The following page may contain information related to upcoming products, features and functionality. It is important to note that the information presented is for informational purposes only, so please do not rely on the information for purchasing or planning purposes. Just like with all projects, the items mentioned on the page are subject to change or delay, and the development, release, and timing of any products, features or functionality remain at the sole discretion of GitLab Inc.
| Stage | Create |
| Group | Code Creation |
| Maturity | minimal |
| Content Last Reviewed | 2024-01-30 |
Thanks for visiting this category direction page on Code suggestions in GitLab. This page belongs to the Code Creation group of the Create stage and is maintained by Kevin Chu (kchu at gitlab dot com).
AI Coding Assistants use Generative AI to suggest relevant code snippets and autocomplete code as software developers type. Coding assistants aim to boost programmer productivity and reduce time spent on rote coding tasks. Some key points:
GitLab Duo Code Suggestions, part of the comprehensive AI-powered DevSecOps platform, empower increased coding productivity, without sacrificing security, privacy, and enterprise control.
GitLab Duo Code Suggestions is part of the larger suite of GitLab Duo, GitLab's AI-powered features and capabilities. The AI Coding Assistant category is among the most obvious (even if it is not the biggest SDLC efficiency driver) and mature among AI categories. By first empowering customers to adopt code suggestions, we can then jointly prioritize and augment the remaining DevSecOps workflow with AI, helping organizations to be fundamentally more efficient in bringing ideas from inception to production.
Having brought Code Suggestions to GA on 2023-12-21, our short-term (3 months) goal is to support the launch of GitLab Duo Pro, improve the fit and finish of the product, and bring critical enterprise controls online.
We need to continue to move quickly, as customers are continuing to actively invest in bringing AI into their software development workflows. We risk alienating customers who are looking to adopt AI if the quality of code suggestions lag behind other available products. By maturing code suggestions and bringing online other GitLab Duo capabilities, we can mitigate this risk while collaborating with customers to drive towards improved overall productivity.
Longer-term, we plan to position code suggestions, along with the rest of our AI offerings, to differentiate on:
Traditionally the creation of code occurs in code editors or IDEs. However in its essence this is a conceptual work performed by trained individuals. By bringing code creation capabilities to other stages of software development life cycle (for example: to CI failed pipelines, or a vulnerability detected with a security scan), we aim to make creating code as easy as possible, empower more members to contribute and redefine perception of code creation as it exists Today.
Managing feature access at scale is an important problem to solve for enterprises. Furthermore, providing visibility into how code suggestions is used, how code suggestion impacts overall productivity will help GitLab customers make data-informed decisions.
Furthermore, Code is enterprise IP. How we enable privacy without compromise even as new AI-powered features become available is a top concern for organizations. Particularly for some GitLab self-manged or dedicated customers, having to send code to 3rd party models may not be a tenable proposition long term. We plan to investigate available options to retain complete privacy, including for air-gapped customers.
The world of Gen AI Models, LLMs, SLMs, is quickly evolving. A flexible design is at the core of GitLab Duo, it enables us to follow dynamically evolving landscape of AI tooling and select optimal models for a given use case. We plan to be granular such that we can use different models by language, deployment type, customization need for code suggestions.
People who code:
In the future, we may expand to include security personas to help write more secure code and review code for security vulnerabilities and fix them early in the software development lifecycle (SDLC), before you commit.
| Theme | Metric | Description | Tactics/Epics | Timing |
|---|---|---|---|---|
| Enterprise Control | N/A | Enable Enterprise to Operate Code Suggestions with the appropriate controls | AI Ban, Enterprise Seat Provisioning, Self-Service Purchasing, Improved User Provisioning | FY25-Q1 |
| Usability | % of errors, Acceptance Rate, user growth | Fix and improve UX issues | Improved Language Support, Reduce error requests, Streaming across all IDEs | FY25-Q1, FY25-Q2 |
| Trigger Logic | Shown Rate, Acceptance Rate | Improve when code suggestions are requested | Improve Trigger Logic, Reduce code suggestion requests | FY25-Q1, FY25-Q2 |
| Decrease latency | Load Time, Acceptance Rate | Decrease the latency between when a suggestion is requested and shown to the user | Code Suggestions Performance Improvements,Architecture Update | FY25-Q1, FY25-Q2 |
| Increase SM/Dedicated upgrade cycle | Minimize the frequency GitLab instances have to be updated in order to get improvements in Code Suggestions and other Duo features | Decouple Code Suggestion Improvements from GitLab Updates | FY25-Q3 | |
| Increased context | Acceptance Rate | Increase the context so that more relevant results are returned | RAG | FY25 |
| ROI Proof | Establish credible and meaningful metric(s) that customers can access directly | ROI Proof | FY25 |
This category is currently at minimal maturity.
Please see the content in our internal handbook
