In an example published by PTC, a save button in a French interface lost its final letter. The text had been translated, but it did not fit. The tool identified a code issue and prepared an instruction to fix it; the change still had to reach the product. That step helps explain what software localization actually sells.
Private Translation Cloud, from OnTheGoSystems, proposes moving translations into the same process a team uses to review software changes. Its growth opportunity depends on reducing the work between a translated string and a version that a customer can use. Counting words processed leaves out the part that determines whether a product reaches another market.
An unfinished translation also consumes development capacity
The Git integration connects language files to GitHub, GitLab or Bitbucket repositories. PTC detects changes on selected branches and returns translations through pull or merge requests. The team retains the decision to review and accept them. This can avoid exporting and reimporting files as a separate task.
Potential savings arise in handovers between people: asking for the correct file, identifying its version, retrieving the translation and checking that it belongs to the change. Automating those steps can release time. The amount cannot be calculated without observing a particular customer’s workflow, and an accepted translation does not create revenue by itself.
PTC is not the only supplier working within the repository. Lokalise’s documentation describes GitHub actions that upload text and return translations through pull requests. That alternative requires PTC to demonstrate the quality and cost of the complete process, rather than presenting a Git connection as an exclusive advantage.
Visual review needs the correct version
PTC’s visual review documentation uses the French button as its own example. For a defect requiring a code change, it generates an instruction for the coding assistant. Detecting the problem and having it corrected in the application are different states, with different owners and timings.
The same documentation acknowledges that a screenshot from an older version may not match the current translations. The most informative commercial demonstration needs the version the buyer intends to release. Reviewing old screens with convincing results may leave the very changes awaiting release unexamined.
The W3C distinguishes localization from translation alone : adapting a product can also affect date and number formats or how the interface is constructed. For a Spanish buyer, a useful evaluation does not end when every string is in Spanish. It ends when the chosen user journeys retain their meaning, readability and operation.
Stack Overflow’s 2026 survey shows conditional trust: 48% of those answering selected trust when they can easily verify the output. It is not an evaluation of translations or PTC. For the sale, the implication is concrete: a promise of automation needs to show what result the buyer can inspect and which work remains the buyer’s responsibility.
48% make trust conditional on being able to verify the answer
Reported trust in AI or agent output within the workflow, Stack Overflow’s international 2026 survey. Single response.
Tap or hover for details. All values are also available in the data table.
Collapse notes
Optional question with 14,304 responses, 23 June to 5 August 2026. Sample recruited through the Stack Overflow community; not an evaluation of translations or PTC. The question and options differ from 2025, so no trend is presented.
Explore the data table
| Category | Respondents |
|---|---|
| When the answer can be verified | 48.0% |
| Many tasks, excluding important decisions | 16.3% |
| Low-risk tasks only | 16.3% |
| Distrust for most tasks | 12.3% |
| Including important decisions | 6.6% |
| Uncertain or no AI use | 0.5% |
Repository access is part of the purchase
The specific GitHub guide requests read and write permissions for contents, pull requests and webhooks, and recommends selecting specific repositories. Approval of that access can matter as much to adoption as the quality of a sentence. A demonstration account without the customer’s restrictions does not reveal this onboarding cost.
There is also an API route , with credentials restricted to a project. It changes the integration method, but does not remove the need to decide which files are sent and who can access them. Choosing the appropriate route makes it possible to evaluate the service under the conditions it will face when handling real content.
PTC’s terms state that it retains resource files until the customer deletes them. They also specify that it does not change other parts of the repository unrelated to translation. Those conditions define a buying decision: what information can be shared and how it is withdrawn when the service is no longer used.
For teams adopting AI, barriers are not merely individual. Stack Overflow’s 2024 professional developer survey separates a lack of codebase context from policies and training. These are multiple responses from an international sample; they do not describe PTC’s customers. They help explain why better translation, alone, may not unblock an integration.
Codebase context outweighs a lack of training
Barriers to using coding assistants or generative AI within the team, 2024 professional developer subgroup. Three selected categories; multiple responses.
Tap or hover for details. All values are also available in the data table.
Collapse notes
24,496 professional developers, surveyed 19 May to 20 June 2024. Each person could select several barriers. Community sample, not Spanish businesses or PTC customers. It is not compared as a trend with the 2026 trust question.
Explore the data table
| Category | Respondents |
|---|---|
| Missing codebase context | 64.6% |
| Inadequate policies | 31.9% |
| Inadequate training | 29.6% |
The pricing table and terms describe different trials
The current pricing table separates a trial of up to 20,000 words and two languages from Pro, excluding Git, API and visual review from the trial. The terms of use , however, promise access to all features for 30 days. Both documents were available when consulted; they do not resolve that difference.
This inconsistency in the documents affects the step PTC needs to demonstrate. Someone uploading a file tests an engine; someone connecting a repository evaluates a release process. The buyer needs to establish the trial’s scope before committing work to that integration. The advertised usage-based structure, without a fixed monthly subscription, does not clarify that scope by itself.
The cost that determines continued use combines consumption, integration and unfinished work. An evaluation can observe how long an update takes to become ready in the chosen languages, which corrections it needs and who carries them out. These are evaluation criteria, not measured PTC results: its public documents do not establish cost reductions or sales increases for a particular customer.
For OnTheGoSystems, PTC’s growth can rest on selling a completed task with clear boundaries. The supplier provides translation, change tracking and review; the buyer retains approval of what enters the product. Savings become defensible when the next version reaches users with less unfinished work, rather than when the word counter rises.
In this article
Marian Vasile
Marian Vasile is a MarTech and growth professional. His work spans product marketing, SEO, paid advertising and deep online tracking, including behavioral analytics and attribution.
