Skip to content
Falcon Nest

Land on an unfamiliar codebase and be useful by lunchtime.

Falcon FDE Solo

A single-seat forward deployed engineering assistant. It reads the whole repository, answers architectural questions with citations, drafts integration code, and writes the handover documents nobody enjoys writing.

  • 1 engineer seat
  • Up to 3 connected repositories
  • Answers cite file and line
  • Runs against private code
Choose a plan

Licence keys are issued the moment payment clears. Prices exclude VAT, which is calculated at checkout from your billing country.

Overview

A forward deployed engineer's first week is mostly archaeology: which service owns this table, why does the retry live here, what breaks if this flag flips. Falcon FDE Solo does that archaeology in minutes rather than days.

Point it at a repository and it builds a structural map — modules, call graphs, data models, configuration surfaces and the seams where integrations attach. Ask a question in plain language and you get an answer with file-and-line citations, so you can verify the claim instead of trusting it. Ask for an integration and you get a diff that follows the conventions already present in that codebase, not generic boilerplate.

Solo also handles the artefacts that surround delivery work. Runbooks, sequence diagrams, migration plans, API reference pages and customer-facing summaries are generated from the actual code and kept in step with it as the branch moves.

  • Whole-repository understanding, with citations you can check
  • Draft integrations that match the existing conventions
  • Delivery documents generated from code, not from memory

Capabilities

  • Whole-repository comprehension

    Indexes structure, call graphs, data models and configuration rather than just chunks of text, so questions about architecture get architectural answers.

  • Citations by default

    Every factual claim about the codebase links to the file and line it came from. If it cannot cite, it says so rather than guessing.

  • Convention-aware diffs

    Generated code follows the patterns, error handling and naming already in the repository, and arrives as a reviewable diff rather than a wall of text.

  • Delivery artefacts

    Runbooks, sequence diagrams, migration plans and integration guides generated from the code and regenerated when the code moves.

  • CLI and editor access

    Use it from the terminal, from your editor, or from the web console. The same index and the same context back all three.

  • Your code stays yours

    Repositories are indexed into an isolated per-tenant store, never used to train shared models, and deletable on demand with a verifiable purge.

Where it earns its keep

  • New engagement ramp-up

    Walk into a client codebase you have never seen and produce a credible architecture summary and integration plan on day one.

  • Integration builds

    Draft the adapter, the retry policy and the tests against the client's actual interfaces rather than against the documentation they forgot to update.

  • Handover documentation

    Produce the runbook and architecture decision records at the end of an engagement without spending the last three days writing prose.

Questions

Is my source code used to train models?
No. Your repositories are indexed into a store that is isolated to your tenant and used only to answer your queries. Nothing from your code enters shared training data, and you can trigger a verifiable purge at any time.
What counts as a task credit?
One substantial unit of work: an architectural question with citations, a generated diff, or a produced document. Follow-up questions within the same thread do not consume additional credits.
How large a repository can it handle?
Up to 500,000 lines per repository on this tier. Larger monorepos are supported on Team and Enterprise, which also allow scoping the index to specific paths.
Does it write code directly to my repository?
Only if you ask it to open a branch, and only with credentials you grant. By default it produces diffs for you to review and apply yourself.
Can I use it on client code under NDA?
Yes, and this is the common case. Data is processed in your selected region, is not retained beyond your configured window, and our data processing agreement is available for your client's review.