Kalisa behind the scenes: How we use GenAI to improve our content marketing

At Kalisa, we build a secure generative AI platform that helps organisations turn their knowledge and expertise into real value. But, before our clients get their hands on it, we use it ourselves in many aspects of our work, every single day.

We thought we would lift the curtain and share how our marketing team leans on Kalisa to brainstorm content ideas, shape client presentations, manage FAQs, and fine-tune SEO. It has become part of how we work. In this blog we explain how we use two types of agents, and why that split helps us ship more content without losing clarity, consistency, or trust.

Using generative AI to improve content marketing strategies, we focus on the workflow, not just the output. In Kalisa, that means using two agent types with clear roles: Sandbox Agents and Knowledge Agents.

Why two types of AI agents matter

Our marketing work has two modes.

• The first mode is fast and exploratory. To explore ideas, options, create drafts, and variations. We always review and edit the AI output. This is where Sandbox Agents are strong.
• The second mode is precise and grounded. We need wording we can stand behind, aligned with our platform, brand, and proof. This is where Knowledge Agents are essential.

So, here is how our marketing team actually uses Kalisa (and a few other AI tools) to make our content marketing faster, more creative, and focused on what really matters, producing quality work that delivers results.

How we use Sandbox Agents for marketing work

Sandbox Agents are where we do most of our day-to-day marketing production. They are designed for open-ended work such as drafting and problem-solving, but in a private, controlled environment.

This is a key difference from public AI tools. Marketing work often involves sensitive material. Draft product messaging, internal positioning, client feedback, sales objections, and early case study content. In Kalisa, Sandbox Agents run in an isolated environment. Kalisa never trains on your information. You control what is uploaded and what the agent can use.

Users can also upload documents to guide the work, so the agent reflects your language and standards rather than producing generic copy.

Here is how we use Sandbox Agents for GenAI for content creation.

1. Idea creation and content planning

Every great article starts with an idea, but finding one that is relevant, timely, and backed by credible insight can take hours. Our marketing team uses Kalisa to make that process faster and more effective. We use Sandbox Agents to generate topic clusters and angles, then we select the best ideas based on audience value.

Example tasks:

• generate blog ideas for a specific role (for example, COO, Head of Innovation, or Head of Operations)
• propose a monthly content plan that includes blogs, LinkedIn posts, and newsletter themes
• draft campaign angles for one theme, each with a different hook and CTA
• turn a single product update into a content series across channels

The output is not the final plan. It is a fast set of options. The team decides what is worth publishing.

2. First drafts for blogs and website content

Our team writes the content. We use a Sandbox Agent to speed up the parts that slow writers down: structure, titles, and rewriting.

In practice, we use it to:

• propose blog structures that match a specific reader and goal
• generate headline options that include SEO keywords and reflect the actual content
• rewrite sections in a clearer style, without changing the meaning
• suggest concrete examples we can refine, replace, or remove

A simple workflow we follow:

1. Create the structure: Generate an outline with headings and key points, based on the topic and target audience.
2. Create SEO-friendly title options: Generate headline variations that include target keywords
3. Draft and rewrite section by section: Write the first version internally, then use the agent to rewrite for clarity, cut repetition, and tighten the argument.
4. Add examples, then edit with judgement: Ask for examples, select the useful ones, and replace anything generic with real details from our work and product.

This keeps the workflow controlled. It also prevents the common failure mode where a tool produces a long draft that sounds fine at first glance, but needs a full rewrite because the structure is weak or the argument is unclear.

3. Repurposing Long-Form Content into Short-Form Assets

High-quality content should never live in one format. Once we publish a blog, report, use case, or article, Kalisa helps us repurpose it into different formats for different platforms.

For instance:

• LinkedIn posts: Kalisa extracts key insights and turns them into concise and interesting captions.
• Checklists: From guides or long-form blogs, it produces step-by-step lists suitable for downloads.
• Newsletters: It generates short intros or summaries tailored to our newsletter format style.
• Presentation slides: For events and webinars, Kalisa automatically structures slides based on article sections and calls-to-action.
• FAQ snippets for the website
• Videos: Short scripts for product videos or demo clips

Every output is then reviewed by our team to ensure it meets our quality standards and reflects our brand voice.

This workflow gives every piece of content a second (and third) life, reaching audiences across channels, from newsletter readers to social followers, without the need to start from scratch each time. It saves time, maintains consistency, and helps us get the most value out of every idea

4. Using Kalisa as an AI writing assistant for marketing copy

We use Sandbox Agents as an AI writing assistant for marketing copy when we need to improve existing text.

Typical improvements:

• tighten long paragraphs into shorter, clearer sentences
• remove vague language and filler
• simplify technical language for non-technical buyers
• produce two versions of the same message (website vs sales follow-up)
• create alternative CTAs that match the page intent

This is where GenAI saves time without taking ownership. The agent supports the writer. The writer stays responsible.

How we use Knowledge Agents for accuracy and consistency

Sandbox Agents help us with creativity. Knowledge Agents help us stay accurate and consistent.

We use Knowledge Agents when content must be grounded in approved materials. In marketing, that often includes product messaging, proof points, and client material. If a piece of content could create confusion or risk, we use a Knowledge Agent.

