In a four-agent Azure DevOps pilot at a Global 500 consumer products company, product leaders spent less time maintaining the backlog.
- of items had piled up
- 1,000s
- from intake to priority
- 4 agents
- backlog changes
- Secure
The extra code waits in review or gets approved unread, while the board asks why AI hasn't paid off. We map where work gets stuck. Then our forward deployed engineers build the fix inside your teams, so delivery speeds up without the added risk.
Our clients include
The licenses arrive and each developer works out alone how to use them. Nobody has agreed who reviews the extra code, what the agents may touch, or how you'll know it's working.
Developers do finish more work, and it lands on review, testing and release steps sized for the old volume.
01
Our coaches and your team map how work gets from request to release.
You see where work waits, and for how long.
02
We set a baseline from the tools you already use, in measures such as lead time and time in review.
You can judge every later change against it, including what it costs to run.
03
Our forward deployed engineers try one change at small scale, such as an agent that handles intake.
A change that does not beat the baseline goes no further.
04
They then build the changes that worked into your teams' daily work.
Your people learn to run them, and we go back to the map for the next place where work waits.
Tools we work with include

A value-stream assessment covers the first two steps, and our senior coaches run it with your team. Assessments start at $5,000, priced by the number of teams and how far the work goes. You approve the scope and the price before any work starts.
Not ready to talk yet? Two free tools give you a first read.
A forward deployed engineer, or FDE, works inside your organization, in your codebase and beside your teams. Ours combine coaching and engineering talent under the guidance of Chris Sims and McCaul Baggett.
What they build depends on what your map found. Their job is to improve what your teams deliver, and they treat token spend as a cost.
Your developers learn on their own delivery work, in courses our coaches write around your codebase, your tools and your policies. Subjects include Lean, Agile and Scrum, DevOps and AI development, from writing a specification an agent can follow to checking the code it returns.
A pipeline that tests and releases code automatically is the foundation of any AI platform. When yours can't keep up with the extra code, our engineers improve it or rebuild it in GitHub Actions, Azure Pipelines or GitOps tools.
Your agents need one place to run, with shared permissions, guardrails and context. We build that platform on tools such as Microsoft Foundry, Claude Managed Agents, LangChain, Warp Factories and our own Cadence.
We implement a full AI software development lifecycle, or AI SDLC. Agents take on defined steps from specification to release, with the context each step needs, and your engineers review the work at each gate.
We build an agent for a job your team needs done, such as triaging intake, reviewing pull requests or preparing a release. Then we deploy it and run it for you as a managed service.
“These are not common staff augmentation engineers. These are people who will consider the system architecture, come up with solutions, and present options for a path forward.”
In a four-agent Azure DevOps pilot at a Global 500 consumer products company, product leaders spent less time maintaining the backlog.
On a VB6 loan platform rebuild, average time from opening a pull request to merge fell from 4.1 days in April–May to 2.5 in July–October 2026.
Story points completed per sprint
+373%More story points in sprint ten than in sprint one, with the same team and point scale, April to early November 2026.
Time to release a feature
1.5 weeksDown from about a month.
Time to merge a pull request
38% lessAverage calendar time from opening to merge, April–May compared with July–October 2026.
Issues caught in review, before merge
14.8 pointsDrop in pull requests where review caught an issue, as the team sped up. Each one was fixed before merge.
These are the client's own engineers, working with our agentic development process. The issues are ones automated review caught and the team fixed before merge. None was released.
“They're communicative and always bring up potential delays before they become issues.”
AI Enablement is the engineering and coaching work that makes AI improve how a whole team delivers software. It goes beyond giving developers a coding assistant.
A forward deployed engineer, or FDE, is an engineer who works inside a client's organization and builds in that client's own codebase and tools. Sigao's forward deployed engineers combine coaching and engineering talent. They are guided by Chris Sims, Sigao's Managing Partner, and McCaul Baggett, who leads organizational coaching. After a value-stream assessment, they fix the places where the map shows work waiting, for example by rebuilding a DevOps pipeline or building a custom agent. Their target is the value your teams deliver, measured against your baseline, and token spend counts as a cost.
Developers do finish more work with AI, and the review, testing and release steps after them have to absorb it. Across 22,000 developers, the Faros AI Engineering Report 2026 compared each organization's lowest and highest periods of AI adoption. Tasks per developer rose 34%. Median time in pull request review rose 441.5%, and median time to first review rose 156.6%. Unreviewed merges rose 31% and incidents per pull request rose 243%. Security is a second cost: in Veracode's 2026 testing, AI models wrote code that failed security tests about 44% of the time when the prompt did not ask for secure code. A value-stream assessment shows where that extra work waits on your own teams.
Possibly nothing. The assessment shows what still needs doing, and sometimes the answer is that more training or a new system is unnecessary.
No. We agree what improvement means and how to measure it, then judge every change against your own baseline.
We agree the people, access and working sessions for your scope before we start. Our team runs the workshop and does the analysis.
No. If the need is already defined, we scope that work directly. An assessment helps when the constraint or the scope is unclear.
No. We work in the tools your teams already use and apply what we have learned to the way they already work.
No. Staff augmentation adds people to your backlog, and you direct their work. A forward deployed engineer takes on a defined problem, such as a review queue your value-stream assessment found, builds the fix with your team and coaches your people to run it. The work is measured against your baseline, and you choose whether we maintain what we build or hand it to your people.
Only if you choose to. We can maintain what we build, run it with your team, or prepare your people to take it over.
A senior coach spends 30 minutes with you on how work moves from request to release on your team and where it gets stuck. You'll hear what an assessment would look at first and roughly what it would cost. You can book it online.
Assessments start at $5,000. The price depends on how many teams take part and how far the work goes, from a map and plan to measured baselines. You can price an assessment online for up to three teams, and you approve the scope and the price before any work starts.
Reviewed by Chris Sims, Managing Partner. Updated November 2026.
Book 30 minutes with a senior coach. You'll talk through how work gets from request to release on your team and where it gets stuck.