SpectraMind AI Private preview
In development · Private preview

The professionalSpectraMindPRO v2

Built to stay with the problem until the work is actually ready.

SpectraMind PRO v2 is our next professional model direction for multi-file software, connected applications and technical creative pipelines. It is being designed for work that has dependencies, tools, review cycles and consequences.

01Professional focusComplex technical work
02Tool-aware directionReview before action
03Not releasedCapabilities may change
SpectraMind PRO v2 model preview artwork with mathematical and spectral graphics
01Understand 02Structure 03Execute 04Verify
Inside PRO v2

The model brief

A professional model should carry the work, not only answer the prompt.

PRO v2 is being designed for assignments that unfold over time: a product with existing decisions, a repository with dependencies, a 3D scene with a real hierarchy or a connected application where every action has consequences.

01 · Why PRO v2

The difficult part of professional work often begins after the first answer.

The current SpectraMind AI Pro v1 is our strongest available model for demanding reasoning. It can analyse a difficult question, produce advanced code and work through complex material in depth. For many requests, that is exactly what a professional user needs.

But a technically convincing answer is not the same as a finished result. A change can be logically correct and still target the wrong file. A generated script can look complete while breaking an existing convention. A Blender automation can run successfully and still damage the scene hierarchy. The work becomes reliable only when the model understands the environment around the answer.

PRO v2 moves the product direction from answering a hard request to maintaining a dependable workflow. The intended model first identifies the relevant system, forms a scoped plan, uses an appropriate connected capability and stays with the task until there is evidence that the result works.

This is not an excuse to make every interaction slower or more complicated. Professional judgement also means recognising when a request is simple. Short questions should remain direct. Planning, tools and verification should appear only when they materially improve the result.

02 · How it should work

One objective, carried through a complete execution loop.

PRO v2 is being shaped around a persistent task state rather than a sequence of disconnected replies. The model should know what the user asked for, what it discovered, which decisions were made, what changed and what remains uncertain.

01

Understand the environment

Inspect the relevant files, dependencies, application state, project conventions and constraints before proposing changes.

02

Form a scoped plan

Separate assumptions from confirmed facts, identify risk points and make the intended outcome explicit.

03

Act through the right capability

Use code, research or a supported application only when it is needed, with access limited to the active task.

04

Verify the real result

Read build output, run tests, inspect a preview or compare scene state instead of treating generation as completion.

05

Leave a clear handoff

Explain what changed, what was validated, what remains open and how another person can continue the work.

The model should be able to recover when a check fails. A failed build is not the end of the workflow and should not be hidden behind a confident summary. It becomes new evidence: the plan is updated, the failure is traced and the next safe action is selected.

03 · Evaluation plan

We will measure completed work, not only impressive responses.

PRO v2 has not been released and we are not publishing benchmark claims for it today. Before presenting performance results, we want evaluations that resemble the work the product is meant to handle. Those evaluations need to test the complete path from initial request to verified outcome.

Multi-file reliability

Does the model find the correct implementation path, preserve conventions and avoid unrelated regressions?

Long-task continuity

Can it maintain requirements, decisions and unresolved questions across a longer sequence of work?

Tool judgement

Does it select the right capability, request confirmation at the right moment and avoid unnecessary actions?

Recovery quality

When a test or action fails, can the model diagnose the evidence and continue without restarting blindly?

Visual and 3D preservation

Can it change a scene while respecting hierarchy, scale, naming and non-destructive production practices?

Efficiency

Does the workflow produce a stronger result without using excessive context, unnecessary tool calls or repeated work?

ABOUT FUTURE RESULTS

When the evaluation suite and model behaviour are stable enough to be meaningful, we intend to publish results together with methodology, limitations and the configurations used. Until then, the capabilities on this page remain product direction rather than measured release claims.

04 · Safety and control

More agency requires clearer boundaries.

A model that can work with repositories and professional applications needs stronger product controls than a model that only returns text. PRO v2 is being designed around scoped access, visible intent and review gates for sensitive or irreversible actions.

The objective is not to interrupt normal work with constant confirmation prompts. Low-risk, reversible activity should remain fluid. Confirmation should appear where the consequence justifies it: sending information externally, changing protected resources, executing a destructive operation or extending access beyond the active task.

Scope
Only the context required for the task

Connected work should begin with the smallest useful set of files, data and application permissions.

Intent
Actions that can be understood before they run

The model should describe material changes and separate a proposal from an executed action.

Review
Human confirmation where consequences are higher

Sensitive steps remain behind explicit approval instead of being hidden inside an automated chain.

Evidence
A traceable result after execution

Outputs, failures and checks should be visible enough for a user to understand what actually happened.

These controls are still part of the development work. Their exact behaviour may vary by application, plan and risk level as the preview evolves.

01Architecture

Built for systems, not isolated snippets

Professional work rarely fits inside a single answer.

A difficult build has existing code, conventions, dependencies and a real runtime. PRO v2 is being designed to treat those constraints as part of the job, keeping a working map of the system while it plans, changes and verifies.

Repository intelligenceTarget workflow

Understand the codebase before changing it.

Trace the implementation path, inspect related modules and preserve project conventions before producing a patch.

Persistent task context

Hold requirements, decisions and unresolved questions together across a longer technical workflow.

