SlideKit

Self-hosted PowerPoint generation for AI agents and pipelines.

SlideKit gives AI agents and other programmatic clients a branded PowerPoint backend. Decks are built from structured data and use the .pptx template you upload, so the output stays consistent: same fonts, same table styles, same chart presets, same two-column layouts. The service runs as a single container inside your own cloud/VPC. See How it works for the end-to-end workflow.

What it produces

• Multi-slide decks, or single slides used as building blocks in a larger pipeline
• Table-heavy pages — KPI tables, deal pipelines, campaign reports, QBR scorecards — paginated across slides when content overflows
• Chart pages — fund performance, portfolio metrics, marketing channel mix, market sizing — built from data, or refreshed in an existing deck without a full regeneration
• Two-column layouts — chart + commentary, table + bullets, photo + bio
• Logo pages — comparable companies, sector maps, customer logo grids, partner ecosystems
• Agenda and section-divider slides derived from the deck’s contents

The visual style — fonts, colours, table formats, chart presets, master layouts — comes from your .pptx template. The API never invents a look; it conforms to yours.

Example slide types

Each pair below shows the empty template on the left and the generated output on the right.

Tables with conditional formatting

• Input: row data
• Output: cells with auto-coloured KPI flags / status pills / traffic lights; column layout, header band, and commentary block from the template
• Common uses: portfolio KPI tables, deal pipelines, campaign reports, QBR scorecards, LP summaries

Logo cells

• Input: row data with a company domain field
• Output: logos rendered into cells alongside the rest of the row data; table layout, column widths, and banding from the template
• Common uses: comparable companies, deal flow, sector / market maps, customer logo grids, partner ecosystems

Charts with data tables and two-column commentary

• Input: chart categories, series, data rows; optional data_table_format for the data table beneath; commentary text for the right-hand panel
• Output: chart rebuilt against the template’s chart preset, data table beneath in the requested font / colour, commentary in the right-hand column
• Common uses: fund performance, revenue / EBITDA bridges, marketing channel mix, market sizing with commentary

Use cases

The same deck shapes — table-heavy pages, chart pages, logo grids, two-column layouts — show up across functions. The common thread is brand-perfect output produced from structured data, at volume.

Investment teams (PE, VC, IB, asset management)

IC memos, pitch books, deal teasers, comparable-companies pages, LP updates, quarterly portfolio reviews, market maps, board decks. Common inputs: financial models, deal pipelines, comp sets, KPI exports.

Marketing and customer success

Campaign performance reports, partner ecosystem pages, customer logo grids, QBR scorecards, monthly sales-enablement decks. Common inputs: campaign analytics, CRM exports, customer-health data.

It is not the right fit if:

• You only need static PDF or HTML output. A generic templating engine is simpler.
• You need interactive editing inside PowerPoint itself. This is a generation service, not an editor.
• You cannot run a container in your environment. The service has no SaaS form.

How it can be deployed

The product runs inside your own cloud / VPC — AWS, Azure, GCP, or equivalent. Templates, request payloads, and generated decks never leave the perimeter you already secure.

The same container image supports five common shapes; pick the one that matches the environment you operate in.

Shape	When to pick it
Docker	Single host, evaluation, developer machines, bare-metal on-prem
Kubernetes	Production with horizontal scaling and managed secrets, including on-prem K8s
AWS tenancy	ECS Fargate, RDS, S3, Secrets Manager, inside your AWS account
Azure tenancy	Container Apps, Azure Database for PostgreSQL, Blob Storage, Key Vault, inside your Azure subscription
GCP tenancy	Cloud Run, Cloud SQL, GCS, Secret Manager, inside your GCP project

The cloud-tenancy shapes are convenience patterns over the same image. If your environment does not match one, the Docker or Kubernetes shape covers it.

Where to start

Generation guides

Build a request payload — slides, tables, charts, two-column layouts, agendas, and more.

Deployment & Operations

How to run the service in production: Docker, Kubernetes, AWS, Azure, GCP, plus configuration, observability, and upgrades.

Security & Compliance

Data handling, authentication, audit logging, dependency provenance, and license terms.

What these docs do not cover

• PowerPoint authoring itself. Build your templates in PowerPoint or Keynote; this service consumes them, it does not edit them.
• Designing a brand system. The service preserves whatever style your templates encode; it does not opine on what good design looks like.