The problem

The next release is clear. In-house capacity is limited, and the final cost is open-ended.

Limited capacity

Your team is booked on the roadmap. The next feature, integration or release waits for people you are still to hire.

Open-ended cost

Time-based billing fixes the rate but leaves the total open. What a feature really costs becomes clear only after it ships.

Unclear accountability

AI agents speed up code production. Who reviews the code, which quality gates apply and who signs off are often not defined in the contract.

Commercial model

Pay for accepted results, not for hours or tokens

Traditional contracts pay for the work. Here you pay for the result: software you have accepted, at a price fixed before the work starts. The delivery risk on the agreed scope stays with us, not with your budget.

Hourly or token-based delivery

You buy hours and tokens

  • The invoice follows hours worked or tokens used
  • You pay as work happens, finished or not
  • AI tools and tokens can appear as extra cost lines
  • What counts as done is settled during the work

Value-based agentic delivery

You buy accepted milestones

  • The price is fixed per milestone before work starts
  • You pay when the milestone passes the agreed acceptance criteria
  • Agent licenses and tokens are included in the milestone price
  • Acceptance criteria define what done means before work starts

How it works

How each milestone runs, from scoping to payment

A Sigma Software team of engineers and AI agents works in your repositories. Every milestone goes through independent checks before your product owner accepts it.

MILESTONE FLOW 01 Scope Free scoping session Scope and dependencies Acceptance criteria Fixed price OUTPUT: MILESTONE DEFINITION 02 Build Engineers and AI agents Planning tasks Implementation tasks Test generation OUTPUT: INCREMENT 03 Verify Independent checks Validator agent checks Engineer review Test evidence OUTPUT: VERIFIED INCREMENT 04 Accept Your product owner Checks against the acceptance criteria OUTPUT: ACCEPTED MILESTONE 05 Pay Fixed price Invoiced on acceptance AI tooling included OUTPUT: INVOICE NEXT MILESTONE FROM YOUR BACKLOG, SCOPED AND PRICED THE SAME WAY
MILESTONE COMMITMENT 01ScopeFree scoping sessionScope and dependenciesAcceptance criteriaFixed priceOUTPUT: MILESTONE DEFINITION 02BuildEngineers and AI agentsPlanning tasksImplementation tasksTest generationOUTPUT: INCREMENT 03VerifyIndependent checksValidator agent checksEngineer reviewTest evidenceOUTPUT: VERIFIED INCREMENT 04AcceptYour product ownerChecks against the acceptance criteriaOUTPUT: ACCEPTED MILESTONE 05PayFixed priceInvoiced on acceptanceAI tooling includedOUTPUT: INVOICE

Getting started

Start with a free scoping session, then pay per accepted milestone

Scoping session

Free · No obligation to proceed

Together with your product owner, we review your roadmap, codebase and constraints and split the scope into milestones.

You get:

  • Milestone plan for your roadmap
  • Acceptance criteria per milestone
  • Fixed-price proposal for the first milestone

Milestone delivery

Fixed price per milestone · Paid on acceptance

A Sigma Software team of engineers and AI agents delivers each milestone in your repositories.

You get:

  • Working increment in your repository
  • Tests and documentation
  • Acceptance review with your product owner
  • Agent licenses and tokens included

Continuous roadmap delivery

Next milestones · Same commercial model

The team continues with the next milestones from your backlog, each scoped, priced and accepted the same way.

You get:

  • Rolling milestone plan
  • Release cadence agreed with you
  • Code, tests and agent configurations owned by you

Human oversight

Engineers direct the work and decide what is ready for your review

AI agents take on planning, implementation, test and validation tasks. Sigma Software engineers make the technical decisions, review the changes and the test evidence, resolve the issues that agents escalate, and decide when a milestone is ready for your product owner’s review.

ROLES AND DECISION RIGHTS YOUR PRODUCT OWNER Client Sets priorities and access Gives domain input Accepts the milestone SIGMA ENGINEERS Technical decisions Split work into tasks Direct agent tasks Review changes and tests Present the evidence AI AGENTS Delivery method Planning tasks Implementation tasks Test and validation tasks
WHO DOES WHAT 01Your product ownerClientSets priorities and accessGives domain inputAccepts the milestone 02Sigma engineersTechnical decisionsSplit work into tasksDirect agent tasksReview changes and testsPresent the evidence 03AI agentsDelivery methodPlanning tasksImplementation tasksTest and validation tasks

Track record

Months of work delivered in weeks, with quality gates intact

About 400 unique features in four months at Ringorang

About 400 unique features in four months at Ringorang

  • Test coverage kept above 75%
  • An average of 100 features a month
  • 12 specialized agents and 40 skills

Ringorang needed more delivery capacity for its next-generation platform without growing the team proportionally or compromising quality and security.

