Text Engineering
3 known investors
dottxt builds structured output technology for large language models, ensuring AI-generated outputs reliably conform to schemas like JSON or grammars. The company serves AI/ML teams, inference providers, and enterprises deploying production AI systems.
Also known as .txt Β· dottxt
Founders & leadership
Investors Β· 3
Company profile
researched Aug 2026Text Engineering, operating as .txt (dottxt.co), develops structured-output technology for large language models. Its core premise is that an application's schema should function as an enforced contract: rather than validating or retrying model output after the fact, the company's constrained-decoding technology makes the model produce output matching a specified shape by construction. Supported specifications include JSON Schema, regular expressions and context-free grammars. The company positions this reliability layer as spanning the full stack, from the tokens a model emits to agents composed from those calls and the specifications those agents fulfill, with the stated goal of making AI systems composable in the way conventional software is.
The product line has three delivery modes. api.dottxt.ai is a hosted platform offering constrained generation on a pay-per-token basis using recent open-source models. For teams that self-host, .txt offers drop-in replacements for inference servers those teams already run, including vLLM, SGLang and TensorRT-LLM, providing JSON, grammar-constrained and function-calling output without post-hoc validation overhead. For inference providers, it offers composable libraries: dotjson for JSON Schema enforcement, dotgrammar for arbitrary context-free grammars, and dotlambda for function calling, designed to integrate into existing inference pipelines so providers can expose structured outputs to their own users.
Business model
The company monetizes structured-generation technology through a hosted API billed per token, self-hosted drop-in inference-server replacements for enterprise teams, and licensable component libraries (dotjson, dotgrammar, dotlambda) for inference providers to embed in their stacks.
Pay-per-token usage pricing is stated for the api.dottxt.ai platform; other offerings are sold to self-hosting teams and inference providers.
βΈFull profile β market position, technology, go-to-market
Market position
The company presents itself as the group behind the widely downloaded open-source outlines project and cites external endorsement of structured generation from Julien Chaumond, cofounder and CTO of HuggingFace.
Emphasis on guaranteed schema conformance on every call rather than probabilistic prompting plus validation, delivered without the performance penalty of post-hoc validation and available across hosted, self-hosted and provider-embedded form factors.
Technology
Constrained decoding that guarantees per-call conformance of LLM output to a supplied specification β JSON Schema, regular expressions or context-free grammars β eliminating retries, validation loops and defensive parsing. The technology is packaged as dotjson (JSON Schema), dotgrammar (context-free grammars) and dotlambda (function calling), and as replacements for the vLLM, SGLang and TensorRT-LLM inference servers. The company states it is built by the team behind the open-source outlines library, cited at more than 65 million downloads.
Go-to-market
Self-serve API sign-up and quickstart documentation for developers, alongside demo requests and direct sales conversations for self-hosting teams and inference providers; the open-source outlines project and its download volume serve as a credibility and adoption channel.
AI engineering teams putting LLMs into production, organizations that self-host inference stacks such as vLLM, SGLang or TensorRT-LLM, and inference providers wanting to offer structured outputs to their own customers.
Compiled by commissioned research from 4 cited public sources β announcements, filings, and press listed under research sources below.
Key figures
latest reportedCompany-reported or press-reported figures, each dated to when it was claimed β not independently audited.
βΈResearch sources Β· 4
primary sources listed
- Text Engineeringdottxt.co Β· web
4 public sources were cited for this profile; the first-party ones are listed here.
Frequently asked questions
- What does Text Engineering do?
- dottxt (.txt) builds constrained-decoding technology that forces LLM output to conform to JSON Schema, regexes or grammars.
- Who founded Text Engineering?
- Text Engineering was founded by Julien Chaumond.
- Who are Text Engineering's investors?
- Text Engineering's investors include Kima Ventures, Elaia, Script Capital.