--- title: NeuralSeek — Brand Rules for AI Tools summary: >- The minimum rules any AI must follow when generating ANY NeuralSeek deliverable in ANY format — Word docs (.docx), PowerPoint decks (.pptx), PDFs, HTML pages, slides, social cards, banners, emails, images, marketing collateral. Canvas background is a rolling purple radial gradient — never dots, grids, mesh, noise, or patterns. This file is the TL;DR; the full brand system lives in 2026_images/neuralseek-brand-guidelines_2026.md. tags: [brand, brand-rules, ai-instructions, canvas, background, gradient, no-dots, no-grids, critical, docx, pptx, pdf, html, slides, all-formats] priority: critical applies_to: [word-docs, powerpoint, pdf, html, slides, social-cards, banners, email, images, markdown, any-deliverable] --- # NeuralSeek — Brand Rules for AI Tools > **If you are an AI agent generating ANY NeuralSeek deliverable, read this file > first. The full brand book is at > [`2026_images/neuralseek-brand-guidelines_2026.md`](2026_images/neuralseek-brand-guidelines_2026.md). > This file is the non-negotiable subset.** ## ⚠ APPLIES TO EVERY OUTPUT FORMAT — not just HTML These rules govern **every artifact you produce** when the work is for NeuralSeek: - 📄 **Word documents** (`.docx`) — reports, whitepapers, RFP responses, letters, memos - 📊 **PowerPoint decks** (`.pptx`) — pitches, keynotes, customer reviews, board materials - 📑 **PDFs** — anything exported as PDF, from any source - 🌐 **HTML pages** — landing pages, dashboards, internal tools, microsites - 🎞️ **Slides** (Google Slides, Keynote, Reveal.js, any deck format) - 🖼️ **Images** — social cards, banners, charts, infographics, hero images, headshots - 📧 **Emails** — HTML email templates, marketing emails, customer-facing comms - 📝 **Markdown / docs** — knowledge base, READMEs, customer-facing markdown - 🧾 **Anything else** with the NeuralSeek name or logo on it If you cannot apply a specific rule because the output format does not support it (e.g., a CSV file has no styling), preserve the rules you *can* apply (file name, metadata, embedded asset names) and **never substitute an anti-pattern as a fallback**. When in doubt, ask the user; never improvise. Format-specific guidance is in section **RULE 7** below. --- ## RULE 1 — Canvas background is GRADIENT ONLY The page background MUST be the rolling purple radial-gradient canvas. Five overlapping radial ellipses on a near-black base. Nothing else on top. **Never use:** - ❌ Dot grids / pin-grids / polka dots - ❌ Square grids / pixel grids / blueprint grids - ❌ Mesh / wireframe / circuit-board patterns - ❌ Noise / grain / film texture - ❌ Particles / star fields / glowing-dot fields - ❌ Diagonal stripes / hatching - ❌ Flat solid dark (must have the rolling gradient — flat is wrong) - ❌ "AI hex" / neural-net decoration patterns - ❌ Any tiled, repeated, or geometric pattern **Required CSS — paste verbatim:** ```css body { background-color: #131316; background-image: radial-gradient(ellipse 1100px 900px at 18% 8%, #301E4C 0%, transparent 55%), radial-gradient(ellipse 900px 750px at 82% 26%, rgba(48, 30, 76, 0.72), transparent 55%), radial-gradient(ellipse 1300px 850px at 8% 72%, rgba(48, 30, 76, 0.55), transparent 60%), radial-gradient(ellipse 1050px 900px at 92% 96%, #301E4C 0%, transparent 55%), radial-gradient(ellipse 800px 600px at 50% 50%, rgba(48, 30, 76, 0.18), transparent 70%); background-attachment: fixed; background-repeat: no-repeat; color: rgba(255, 255, 255, 0.95); } ``` **If gradients are unavailable** (raster image generators that struggle with them): fall back to flat `#1a1424` and state "gradient unavailable in this format." **Never** fall back to dots or grids. **Why:** Dots and grids communicate "AI sparkle" — the exact aesthetic NeuralSeek positions *against*. The gradient communicates "institutional, calm, governed" to CISO / compliance buyers. --- ## RULE 2 — Logo is full-color, always — and pick the right variant for the background - The NeuralSeek mark sweeps **cyan → trust blue → indigo → magenta**, top-left to bottom-right. - **Never** recolor it. **Never** flatten it. **Never** place it on busy imagery. - Minimum clear space: 1× the cap-height of the wordmark on every side. ### Logo variant selection — pick by background The brand canvas is dark by default, so the **colored N + white text** variant is the right