We built an agentic delivery lifecycle on Claude Code and Claude Cowork, with product and engineering agent workstreams, independent validator gates and human approval at each phase boundary.

View Case Study
DripDrip Pharma platform in production within ten weeks

DripDrip Pharma platform in production within ten weeks

  • Conventional estimate: around five months
  • Six production releases in ten weeks
  • Desktop page build: one hour to 18 minutes

DripDrip Pharma, a European healthcare company, needed to launch a B2B medical e-commerce platform on WordPress, WooCommerce, Dokan and Stripe on an accelerated schedule.

We set up an AI-native SDLC across engineering, design, product and project management, with Claude Code as the shared working layer for both teams.

View Case Study

Have a feature, integration or release in mind?

Tell us the objective, your target date and the main constraint. Code and confidential details can wait until the scoping session.

Why Sigma Software

Agent teams with independent quality gates and terms that protect you

Independent quality gates

Every stage, from planning to security review, passes through a validator agent that is separate from the agent that produced the work. Unresolved issues go to an engineer after three auto-fix attempts, and engineers approve every release.

Built on our proprietary agentic stack

Delivery runs on our proprietary Enterprise Agentic Engineering Stack: the agent workflows, reusable skills and built-in checks proven in our client work. Your milestone runs on a battle-tested setup instead of one assembled for each project.

Anthropic Select Services Partner

Sigma Software is an Anthropic Select Services Partner with 139 Claude-certified engineers.

Code and agent setup stay yours

The code, specifications, tests and agent configurations built for your product belong to you, together with the IP, so your team can continue the work after handover.

Questions before you start

What is value-based agentic delivery?

It is a way to buy software product engineering services by the accepted milestone. Sigma Software engineers use AI agents to deliver agreed software milestones for product teams, and you pay a fixed price per milestone after your product owner accepts it. Your team provides a product owner, domain input and the access agreed in the delivery plan.

How is the milestone price set?

In a free scoping session we agree the milestone’s scope, dependencies and acceptance criteria, then propose a fixed price. The AI tooling used to deliver the milestone, including agent licenses and tokens, is included. The costs of running the software you operate, such as hosting or AI model usage in production, are agreed separately where they apply.

What happens if a milestone does not pass acceptance, or requirements change?

A milestone is invoiced only once your product owner accepts it. The fixed price covers the agreed scope and acceptance criteria. If a requirement changes, we review the change with you and agree any revised scope, criteria or price before going further. Corrections and any disagreement follow the engagement terms set before delivery starts.

Who owns the code?

You do. The code and IP delivered in the milestone belong to you. Pre-existing and third-party components, such as open-source libraries or Sigma Software’s own reusable tooling, are identified and treated in the engagement terms. Handover items such as repository access and documentation, and who maintains the software, are agreed for each engagement.

How is quality controlled when AI agents write code?

AI agents take on planning, implementation, test and validation tasks. Sigma Software engineers make the technical decisions, review the changes and the test evidence, and decide what is ready for your review. Your product owner then accepts the milestone against the agreed criteria. The tests, tools and release approvals for each milestone are set in its delivery plan.

How does this compare with AI pods or time-and-materials outsourcing?

Time-and-materials contracts bill hours, and AI pod offers typically bill team capacity, a subscription or usage. Here you buy one defined milestone at a fixed price and pay after acceptance, with AI delivery tooling included. For continuing capacity across evolving priorities, dedicated development teams remain the better fit.

Can agent teams work on code that must not leave our perimeter?

We assess that during scoping, before any work starts. Together we establish which AI models may process your code, how repository and environment access works, and any other constraints. The milestone’s delivery plan records the tools and controls that apply.

Can our own engineering teams work this way?

Yes, as a separate engagement. If your own engineering team should adopt AI agents in its delivery, see AI-Powered SDLC. Milestone delivery fits when one feature, integration or release can be scoped and accepted on its own.

Related engineering services

Digital Product Development

Product development, enhancement and maintenance.

AI-Powered SDLC

Bring AI agents into your own engineering lifecycle.

Dedicated development teams

Continuing engineering capacity for evolving priorities.

Tell us what you need delivered next

Describe one feature, integration or release, your target date and its main constraint. Free scoping turns it into a milestone definition, acceptance criteria and a fixed-price proposal.

Sigma Software has offices in multiple locations in Europe, Middle East, Northern and Latin America