Skip to content

Frequently Asked Questions#

The platform#

What is LIT?#

LIT is the workplace for AI agents. The labs make the models smarter; LIT gives your AI a seat on your team, on your machines, with rules and a record. A seat is a team-held credential a member uses. Rules are per-team policy over providers, models, tools and spend. The record is the channels, transcripts and audit log that belong to you rather than to the AI vendor. One workplace, three doors: a server IT governs, a desktop engineers work on, and a phone managers supervise from. Platform overview →

Do I need to be an engineer?#

No. The member experience is: sign in, find a seat and a provisioned agent already waiting, complete one OAuth with your AI provider, and start talking to your agent in a channel. Engineers get more (local files, local models, the desktop), and admins get the policy and audit tabs, but a member never touches a terminal.

Which models and harnesses does it run?#

LIT spawns the vendor's own harness, Claude Code, Codex CLI or Gemini CLI, rather than reimplementing the loop around the model. Direct API access to Anthropic and Google is there too, and local models run through Ollama. Policy can allow or deny by provider, by model, or by model family, and you can switch models without losing the record. Any harness, any model →

Subscription or API key?#

Both, side by side. A Claude Max or Team subscription seat is held by the team and used by a member through Claude Code, the way the vendor intends. A pay-per-token API key sits next to it in the same credential pool, under the same policy, writing to the same record. Teams typically start on the seats they already pay for and add tokens when they're ready.

Is it self-hosted?#

Yes. Everything runs on your machine, or on a partner's infrastructure on your behalf. Sessions, channels, transcripts, memory and credentials live under ~/.config/lit/ per user, and nothing is sent to LIT.AI servers. On team deployments, tools run as the authenticated user over SSH with that person's own permissions, and every action is on the audit log. Identity & Security →

How does LIT handle data security?#

Self-hosted deployment keeps data on your infrastructure. Single sign-on comes from your existing identity provider through Keycloak, with owner, admin, member and group roles. Team policy governs which providers, models, MCP servers and paths an agent may touch, and deny always wins. Every action is written to the audit log and every conversation to a transcript. Seats & Policy → · Security reference →

Do you support self-hosted LLMs?#

Yes. LIT integrates with Ollama, so a local model can take mechanical work while a frontier model takes the hard reasoning, in the same channel.

What happens to a person's channels when they leave?#

They go to a successor. When an admin offboards someone, or the identity provider removes them, their seat closes and their channels and transcripts are assigned to a named colleague instead of vanishing with the account. Retention holds and transcript grants cover the cases where a record needs to outlive both people. Channels & the Record →

Buying and licensing#

How do I buy it?#

Teams buy through a partner. Partners deliver the platform under their brand, administer your seats and policy, connect your identity provider, and support you directly; Positronic builds the platform behind them. Partners →

Individuals don't buy it. LIT Desktop and the pip install are free for personal and non-commercial use, and they are the same workplace, for one person.

How much does it cost?#

See the Pricing page. Platform licensing for teams is arranged through partners; partner licensing is arranged directly with us. Professional services are priced on the same page.

We deploy AI for clients. Can we become a partner?#

That is who partners are: fractional CTOs, consultancies, agencies and managed service providers who are accountable for what AI does in a customer's environment. Partners own the accounts they bring, and Positronic never sells direct into them. Become a partner →

Can I run LIT on my own infrastructure?#

Yes. On-premises servers, your own cloud account (there is an AWS Marketplace image), Docker, or a single machine with pip install. Installation →

Services#

What is a Fractional CTO service?#

Strategic technology leadership on a part-time basis, for companies that need senior AI leadership but not a full-time executive. Our fractional CTO work covers AI strategy, architecture decisions, team building, and hands-on implementation guidance. Fractional CTO →

How does Vibe Coding work?#

Vibe Coding is our methodology where senior engineers provide intent and direction while AI agents handle implementation. In practice it compresses timelines: we routinely complete in days what traditionally takes weeks. See the sprint where two complete applications were built in fourteen hours.

Do you work with existing AI infrastructure?#

Yes. LIT runs alongside whatever you already have. The harnesses and models are the vendors' own; the identity provider is yours; MCP connects the tools your team already uses.

What if we need custom development?#

Our Professional Services team builds production AI systems: agentic applications, skills and apps on the platform, real-time data systems, and integration with what you already run.

Still have questions?#