choice for nearly every NeuralSeek deliverable. Only switch to the colored-text variant when the background is genuinely white or near-white (print docs, light-mode exports, formal letterhead). | Background | Use this file | Path | |---|---|---| | **Dark / brand canvas / any non-white surface (DEFAULT)** | `color_logo_white_text.svg` — colored N mark with white wordmark | [`2026_images/NeuralSeek Logos/color_logo_white_text.svg`](2026_images/NeuralSeek%20Logos/color_logo_white_text.svg) | | **White / near-white / light-mode print** | `neuralseek__colored_logo_and_name.png` — colored N mark with colored wordmark | [`2026_images/NeuralSeek Logos/neuralseek__colored_logo_and_name.png`](2026_images/NeuralSeek%20Logos/neuralseek__colored_logo_and_name.png) | | **Mark only — no wordmark needed (small UI, favicon, footer chip)** | `neuralseek_colored_just_logo.png` — colored N mark alone | [`2026_images/NeuralSeek Logos/neuralseek_colored_just_logo.png`](2026_images/NeuralSeek%20Logos/neuralseek_colored_just_logo.png) | | **All-white fallback** when even the colored N would clash (overlaid on photography, single-color print, etched/embossed) | `neuralseek_all_white_logo_and_name.png` — monochrome white wordmark + mark | [`2026_images/NeuralSeek Logos/neuralseek_all_white_logo_and_name.png`](2026_images/NeuralSeek%20Logos/neuralseek_all_white_logo_and_name.png) | **Default presumption:** If you cannot inspect the background or you are unsure, use `color_logo_white_text.svg`. The brand canvas is dark; the colored-N + white-text combination is the workhorse logo for almost every NeuralSeek visual. --- ## RULE 3 — Two fonts only - **Open Sans** — every word a human reads (headings, body, navigation, forms). - **JetBrains Mono** — every identifier the system writes (model names, run IDs, policy references, code). - No other fonts. --- ## RULE 4 — Color discipline - **Blue governs.** Governance, audit, security, admin surfaces stay blue. - **Purple creates.** AI-creation surfaces use indigo (`#5B2DFF`) → magenta (`#A24BFF`). - **Cyan signals.** `#1FB6FF` for highlights and pointers. - **Magenta accents.** Sparingly. Only on creation surfaces. - **Never** mix governance-blue and creation-magenta on the same surface. Brand hex palette: | Token | Hex | Use | |---|---|---| | Canvas | `#131316` | Page background base | | Regal Purple | `#301E4C` | Background glow accent (5 radial gradients) | | Cyan | `#1FB6FF` | Highlights, pointers, signal | | Trust Blue | `#2D5BFF` | Governance, system, admin | | Indigo | `#5B2DFF` | Creation surfaces, action gradient start | | Magenta | `#A24BFF` | Creation surfaces, action gradient end | | Ink (cards) | `#2A2638` | Default card surface (lifted ~2 shades above canvas) | | Surface-2 | `#1F1C28` | Recessed surfaces (table headers, code) | | Approve / Warn / Block | `#22C55E` / `#F59E0B` / `#EF4444` | Status colors | --- ## RULE 5 — Gradient text on important phrases - Title gradient: `linear-gradient(90deg, #FDFEFF 0%, #7BABFF 100%)` for governance phrases. - Pink variant: `linear-gradient(90deg, #FDFEFF 0%, #E170FC 100%)` for creation phrases. - **One gradient phrase per heading.** Never two. Never on body text. Never on labels or mono identifiers. --- ## RULE 6 — Voice & tone NeuralSeek sells to CISOs, CROs, CCOs in regulated industries (gov, healthcare, banking). Voice is: - Calm, considered, declarative. - Built for governance, not for vibes. - "Built to be trusted in regulated rooms." - Use the punch-line talk tracks in [`NeuralSeek Knowledge/Knowledge from Calls/12-punch-line-talk-tracks.md`](NeuralSeek%20Knowledge/Knowledge%20from%20Calls/12-punch-line-talk-tracks.md) for sales angles. Avoid: "AI sparkle" language, hype phrases, generic "transform your business" copy, anything that sounds like a generic AI startup. --- ## RULE 7 — Format-specific application The rules above (background, logo, fonts, color, voice) apply to **every** output format. Below is the concrete guidance for each common deliverable type. If your output format is not listed, apply the closest match and preserve the spirit: dark canvas, rolling purple gradient (where renderable), Open Sans + JetBrains Mono, brand-palette colors, full-color logo, no anti-patterns. ### 📊 PowerPoint (`.pptx`) — pitches, decks, customer reviews, board materials - **Slide master background:** solid fill `#131316`. If