Back To All

Waydev SDLC 30B, OLA Agent, Turing OTEL

November 21st, 2026
Topics
AI Agents
AI IMPACT
AI ROI
AI SDLC
Harness AI SDLC
SI SDLC
SI SDLC ROI
Share Article

/

Waydev Product update November 2026

A bigger, multilingual model. And two new ways to see delivery.

Three launches on one delivery line: Waydev SDLC 30B, OLA Agent and Turing

This week we shipped three launches, all live in Waydev today: a larger model that speaks your team’s language, an agent that sets delivery targets from your own data, and a way to know who wrote every commit, a person or an AI agent.

AC Alex Circei CEO and Co-Founder, Waydev

Each launch maps to one part of what we are building at Waydev: private agent intelligence for engineering organizations. Turing measures what AI agents actually write. The OLA Agent turns your delivery data into targets you can hold. And Waydev SDLC 30B is our own model, improving with every release, running on Cloud or inside your own network.

1 MODEL Waydev SDLC 30B Twice the parameters, now multilingual. 2 NEW AGENT OLA Agent Delivery targets Waydev suggests, then tracks. 3 OPEN SOURCE Turing Know who wrote every commit, a person or an AI agent.
Model upgrade

Waydev SDLC 30B. Twice the size, now multilingual.

Our own model, fine-tuned on software delivery data, grows from 15B to 30B parameters. Teams ask in the language they work in and get the answer, the chart and the SQL back in that language. It is the default in Waydev, on Cloud and on-prem.

A question in English and German goes into Waydev SDLC 30B and comes back as an answer, a chart and SQL
One model for global teams. Answers come back in the language of the question.
Answers you can check. Every answer shows the metric and the SQL behind it.
No outside AI vendors. On-prem, the model runs inside your own network.

Read the model docs →

New

OLA Agent. Delivery targets set from your own data.

OLAs (Operational Level Agreements) are targets for any delivery metric, for the whole organization, a team or an attribute. The OLA Agent proposes them from your recent baseline, so they are realistic from day one, then tracks every OLA you approve.

Lead time for changes falling from a 7 day baseline into a 5 day target suggested by the OLA Agent
1 Suggest. The OLA Agent proposes an OLA with the baseline, the reason and how to improve.
2 Approve. Adjust the target if needed, then approve it for the org, a team or an attribute.
3 Track. See which OLAs are met, which are not, and how far each one is from target.
4 Report. Weekly or monthly status to Slack, Microsoft Teams or email.
The OLA Agent suggesting an OLA to bring lead time from 7d 1h 6m to under 6d 8h, with how to improve and Approve

A suggestion: the baseline, the target, and how to get there.

Operational Level Agreements (OLA) Agent: 53.6% compliance, 15 of 28 OLAs met, organization compliance trend and metric cards

Compliance across every OLA, with met and not met at a glance.

Read the OLA docs →

New · Open source

Turing. Know who wrote every commit, a person or an AI agent.

The name is based on the Turing test: for every line in a commit, the Turing attribution engine knows whether a machine or a person wrote it. We built Turing to replace the previous, Entire-based setup, and minimize the amount of data collected and stored.

What one commit’s WD: trailer tells Waydev
Agentclaude-code
AI lines2
Human added1
Human modified1
Tokens165
AI share66.67%
An AI agent's OpenTelemetry spans lined up with the PR it merged and the deploy that followed
Open source. The code is public on GitHub, so your security team can review every line.
Every major coding agent. Claude Code, OpenCode, Codex CLI, Gemini CLI, Factory Droid, Copilot CLI and Cursor.
Line by line. Each commit gets AI lines, human added, modified and removed lines, and the tokens it took.
Private by design. Lines are stored as hashes, never as text. Only the numbers in the trailer leave the machine.
One installer. A single Node.js file with no dependencies sets up attribution and OpenTelemetry for Claude Code and OpenCode.
Safe switch from Entire. Compare mode runs both tools on the same commits so you can check the numbers first.
Turing showing coding agent runs as OpenTelemetry traces, linked to PRs and deploys

Agent runs as OpenTelemetry traces, linked to the PRs and deploys they produced.

Read the Turing docs → | View Turing on GitHub →

Why these three, together

AI agents now write a growing share of the code in every engineering organization. Leaders need to know three things: what the agents actually wrote, whether delivery is getting better because of it, and whether they can get answers about all of it without sending their data anywhere. Turing answers the first. The OLA Agent answers the second. Waydev SDLC 30B answers the third, in your team’s own language.

Measure the agents, set targets from your own baseline, and keep the intelligence inside your network.

Alex Circei, CEO, Waydev

For your security and procurement teams

DeploymentWaydev Cloud or on-prem
Outside AI vendorsNone. On-prem runs inside your network.
Commit attributionOpen source, lines stored as hashes
ComplianceSOC 3 report

See all three on your own data

A 30-minute walkthrough with our team.

Book a demo
Starting a new annual plan? Take 25% off your first year when you sign by December 1, 2026.* Claim the offer →

Read more

DOCSWaydev SDLC 30B → DOCSOLA Agent → DOCSTuring → PRODUCTWaydev Brain → GUIDEHow to measure AI adoption in engineering →

* 25% off the first year of a new Waydev annual plan signed by December 1, 2026. Cannot be combined with other offers. Talk to us for full terms.

Ready to unlock your SDLC productivity?

Request a Demo Call