bright-bright is a private, dark AI workspace for doing business work in one sandbox with an all-in-one AI assistant, agent, and bots. The workspace supports memory, drafting, coding, and open module management, alongside bot scaling, competition entry, and visibility into sales and income, including crypto and bitcoin.
This document is the requested system requirements document for bright-bright. Its audience is the Workspace Owner and Bot Operator who use and modify the private workspace.
bright-bright is a private AI workspace with a custom React interface and a shared FastAPI backend. Its accepted human users are the Workspace Owner and Bot Operator. The owner uses the sandbox for business work with an AI assistant, agent, and bots; the operator’s accepted work centers on scaling bots and modifying open modules. The product includes the named areas and workflows in this document, without assuming unspecified business capabilities.
The workspace is private and uses application-owned identity. The current page contract includes an anonymous Landing and Login, followed by protected workspace pages. The product includes the free-tier capabilities referenced by the user, but no specific free-tier feature list or tier entitlements were supplied. The three stated prices are preserved as values; no billing, payment, subscription, or tier-enforcement workflow is specified.
The current delivery is a private, application-owned workspace. A user can reach Landing anonymously, establish identity through self-service enrollment before first use, and use Login to verify identity on return. Protected workspace pages require login. Identity supports private ownership and continuity; it does not imply differentiated permissions between the accepted personas.
The supplied page contract is the current information architecture: Landing, Login, Sandbox, memory, drafting, coding, module management, Bots, Competitions, and Revenue. The shared backend supports application APIs and domain modules. The provisioned infrastructure is backend (FastAPI), database (MariaDB), frontend (React), litellm (LiteLLM gateway), postgres (PostgreSQL supporting LiteLLM), and redis (Redis for the accepted shared caching, state, queue, or coordination need). PostgreSQL is separate from application-owned storage. No additional service is specified.
The product supports the stated business-work, bot, competition, revenue, and open-module scope. It does not promise outcomes, invent free-tier entitlements, or add adjacent business, financial, competition, or account-management capabilities.
As a Workspace Owner, I should be able to enter the public bright-bright workspace entry and proceed toward my private workspace.
Provenance: required_inference for the anonymous entry and transition; the product and private workspace are explicit.
Lifecycle and access: From Landing, the owner selects “Enter workspace” and reaches Login. Landing is anonymous; no private state is shown. If navigation fails, the owner can retry from Landing.
Acceptance: Landing presents the supplied hero meaning and an entry action that leads to Login.
As a Workspace Owner, I should be able to establish application identity before first use and verify identity on return.
Provenance: required_inference.
Lifecycle and access: The owner completes self-service enrollment before first use, or verifies identity on return through Login. Successful verification opens the protected Sandbox. Failed enrollment or verification leaves protected state unavailable and permits retry.
Acceptance: First use cannot enter protected pages without identity establishment; returning access requires verification. No additional account-management capability is implied.
As a Workspace Owner, I should be able to do business work from one sandbox with an all-in-one AI assistant, agent, and bot.
Provenance: explicit.
Lifecycle and access: From the protected Sandbox, the owner initiates business work with the assistant, agent, or bot and observes the interaction’s result or an explicit failure. The owner can continue in the Sandbox or move to another accepted workspace area.
Acceptance: The Sandbox provides the accepted assistant, agent, and bot business-work context. It must not claim unsupported task coverage or successful results when an interaction fails.
As a Workspace Owner, I should be able to use a distinct memory area for durable workspace memory.
Provenance: The named area and durable memory are explicit in Planning Scope; the minimum observable loading, empty, and recovery states are required_inference.
Lifecycle and access: From a verified session, the owner opens memory and sees available memory or an empty state. A failed load or save is reported and can be retried; the owner can return to the Sandbox or another accepted area.
Acceptance: memory is a distinct protected destination and represents durable owner-associated memory without inventing unspecified memory operations.
As a Workspace Owner, I should be able to use a distinct drafting area for drafts.
Provenance: The named area and drafts are explicit in Planning Scope; minimum observable states and recovery are required_inference.
Lifecycle and access: From a verified session, the owner opens drafting and works with drafts. Loading or saving failure is visible; retry is available where applicable, and the owner can continue to another accepted area.
Acceptance: drafting is a distinct protected destination. No sending, publishing, or specific draft schema is implied.
As a Workspace Owner, I should be able to use a distinct coding area for code work.
Provenance: The named area and code work are explicit in Planning Scope; minimum observable states and recovery are required_inference.
Lifecycle and access: From a verified session, the owner opens coding and works with code. Loading or saving failure is visible; retry is available where applicable, and the owner can continue to another accepted area.
Acceptance: coding is a distinct protected destination. No runtime, language, or deployment behavior is implied.
As a Workspace Owner, I should be able to modify and add brain, core, weights, and nodes.
Provenance: explicit.