gradient fills are supported by your generation library (python-pptx `gradient_stops`, python-pptx-templater, etc.), use a linear or radial gradient from `#301E4C` (top-left) to `#131316` (bottom-right) instead. - **Title font:** Open Sans Semibold 32pt (fallback chain: Open Sans → Calibri → Arial). Use white `#FFFFFF` by default. Use cyan-to-blue gradient `#FDFEFF → #7BABFF` for the single most important phrase per title (if your library supports text gradients; otherwise solid `#7BABFF`). - **Body font:** Open Sans Regular 18pt, white at 95% opacity (`#F2F2F2`). - **Code / identifiers:** JetBrains Mono 14pt (fallback: Consolas, Courier New). - **Accent color for key callouts:** `#A24BFF` (creation/AI surfaces) or `#1FB6FF` (signals/highlights). Never both on the same slide. - **Logo placement:** the slide canvas is dark, so use `2026_images/NeuralSeek Logos/color_logo_white_text.svg` (colored N + white text) in the masthead or footer of the slide master so it appears on every slide. See the logo variant table in RULE 2 — use the white-text variant on any non-white background. - **Charts:** use the brand palette — cyan, trust-blue, indigo, magenta — in that order for series. Use Approve/Warn/Block colors only for status data. - **Forbidden:** dot-grid backgrounds, mesh patterns, noise textures, particle effects, generic "tech" or "AI" slide templates, default Office themes. ### 📄 Word documents (`.docx`) — reports, whitepapers, RFP responses There are two acceptable Word modes — choose based on intended use: - **Digital-only mode (preferred default):** dark page background `#131316` with white body text. Cover page uses the full rolling-purple gradient (insert as a full-bleed image generated from the HTML brand spec, or use Word's gradient fill: linear `#301E4C → #131316`). - **Print mode (use only when the doc will be physically printed):** light background (`#FFFFFF`), dark body text (`#0B1220`). All brand colors still apply for headings, accent rules, and callouts. Cover page can still use the dark canvas as a full-bleed image. In both modes: - **Headings:** Open Sans Bold — H1 24pt, H2 18pt, H3 14pt. H1 in `#7BABFF` (governance) or `#E170FC` (creation). H2/H3 inherit from H1 family. - **Body:** Open Sans Regular 11pt. - **Code / identifiers / API refs:** JetBrains Mono 10pt. - **Footer:** include "NeuralSeek · Built to be trusted in regulated rooms" and a small color logo aligned right. - **Forbidden:** Calibri-default Word documents with no styling, generic Microsoft templates, dot/grid page borders, any "AI"-themed clipart. ### 📑 PDFs — exported from any source PDFs inherit the rules of their source format: - **PDF from Word:** apply Word rules above. - **PDF from HTML:** apply HTML rules (RULE 1's CSS spec). Use `--print-background` in headless Chrome / wkhtmltopdf / weasyprint so the rolling-gradient canvas actually renders in the PDF. - **PDF from PowerPoint:** apply PowerPoint rules above. - **Standalone PDF generators (reportlab, fpdf, etc.):** dark canvas `#131316`, Open Sans (embed the font), brand colors, no anti-patterns. ### 🌐 HTML pages — landing pages, microsites, dashboards, internal tools - Use the exact CSS in **RULE 1** verbatim. - Load Open Sans and JetBrains Mono via Google Fonts: ```html ``` - Set `body { font-family: 'Open Sans', -apple-system, BlinkMacSystemFont, sans-serif; }`. - Set `pre, code { font-family: 'JetBrains Mono', ui-monospace, monospace; }`. - All other rules from RULES 1–6 apply. ### 🖼️ Images, banners, social cards, hero graphics - **Canvas:** dark `#131316` with the 5-blob rolling purple radial gradient (`#301E4C`). If your image generator (DALL·E, Imagen, Midjourney, Stable Diffusion) cannot reliably produce gradients, fall back to flat `#1a1424` and explicitly state "gradient unavailable in this format." **Never** fall back to dots, grids, mesh, or "AI"-aesthetic patterns. - **Logo:** always full-color, from `2026_images/NeuralSeek Logos/`. Never recolor, flatten, or place on busy imagery. **For the dark brand canvas (the default case for almost every image), use the colored-N + white-text variant: `color_logo_white_text.svg`.