Create and edit images

Create illustrations and presentation artwork in GovTribe AI, refine them in conversation, and use saved images in documents, slides, and PDFs.

Start in a GovTribe AI conversation and describe the image you need. GovTribe AI can create an image, show its progress in chat, and save the completed result as a PNG User File. You can then request an edit or use the image in another deliverable.

The agent uses the native image_generation tool, which is automatically available with all supported GovTribe AI conversation models. This is a hosted GovTribe AI capability; it is not a public GovTribe MCP tool exposed to connected clients. Keep the conversation's selected reasoning model. There is no separate image model chooser or activation setting, and image size and quality are selected automatically.

Choose an appropriate visual

Need	Recommended approach
Proposal cover art, a presentation background, or an illustration of a service or mission.	Generate an image, describing its subject, style, colors, and intended placement.
A conceptual visual to accompany a solution narrative.	Generate illustrative artwork, then review whether it communicates the intended idea.
Spending trends, vendor comparisons, geographic data, or other measured results.	Build a chart or map from verified data using the shell or interactive MCP apps.
A process diagram with exact labels, editable connectors, or required wording.	Build it with editable shapes and text in the destination document or presentation.
An approved logo, required seal, or existing company image.	Reuse the authorized source file through User Files.

A generated PNG is a raster image. Placing it in an editable PowerPoint or Word file does not make its internal text or shapes separately editable. Keep titles, numbers, citations, and other text that must be exact in document text boxes, tables, or chart labels.

Use an illustration when you request one or when it helps explain the deliverable. Ordinary pricing workbooks, compliance matrices, contact lists, and data exports do not need generated decoration. A conceptual depiction is not evidence of a company's installed equipment, proven capability, past performance, or customer endorsement.

An existing brand, approved template, or previously selected theme carries forward to new artwork. You do not need to choose a fresh theme for every image.

Describe the audience and decision

Include who will read the document, their mission or operating context, and the decision it supports. A proposal for government program, technical and acquisition reviewers needs a different emphasis from an internal bid/no-bid brief or a partner presentation. Your stated audience takes precedence over assumptions from the document type. When the conversation already establishes the audience, reuse that context.

For government-facing material, describe the actual mission and what makes the proposed solution relevant. A transportation-data proposal might use roads, sensors and information flow; a sustainment proposal might use maintainers and equipment. Generic flags, military imagery and seals do not establish credibility. Keep capability claims grounded in evidence and use approved branding.

For a transportation proposal cover:

Create portrait cover artwork for an FHWA transportation-data proposal. The readers are government program, technical and acquisition reviewers assessing how our proposed approach supports reliable roadway data collection and analysis. Show a realistic roadway setting with a restrained conceptual suggestion of sensors and information flow. Use our existing navy and teal palette, with open space for the title and submission details to be added as editable Word text. Do not include text, logos, seals, invented metrics or imply an installed FHWA system or endorsement.

The same audience should guide the written explanation and caption. Exact titles, solicitation numbers, company identifiers and dates belong in editable document text. For internal capture or teaming material, state the pursuit or partnership decision so the image supports that use. Audience guidance does not make illustrations necessary for tables, workbooks or measured charts.

Create an image

Describe the purpose and subject. Name the deliverable, intended reader, decision and what the image should show. For example, request a cloud migration illustration for a proposal cover or a field operations illustration for a briefing slide.

Specify the visual direction. Include style, colors, composition, and what to leave out. Describe a wide or portrait composition and any blank space needed for a title. These are design instructions; review the result before relying on an exact layout.

Send the request and review the completed image. The chat shows an animation and previews while the image is being created. Once complete, click the image to open a larger preview fitted to your screen. Use 100% to inspect native-resolution detail, Fit to screen to see the whole image again, or Download to save the PNG. Check the subject, visual details, and fit for the intended document before requesting further work.

For a proposal cover:

Create a portrait illustration for a federal cloud migration proposal cover. Show a secure connection between an existing data center and a modern cloud environment. Use a restrained navy, teal, and white palette with a clean editorial illustration style. Leave the top third open for a title that I will add in Word. Do not include text, logos, or seals.

For a presentation:

Create a wide illustration for a briefing slide about coordinating field maintenance teams. Show technicians, equipment, and a central operations team in one coherent scene. Use a simple flat style with muted blue and orange accents. Leave open space on the left for editable slide text. This should be a conceptual illustration, with no invented metrics or labels.

Refine an image in conversation

Continue in the same conversation and identify the image to edit. When several images exist, name the subject or version so the agent can select the intended source. State both the requested changes and the features to preserve.

Edit the cloud migration cover illustration you just created. Replace the teal accents with orange and move the cloud illustration lower to leave more room for the title. Preserve the data center, overall style, and portrait composition. Keep the image free of text.

An edit produces a new saved image; the original remains available unless deleted. Review the new version against the original because a request to preserve a detail does not guarantee that every pixel will remain unchanged.

