By the time an engineer at Modelit opens a Jira ticket, Claude has already investigated the codebase, scoped the change, and planned the work. By the time a reviewer sees a pull-request, the development changes have been repeatedly challenged and independently graded via multiple adversarial reviews. These changes are then mapped to end-to-end test suites responsible for testing the specific code being shipped.
This is Caidence, the AI delivery framework PeerNova built with implementation partner Modelit to develop Cuneiform, PeerNova's Salesforce-native applied data intelligence product. After the Cuneiform team adopted it in March 2026, delivery volume doubled, and velocity increased by 43%.
With Claude, PeerNova was able to:
- Double delivery volume, from 637 to 1,286 completed tickets across matched 170-day periods
- Boost team velocity, shipping tickets 2.2 days faster at the median
- Ship changes after automating regression tests bound to the specific code being released
The challenge
Prompts made individuals faster. Delivery stayed flat.
By early 2026, PeerNova was no longer asking whether AI could help engineers write code faster. The team was already using Claude and seeing individual productivity gains. The harder question was why those gains were not translating into faster, more reliable software delivery.
The team started with the obvious lever: better instructions. Engineers refined prompts, documented standards, and carried lessons from one session into the next. Claude became better at producing the work, but the surrounding delivery process stayed largely the same.
The bottleneck was not writing code. It was everything required to trust the result: reviewing changes, checking that acceptance criteria were defined well and actually met, finding missing tests and regressions, and sending incomplete work back for another pass.
Instructions could guide Claude, but they could not guarantee that work was performed consistently, prevent a prohibited action, verify the final result, or preserve every lesson across sessions.
PeerNova and Modelit concluded that improving prompts would not be enough. They needed to turn the delivery process itself into a software product that enabled flow, comprehension, traceability, and delivery observability. They wanted to create a system that engineers could learn how to work with and manage AI.
“We were having inconsistent experiences with AI delivery that varied too much across engineers. Some engineers were doing really interesting things; others were too tentative. We decided to create an internal product that would scope, build, test, and deploy code, following human ceremonies in an observable way. We needed to experience and learn AI delivery together.”
Abraham Lloyd
CTO at PeerNova Inc.
The solution
Turning delivery into an AI-driven pipeline
PeerNova and Modelit built Caidence around Claude so the model would operate inside a defined delivery system rather than as a standalone coding assistant.
A ticket begins with one command. Claude starts with a framed goal from a human, investigates the codebase, and builds a scope that takes the shape of a task, story, bug, or collection. The analysis repeats until a complete pass surfaces nothing new. Caidence then decomposes the work into subtasks and Claude implements them individually, on isolated branches, with commits tied back to the acceptance criteria they satisfy.
The goal is not simply to generate code faster. It is to carry the work from desired intent through to proven outcome using the same repeatable, observable, and audited process every time.
Verification became part of the work
A validation loop searches for issues such as missing tests, dead code, unhandled errors, and acceptance criteria that were not 100% met. The loop can run for up to ten rounds and stops only after two consecutive passes find nothing new. When it identifies a problem, the finding enters a fix loop and validation begins again.
Caidence separates execution from judgment. Claude Opus handles well-defined implementation work, while Claude Fable handles higher-judgment tasks such as scoping and adversarial review. Every dispatched agent declares its model explicitly so routing is part of the framework rather than an informal convention.
Another convention Caidence introduced is that the model and session producing the work are not allowed to judge it. Caidence leverages a multi-model panel that includes independent models from other providers and persons to grade work. Any work below an A grade returns to the remediation loop.
“Working with AI introduced some really interesting conversations about quality and consistency. What we found was most effective was this idea of "convergence" where we'd align different models, personas, and perspectives to review work. We would allow review cycles to continue until all parties (managing their own independent context and sessions) met our delivery minimum A grade.”
Abraham Lloyd
CTO at PeerNova Inc.
"Done" became something the system proves
Once validation and grading pass, Caidence creates review evidence, identifies the end-to-end tests that are impacted, and updates the pull request and Jira ticket in preparation for the release.
If the evidence is missing, invalid, forged, or belongs to an earlier version of a change then the pull request is blocked with feedback explaining what is preventing the merge.
That changed the role of review. Instead of asking a person to reconstruct whether every delivery step happened, the pipeline produces evidence that those steps ran against the code being shipped.
"Done" became something the system proves rather than something anyone declares.
Guardrails enforce delivery principles
PeerNova and Modelit also separated rules that can be enforced from guidance that still requires judgment.
Delivery rules that can be deterministically enforced became hooks. If Claude or an engineer attempts a prohibited action, such as committing directly to a protected branch or bypassing a required code check, the framework blocks it. The denial also explains what must happen before the action can proceed.
Rules that cannot be deterministically enforced remain written guidance and are identified as such.
Around the pipeline sits persistent project memory and live connections to systems including Jira, GitHub, and Salesforce. Claude can work from current project state and accumulated knowledge rather than relying on whatever context happens to fit inside one session.
The Cuneiform team adopted the framework on March 11, 2026. Repository history shows the transition clearly: commits following Caidence conventions went from zero in February to complete adoption by April.
The outcome
Twice the delivery
Across matched 170-day periods before and after Caidence adoption, the Cuneiform team increased completed tickets from 637 to 1,286.
The result was not simply more AI-generated code. Verification that previously depended on people reconstructing each change now runs inside the delivery pipeline. Engineers can run multiple workstreams while the framework continuously checks whether each one meets the team's standards.
“The most important improvements came from constant experimentation and evolution. Our speed, quality, and concurrency gains all came from measuring delivery patterns and artifacts, constantly questioning the outcomes we generated, and staying in-touch with new model capabilities. Our biggest improvements came out of this collaboration model.”
Abraham Lloyd
CTO at PeerNova Inc.
Caidence has since expanded beyond code delivery. PeerNova and Modelit are applying the same approach to knowledge management, product documentation, content creation, and backlog management. Using the ever-evolving set of capabilities that Anthropic provides, coupled with the shared learnings of the Modelit and PeerNova partnership.
For PeerNova, the shift moved from using AI to accelerate code development to building a delivery system powered by Anthropic models that could deliver the efficiency gains AI promises while unlocking the creativity and experimentation needed to reimagine work.
Modelit is PeerNova's implementation partner for Cuneiform.