  • Constraints remain visible
  • Decisions stay attributable
  • Recovery paths are explicit

Runtime-first validation

Move beyond plausible code by checking the build, reading failures and iterating on the real result.

verify --workspacePASS
02Mission control

One continuous execution loop

See the work move from intent to evidence.

PRO v2 is planned around a disciplined loop. Explore each stage to see what the model should preserve and produce before the next stage begins.

SPECTRAMIND / PRO_WORKSPACESIMULATED PREVIEW
SYSTEM MAP01 / 04

Build a reliable map before touching the code.

Inspect entry points, dependent files, project conventions and the checks that define success.

OUTPUTimplementation-map.md · 8 dependencies · 3 risk points
03Connected 3D

From prompt to production scene

A new direction for technical creative work.

We are exploring how PRO v2 could work with professional 3D applications through reviewed, scoped actions. Blender is the first reference workflow: the model reasons about the scene, proposes a plan and prepares Python automation without hiding what it intends to change.

PRO v2 ACTION PLANProduct lighting pass
  • Audit current scene scale
  • Create non-destructive light collection
  • Render preview before applying
SCENE

Scene intelligence

Inspect hierarchy, naming, scale, materials and dependencies before generating changes.

Planned workflow
PYTHON

Auditable automation

Prepare Blender Python that is scoped, readable and ready for review before execution.

Planned connector
RENDER

Visual checkpoints

Use preview renders and scene state to verify the result rather than assuming it worked.

Research direction
04Connected work

Capability with control

Applications become part of the workflow, not invisible shortcuts.

The product direction is to let PRO v2 choose the right supported tool while preserving clear boundaries. Sensitive or irreversible actions should remain reviewable and require confirmation.

Connected work is not meant to turn the model into an invisible macro. The model should understand what an application contains, explain the operation it proposes and use the application as part of a traceable workflow.

Different tools require different boundaries. Reading a build log is not equivalent to publishing a deployment; preparing a Blender script is not equivalent to applying it to a production scene. PRO v2 is intended to distinguish between observation, preparation and execution rather than treating every tool call as the same kind of action.

Initial research is focused on software work and technical creative pipelines. The final set of supported applications, permissions and access rules has not been announced.

01Scoped access

Only the tools and data needed for the active task.

02Visible intent

Explain what will change before an external action.

03Review gates

Keep sensitive actions behind explicit confirmation.

04Evidence

Return outputs, failures and validation, not just completion claims.

05PRO v1 and PRO v2

A different kind of upgrade

PRO v1 answers harder questions. PRO v2 is being built to run harder workflows.

PRO v2 is not simply a renamed version of the current model. It is a product direction built around system awareness, connected execution and verification. PRO v1 remains the available professional model while this next generation is developed and tested.

CapabilitySpectraMind AI Pro v1SpectraMind PRO v2
Primary role

Deep reasoning and advanced answers

Professional workflow and execution partner

Working context

Conversation, files and supplied material

Planned workspace, repository, scene and decision map

Software work

Generate, explain and debug sophisticated code

Inspect, change and verify multi-file systems as one task

Connected tools

Use supported tools within the conversation

Plan and coordinate reviewed actions across supported applications

3D workflows

Reason about scenes and produce scripts or guidance

Planned scene inspection, auditable Python actions and render checkpoints

Verification

Review the generated answer and supporting reasoning

Use runtime, test and visual evidence as part of the execution loop

Availability

Available now · Pro plan

In development · No launch date

What should feel different

Instead of repeatedly asking the user to carry context between steps, PRO v2 is intended to preserve the plan, understand what changed and decide what must be checked next. The experience should feel closer to handing a task to a careful technical collaborator.

What will remain familiar

The SpectraMind identity, direct communication and user control remain central. The model should still adapt to the person and the task, while exposing more of the structure needed for serious professional work.

06Product direction

Why a second generation of Pro

More depth is useful. More ownership is the real upgrade.

Professional users do not only need a stronger answer. They need a model that can enter an unfamiliar system carefully, maintain a plan, use the correct application and prove that the result works.

PRO v2 is being shaped around that responsibility. The intended experience is closer to a technical collaborator: one that asks when a decision is genuinely blocked, moves independently when the path is safe and leaves the work in a state another person can understand.

This page is a preview of that direction, not a release promise. The final model, supported applications, access rules and capabilities may change as testing progresses.

AVAILABLE
SpectraMind AI Pro v1

The current professional model for deep reasoning and demanding technical requests.

DEVELOPMENT
Workflow foundations

System context, planned execution, connected applications and verification loops.

NOT ANNOUNCED
Launch date and final access

Model configuration, supported connectors, plans and pricing may change before release.

07 · Availability

PRO v1 remains available. PRO v2 does not have a launch date.

SpectraMind AI Pro v1 remains the professional model available to Pro members. PRO v2 is a separate generation in development, and the page describes the direction being explored rather than functionality that customers can enable today.

A future preview would begin with limited workflows where permissions, verification and recovery can be observed closely. Wider access would depend on reliability, safety and whether the model delivers a meaningful improvement over the current Pro experience.

PRO v1Available nowPRO v2In development
Private preview

The next professional SpectraMind is being built now.

Follow the development of PRO v2 and tell us which technical or creative workflow should be tested next.

No launch date has been announced.