** Only switch to the colored-text variant (`neuralseek__colored_logo_and_name.png`) if the background is genuinely white or near-white. See the variant table in RULE 2 for the full picker. - **Templates to copy from:** the existing LinkedIn banners ([`neuralseek-linkedin-banner-company.svg`](2026_images/NeuralSeek%20Logos/neuralseek-linkedin-banner-company.svg), [`neuralseek-linkedin-banner-personal.svg`](2026_images/NeuralSeek%20Logos/neuralseek-linkedin-banner-personal.svg)) are good reference templates for proportions and treatment. ### 📧 Emails — HTML and plain-text - **HTML emails:** use the dark canvas gradient, but include a flat-color fallback `background-color: #1a1424;` because some email clients (Outlook desktop, certain mobile clients) do not render CSS gradients reliably. - **Plain-text emails:** include the color logo as an inline attachment. Keep voice consistent with RULE 6 — calm, declarative, no AI sparkle. ### 📝 Markdown / docs (READMEs, knowledge base entries) - The rendering host (GitHub, GitLab, etc.) controls the styling, so you can't apply the canvas directly. - What you *can* control: descriptive image filenames, full-color logo embeds, consistent code-block language tags, brand-voice copy throughout. - Add YAML frontmatter (`title`, `summary`, `tags`, `source`) so AI retrieval works well. ### 🧾 Anything else If you cannot find your format in the list above, default to: 1. Dark canvas (`#131316`) with the rolling purple gradient where renderable. 2. Open Sans + JetBrains Mono. 3. Brand-palette colors only. 4. Full-color NeuralSeek logo, never recolored. 5. Voice from RULE 6. 6. **Never** substitute an anti-pattern as a fallback. Ask first. --- ## RULE 8 — Visual asset catalog: use these exact files When a deliverable needs to show a NeuralSeek workspace node, an enterprise app connector, a third-party data source, or a logo lockup, **use the canonical brand-approved asset from `2026_images/`**. Do not pull logos from Google Images, vendor websites, or any other source — those won't match the brand treatment, and many vendor logos have specific monochrome/color restrictions the brand-approved versions already handle. This catalog mirrors the assets shown in [`2026_images/neuralseek-brand-guidelines_2026.md`](2026_images/neuralseek-brand-guidelines_2026.md). The two files stay aligned — that brand book is the priority source if anything diverges. ### 8.1 Logo lockups (4 variants) Already covered in detail in RULE 2 above. Summary: - Dark / non-white background (DEFAULT): `2026_images/NeuralSeek Logos/color_logo_white_text.svg` - White / light-mode background: `2026_images/NeuralSeek Logos/neuralseek__colored_logo_and_name.png` - Mark only (small UI, favicon): `2026_images/NeuralSeek Logos/neuralseek_colored_just_logo.png` - All-white monochrome fallback: `2026_images/NeuralSeek Logos/neuralseek_all_white_logo_and_name.png` ### 8.2 Workspace nodes — the visual grammar of mAIstro flows Every capability in the NeuralSeek workspace is rendered as a square, neutrally-styled node icon. Use these when showing mAIstro flow diagrams, architecture slides, or "what NeuralSeek does" overviews. All paths are under `2026_images/nodes/`. **Models** (Send To LLM variants — extend with additional `send_to_*.png` files in the folder when you need a specific model): - `nodes/send_to_claude_4_sonnet.png` - `nodes/send_to_managed_gpt_5_4.png` - `nodes/send_to_llama_3_2_90b_vision.png` - `nodes/send_to_deepseek_v3.png` - `nodes/embeddings.png` - `nodes/translate.png` **Knowledge & Retrieval:** - `nodes/seek.png` — the canonical Seek node (= one drag-and-drop = ~6,000 lines of TypeScript per TT-13) - `nodes/ns_kb_search.png` - `nodes/ns_kb_add_document.png` - `nodes/ns_kb_delete_document.png` - `nodes/semantic_score.png` - `nodes/curate.png` **Guardrails & Governance:** - `nodes/protect_prompt_injection.png` - `nodes/remove_pii.png` - `nodes/profanity_filter.png` - `nodes/http_guardrails.png` - `nodes/condition.png` - `nodes/stop.png` **Data & Connectors:** - `nodes/postgres.png` - `nodes/snowflake.png` - `nodes/databricks.png` - `nodes/db2_database.png` - `nodes/neo4j.png` - `nodes/redshift.png` **Generation & Output:** - `nodes/create_doc.png` - `nodes/create_pdf.png` - `nodes/create_powerpoint.png` - `nodes/generate_image.png` - `nodes/generate_speech.png` - `nodes/generate_video.png` **Logic, Protocol & I/O:** - `nodes/mcp.png` - `nodes/a2a.png` - `nodes/rest.png` - `nodes/extract.png` - `nodes/mathematical_calculation.png` - `nodes/read_cache.png` > The brand book calls out "60+ workspace nodes" — the