passes the test set, cache stable answers and cap usage per customer.

Rollout with feature flags

AI features ship dark, reach a small group first and can be switched off per customer without a deploy. Every customization falls back to the standard behaviour when it fails.

Quality measurement

An evaluation set built from your real inputs, scored on every prompt and model change. You see quality move as a number before customers notice anything.

Retrieval over your own data

Question answering over your documents with a vector store, re-indexed safely when the model changes and isolated per customer all the way into the search itself.

Proof

Integration work from the products we run and from enterprise backends

Product, wholly owned

Notermed provider switch

The AI scribe ran on OpenAI. A HIPAA requirement moved the whole AI layer to AWS Bedrock: a new model, re-indexed document stores with a rollback path, strict input limits for real PDFs. The product’s core logic barely changed.

Read the seven lessons

Product, wholly owned

AgentlyLeads

AI added to a working CRM. The 130+ REST endpoints and 49 data models existed first; the MCP connector with 57 tools came on top, with OAuth 2.1 and tenant isolation enforced on every call.

Read the case study

Past performance, enterprise

Vodafone Kabel Deutschland

Integration into large systems is backend work. CS2 spent three years (10/2022 to 10/2025) on Vodafone Kabel Deutschland’s field-operations REST backend under the GDPR: Java, Spring Boot, Kubernetes, AWS, Jenkins.

See past performance
How we work

Four steps from your codebase to a live AI feature

  1. Audit the system and the data

    We read your architecture, find the integration points and classify the data each feature would touch. That decides which providers are even allowed.

  2. Prototype one feature behind a flag

    The first feature goes into your real codebase, dark, against real inputs. Your team sees it working in context within weeks.

  3. Harden it

    Provider abstraction, privacy rules in code, cost and latency budgets, and an evaluation set that runs on every change.

  4. Roll out and measure

    Flag on for a small group, then wider, with quality and spend tracked per feature. Your team gets the runbook and the kill switch.

Stack

What we integrate with and build on

  • Provider-agnostic model layer (AWS Bedrock, OpenAI, Anthropic)
  • Spring AI (Java)
  • Model Context Protocol (MCP)
  • PostgreSQL + pgvector
  • Redis caching
  • Feature flags
  • Java 17 + Spring Boot
  • NestJS and Node.js
  • Next.js + TypeScript
  • REST and event-driven integration
  • AWS (ECS Fargate, Lambda, S3 + CloudFront)
  • Terraform, GitHub Actions, Jenkins
FAQ

Questions about adding AI to existing software

Can you add AI to software CS2 didn’t build?

Yes, and that’s the usual case. We start by reading your codebase and finding the right integration points, whether that’s a REST API, a message queue or the database layer. Our backend experience includes three years inside Vodafone Kabel Deutschland’s field-operations platform, which CS2 didn’t originate either.

How hard is it to switch AI providers later?

With a provider abstraction, the switch itself is configuration. The work is around it: re-indexing any vector stores, re-running the evaluation set, and testing with real documents, because model limits differ. For Notermed the move was a short project, and the rest of the product kept running as before.

How do you keep our data private?

We classify the data each feature touches and write rules in code about which provider may receive it. For Notermed, the app cannot start if it is pointed at a provider outside the HIPAA boundary. Customer separation is enforced inside the AI search too, not only in the database.

What does LLM integration cost per month?

It depends on volume, model and how much text each request carries, so we set a monthly budget per feature before launch rather than guess. Smaller models, caching and strict input limits usually bring the bill well below the first estimate. Usage is logged per feature and per customer so finance can see it.

How do you measure whether the AI feature is any good?

With an evaluation set made from your real inputs and the outputs your experts would accept. Each prompt or model change is scored against it before it ships. After launch we track acceptance, edits and rollbacks per feature, so quality is a number you can watch.

What happens if the AI feature misbehaves in production?

You switch it off with a feature flag, per customer or globally, without a deploy. Customizations fall back to the standard behaviour on error, and the old index or model stays available for a rollback. We design for that before launch, because it always comes up.

Which stacks do you integrate with?

Java and Spring Boot, Node.js and NestJS, Next.js and TypeScript on the application side; PostgreSQL and Redis for data; AWS for infrastructure. If your system speaks REST or publishes events, we can connect to it. For other stacks, we assess the integration points in the first call.

Tell us which system should get AI first.

Bring the product or internal tool and the feature your users keep asking for. In 30 minutes we’ll tell you where it plugs in, where the data may go and what the first flag-protected version looks like.