An agent should identify the intended source explicitly. An image being available in conversation context does not mean it is the edit target. If the target is ambiguous, resolve that choice before editing.

Understand rendering progress

What you see in chat	What it means
Animated placeholder or changing previews	Generation is in progress. A preview is unfinished and cannot be downloaded from the card.
Preview remains visible	GovTribe may still be rendering or saving the image. Wait until the completed image becomes clickable.
Clickable image	The completed PNG is saved. Click the image, or focus it with the keyboard and press Enter, to open the larger preview and its Download control.
Image generation stopped	The work was stopped. Any visible partial image is not a completed deliverable.
Image generation failed	Generation failed. Read the status before deciding whether to try again.
Image created. Saving failed.	GovTribe could not save the result. Recovering the existing result does not require another generation.

The larger preview fits the image to your viewport initially. At 100%, scroll to inspect images larger than the screen. Close it with the close button or Escape to return to the conversation.

Use the conversation's existing Stop control to stop ongoing work. Previews are temporary and do not appear in User Files. Progress changes are announced to screen readers. Reloading the conversation restores the recorded status; it does not request another image. A saved image that cannot load in the card can show Image saved. Preview unavailable. Check the saved file in User Files.

Use the image in a deliverable

Ask for the destination and placement as part of the request:

Use the completed cloud migration illustration on the cover of an editable Word proposal outline and on the opening slide of a PowerPoint briefing. Keep headings and captions editable. Also export the Word document to PDF. Inspect the rendered files for cropping, stretching, and overlap before delivering them.

The agent follows this workflow:

Generate or select the image. Use native image generation for new artwork or edits. Reuse a completed result when the user wants that same image in another file.

Stage the saved image when the shell needs it. Pass the returned call ID to Add_To_Container in image_generation_call_ids, or use the saved User File ID through items. If no usable identifier is available, resolve the saved image through Search User Files and use its returned file identity; do not invent an identifier or select an ambiguous match. Generation alone does not attach an image to the container. Follow the hosted-shell staging guide for the exact request and polling contract.

Inspect the image and build the deliverable. Wait for status: completed and hosted_tool_ready: true, then use the returned container_files.path_hint. Inspect the image before embedding it, preserve its aspect ratio, and keep required text and data editable in the destination file.

Review the finished artifact. Render the PPTX, DOCX, or PDF and check image placement, cropping, legibility, and surrounding content. Inspect the actual file being delivered after any later change.

To revise an existing deliverable, identify both the image change and the document scope, such as: “Use the revised cover image in the proposal and briefing; keep all other text and slides unchanged.” Rebuild using the new saved image and verify the unchanged content. Do not assume that editing the image automatically updates files that already contain its earlier version.

After container replacement, stage the same image call or User File again and use the newly returned path. Reuse the saved image; do not generate it again just to restore shell access. Current conversation and file permissions still apply.

Combine illustrations with source-backed charts

One deliverable can use both approaches. A satellite communications briefing might show a conceptual optical-link scene alongside a chart built from verified award data. The illustration explains the setting; the chart communicates measured evidence. Keep their captions and source notes distinct.

Create an editable three-slide satellite communications briefing for our internal capture leadership deciding where to invest pursuit effort, using the attached award data. Add one conceptual optical-link illustration in our existing navy and teal style. Build the spending chart from the supplied values, with editable labels and source notes. Keep the illustration separate from the chart and identify it as conceptual. Inspect the final slides and provide the PowerPoint, authoring source, and selected image asset.

For a Word proposal, state the buyer's limits and where the figure belongs:

Add a conceptual field-maintenance illustration to this two-page Word proposal and export it to PDF. Preserve the approved wording and template. Keep the figure and its editable caption together, add descriptive alt text, and stay within two pages. Do not imply that the pictured equipment is an installed customer system. Inspect the final Word and PDF files at normal reading size.

The agent checks the selected image before embedding, then reviews its crop, aspect ratio, resolution, captions, and surrounding content in the finished artifact. If a requested figure cannot be generated or inspected, that limitation remains visible rather than being reported as completed work.

For a later text-only revision, reuse the approved image. For a targeted picture replacement, identify the picture and what should change; other occurrences of the same image should remain unchanged unless requested. Keep the original image and editable document source so the deliverable can be rebuilt without relying on a temporary download link.

Availability and credits

Image generation is included in supported GovTribe AI conversations without selecting a skill or turning on a setting. If an agent is working in a different client or lacks the native tool, it should report that limitation rather than claim an image was generated.

Creating or editing images adds image usage to the conversation's credits. Image inputs use non-cached rates with GovTribe's standard markup and credit conversion. Rendering previews is included in image output usage and is not charged a second time. The reasoning model retains its normal rates. See Credits for billing guidance.

Use search results and user files in the hosted shell: Stage images alongside search results and other files for authoring work.
Interactive MCP apps: Create charts, maps, and document review views from data.
User files: Find, manage, preview, and download accessible saved files.
GovTribe AI capabilities and limits: Understand available workflows and review requirements.

On this page