catalog above is the curated subset shown in the brand book itself. For other nodes (additional models, additional connectors), browse `2026_images/nodes/` directly and follow the same square-icon visual grammar. ### 8.3 Enterprise app connectors — systems of record Use these when showing NeuralSeek's native connector ecosystem (one-pagers, RFP responses, integration diagrams). All paths are under `2026_images/Enterprise Apps/`. - `Enterprise Apps/salesforce.webp` - `Enterprise Apps/hubspot.png` - `Enterprise Apps/slack.png` - `Enterprise Apps/teams.png` — Microsoft Teams - `Enterprise Apps/gmail.png` - `Enterprise Apps/zoom.png` - `Enterprise Apps/sharepoint.png` - `Enterprise Apps/googledrive.png` - `Enterprise Apps/box.png` - `Enterprise Apps/googleCalendar.webp` - `Enterprise Apps/jira.png` - `Enterprise Apps/trello.svg` - `Enterprise Apps/clickup.png` - `Enterprise Apps/workday.png` - `Enterprise Apps/snowflake.png` - `Enterprise Apps/databricks.png` - `Enterprise Apps/bigquery.svg` - `Enterprise Apps/Postgresql.png` - `Enterprise Apps/oracle.png` - `Enterprise Apps/db2.webp` — IBM Db2 - `Enterprise Apps/neo4j.png` - `Enterprise Apps/elastic.webp` - `Enterprise Apps/azure.png` — Microsoft Azure - `Enterprise Apps/watsonx.png` — IBM watsonx - `Enterprise Apps/epic.webp` - `Enterprise Apps/erp.webp` ### 8.4 Third-party data sources — external signals Use these when showing the search/data/regulatory/prospecting sources NeuralSeek brings inside the policy perimeter. All paths are under `2026_images/3rd Party Data Sources/` unless noted (Hugging Face lives in `LLM Logos/`, PowerPoint in `Enterprise Apps/`). - `3rd Party Data Sources/bing.png` - `3rd Party Data Sources/brave.png` - `3rd Party Data Sources/google search.svg` - `3rd Party Data Sources/tavilylogo.png` - `LLM Logos/huggingface.svg` — Hugging Face - `3rd Party Data Sources/sec_edgar.png` - `3rd Party Data Sources/nasdaq.png` - `3rd Party Data Sources/zillow.png` - `3rd Party Data Sources/attom.png` - `3rd Party Data Sources/clay.webp` - `3rd Party Data Sources/instantly.png` - `3rd Party Data Sources/heyreach.png` - `Enterprise Apps/powerpoint.png` — PowerPoint (used as a data source for ingestion) ### 8.5 LinkedIn banner templates For social posts / banners, copy the proportions and treatment from these existing SVGs rather than designing from scratch: - `2026_images/NeuralSeek Logos/neuralseek-linkedin-banner-company.svg` — company-page banner ("Deploy AI under policy.") - `2026_images/NeuralSeek Logos/neuralseek-linkedin-banner-personal.svg` — personal/exec banner ("Built to be trusted." + 118 guardrails callout) ### 8.6 Product screenshots — real UI beats mockups 170+ product-UI captures live under `2026_images/Product Screenshots/`, organized for navigation: **start at [`Product Screenshots/00-INDEX.md`](2026_images/Product%20Screenshots/00-INDEX.md)**. The `full-screen/` subfolder holds the 18 full-page hero shots (use these first); component close-ups are grouped under `maistro-governance/`, `seek-governance/`, `maistro-editor/`, and `neuralworks/`. Never mock up a NeuralSeek UI when a real screenshot exists — check the index first. ### Rules for using these assets - **Use the exact filename** — do not substitute variants from elsewhere on the web. - **Vendor logos are full color by default.** If a deliverable requires monochrome (single-color print, embossed surface), check `2026_images/whited_logos/` for white-only variants before recoloring anything. - **Keep aspect ratios.** All logos and node icons are designed at the proportions stored on disk; do not stretch or squash. - **Sizing.** Logo lockups need 1× the cap-height of the wordmark as minimum clear space on every side (per RULE 2). Node icons render at square aspect, sizes typically 40–80px in flow diagrams, 120–240px when featured. - **If you need a logo or node icon not listed here**, browse the relevant `2026_images//` first. If it's not in the repo, ask before improvising — the brand-approved version may exist but not yet be linked. --- ## If this file conflicts with the full brand book The full brand book at `2026_images/neuralseek-brand-guidelines_2026.md` is the source of truth for everything not listed here. This file is the AI-critical subset. If you see a conflict, follow the full brand book — and please flag the conflict so it can be reconciled.