Lifecycle and access: From a verified session, the owner opens module management, modifies or adds one of the named elements, and observes whether the change was applied. On failure, the last confirmed state remains authoritative and the owner can correct or retry. The owner can continue to another accepted area.
Acceptance: brain, core, weights, and nodes remain open to modification and addition; the system does not falsely report an unconfirmed change as successful.
As a Workspace Owner, I should be able to work with 4 to 7 bots to scale business work.
Provenance: The bot range and scaling goal are explicit; the Bots destination and minimum observable state are required_inference.
Lifecycle and access: From a verified session, the owner opens Bots and reviews or works with the bot set. The page represents the stated 4–7 range and the three price values. If information is unavailable, the owner sees an error and can retry or continue elsewhere.
Acceptance: The product supports the stated 4–7 bot scope without inventing bot deployment, billing, or tier entitlements.
As a Workspace Owner, I should be able to use the three stated price tiers: $150, $500, and $2,000.
Provenance: explicit.
Lifecycle and access: From a verified session, the owner can see the three price values in the bot-scaling context. If tier information cannot be loaded, the page reports that failure and permits retry.
Acceptance: The values are preserved exactly. No currency period, included features, payment, subscription, or enforcement behavior is inferred.
As a Workspace Owner, I should be able to enter Kaggle and other competitions.
Provenance: Competition entry and Kaggle are explicit; the Competitions destination and minimum observable recovery are required_inference.
Lifecycle and access: From a verified session, the owner initiates competition-entry work for Kaggle or another competition and sees a confirmed result or an explicit unavailable/failed state. The owner can retry where applicable or continue elsewhere.
Acceptance: The product supports competition-entry work without claiming a submission was accepted or a competition was won. No external account or submission mechanism is assumed.
As a Workspace Owner, I should be able to view sales and income in real time, including crypto and bitcoin.
Provenance: explicit; the Revenue destination and minimum freshness/error states are required_inference.
Lifecycle and access: From a verified session, the owner opens Revenue and reviews available sales and income information, including crypto and bitcoin. The page communicates unavailable or stale information and allows retry.
Acceptance: The product does not promise income or gains and does not present unavailable or unconfirmed information as real-time results.
As a Workspace Owner, I should be able to use anything available on the free tier.
Provenance: explicit.
Lifecycle and access: In the private workspace, the owner can use the capabilities available on the free tier. If a free-tier capability is unavailable, the system must not represent it as available.
Acceptance: The free-tier inclusion is preserved, but no specific feature list, entitlement, or tier comparison is invented.
required_inference, as supplied by the closed Planning Scope persona catalog.required_inference, as supplied by the closed Planning Scope persona catalog.The authoritative creative direction names Gleb Kuznetsov as the muse and calls for a cinematic, dark, polished private AI workspace that feels capable, open to change, and controlled. The supplied screenshots are optional visual context only; they do not prescribe page structure, navigation, or additional product capabilities.
#080B16; raised surface #11182B; text #EAF1FF; primary #18234A; accent #00D9F5; muted text #8192B2.The public Landing hero pairs the headline “One workspace. More room to build.” on the left with a large, layered dark workspace preview on the right. The preview recomposes only accepted content: compact memory, drafting, coding, and module-management panels, with a narrow cyan status rail on the active area and readable configuration rows. A cyan “Enter workspace” action anchors the composition. On mobile, the preview stacks below the headline. The concept communicates a private, editable workspace without implying unaccepted features or guaranteed business outcomes.
Interaction Model: Animated
Motion Tempo: restrained
Hero Dimensionality: layered_2d
explicit constraint; identity mechanics required_inference: Protected workspace state is unavailable until identity is established and verified. Landing and Login remain anonymously reachable. Returning users verify identity before entering protected pages.explicit Planning Scope requirement: The system must support durable application-owned workspace state for accepted durable content, including memory. Do not infer additional data categories or operations.explicit: Brain, core, weights, and nodes remain open for user modification and addition; the system must not impose a closed structure that prevents this accepted work.required_inference: Revenue information must communicate freshness and distinguish unavailable or stale data from confirmed current information. This supports the stated real-time visibility goal without promising gains.required_inference: For accepted stateful interactions, show whether the operation succeeded or failed; do not represent unconfirmed changes, competition entry, or revenue information as successful or current.litellm) for backend model-provider calls.postgres) supporting LiteLLM, separate from application-owned storage.redis) for the accepted shared caching, state, queue, or coordination need.No additional service is specified. The provisioned infrastructure does not itself establish product requirements, domain entities, authentication behavior, or additional workflows.

Private AI workspace
Run business work with an AI assistant, agents, bots and open modules in your private workspace.
// work in progress
const workspace = open;
Private AI workspace
Run business work with an AI assistant, agents, bots and open modules in your private workspace.
// work in progress
const workspace = open;
No comments yet. Be the first!