1. Product-specific content that must be correct

We maintain a Knowledge Agent based on our product information. This includes:

• platform capability descriptions
• differentiators and positioning
• approved language for security and privacy
• standard explanations of how Kalisa works

When we write a blog that describes the platform, we use the Knowledge Agent to validate key claims and to propose wording that matches our tone. When we write a landing page section, we use it to keep language consistent across pages and campaigns.

2. Client-specific assets and proof

Case studies, testimonials, and outcomes are high-value. They are also easy to distort over time if people rely on memory or old notes.

We build Knowledge Agents that include:

• client case studies
• approved quotes and testimonials
• agreed metrics and context
• client-specific language rules, where relevant

Then we use the agent to draft:

• case studies and structured stories
• proof point libraries for the team
• sales emails that reference the correct outcomes
• short summaries for LinkedIn and newsletters

This is one of the clearest ways to use GenAI safely in marketing. The output stays tied to what you have approved.

3. Internal FAQs and Knowledge Access

We use Kalisa not just for external marketing, but also for internal knowledge management to support our marketing strategy.

Our team has created an internal FAQ Knowledge Agent that gives everyone instant access to accurate, up-to-date information. For example, we ask:

• “What is the latest proposal we sent to X client?"
• “What are our brand colours and design rules?"
• "What are the latest statistics on AI use in legal services?”
• “Where can I find our latest presentation for the finance sector?”

Kalisa analyses our shared folders, past campaigns, and brand documentation to provide quick, verified answers. This means team members no longer have to search through multiple files or ask colleagues for links, they can find what they need instantly.

This setup keeps everyone aligned, reduces repetitive questions, and ensures that knowledge flows freely and securely across the team

4. Marketing brochures and client presentations

Client and prospect presentations once took days to prepare. Now, they take just a few hours.

Our marketing team has uploaded all our previous Kalisa decks: proposals, product demos, and supporting materials such as client case studies, testimonials, one-pagers, and brand templates. Combined with a set of predefined rules, this forms the foundation of a dedicated Kalisa AI Knowledge Agent built specifically for presentation creation.

Here is how it works. When a new opportunity arises, we brief the Agent on:

• The client’s sector and size.
• Their key challenges and goals.
• The intended audience (for example, Chief Marketing Officer).

Within minutes, the AI Agent produces a tailored presentation structure. It automatically suggests relevant client stories, use cases, suitable testimonials, and recommends talking points drawn from our existing materials.

Previously, we had to search our library manually, to find relevant case studies and testimonials, copy and paste content from past decks and so on. Now, Kalisa does the heavy lifting, gathering and organising everything we need in one go.

Once the structure is ready, our team focuses on crafting the narrative and adding the copy. We then move to Canva for design and layout, and use Sora to generate complementary visuals.

The result is a complete presentation workflow that balances efficiency with creativity. Every deck we produce stays consistent with our brand, rooted in verified content, and adapted to the client’s specific needs, all in a fraction of the time it used to take.

Quality and Consistency Checks with AI Workflows

When we write long-form content or create supporting materials such as presentations, guides, and resources, we need to ensure they meet our quality standards.

For example, to make sure our blog articles are SEO optimised, we check whether they include the right keywords, exceed 1,000 words, and have a concise meta description within 150 characters. If it is a presentation, we verify that it follows our approved structure, uses consistent formatting, and aligns with our visual guidelines.

Rather than spending hours manually reviewing every piece of marketing collateral, we use Kalisa Workflows to automate these checks. Each workflow is built around predefined questions and rules, tailored to our own content standards.

With a single command, Kalisa scans the content and reports back:

• Does it meet SEO criteria?
• Is the tone of voice consistent?
• Are visuals and layouts aligned with our brand?

This process ensures every asset we publish, whether a presentation, blog, or downloadable resource, is accurate, compliant, and on-brand. It saves valuable time while giving us confidence that all content meets the same professional standard.

The Impact: More Time for Creative and Strategic Work

Using Kalisa internally has changed how our marketing team works.

• Time savings: Routine tasks like research, slide formatting, and FAQ updates are automated.
• Quality improvement: Content ideas and drafts are grounded in verified data, avoiding speculation or errors.
• Consistency: Brand tone, visuals, and messaging stay uniform across all outputs.
• Scalability: We can produce more materials, faster, without increasing headcount.

These efficiencies allow us to focus on what cannot be automated, creative thinking, storytelling, and strategy.

We have seen a measurable reduction in production time per article and presentation, while engagement and lead quality have improved. Most importantly, our content is now more aligned with our audience’s needs and interests, thanks to the data-driven insights Kalisa provides.

Lessons Learned

Implementing AI into marketing workflows is not about replacing people. It is about giving them better tools to think, create, and act faster.

Some of our key lessons:

• Human oversight remains essential. Every AI-generated output, from titles to visuals, goes through a final review to ensure it reflects our voice and standards.
• Quality inputs produce quality results. Uploading outdated or irrelevant sources leads to noise. We maintain strict data hygiene to keep Kalisa’s insights trustworthy.
• Context is everything. Before asking Kalisa for help, we provide clear instructions on audience, goal, and tone.

This combination of automation and human review makes our marketing operation both efficient and reliable.

Why We Share This

We built Kalisa to help organisations like ours use AI safely and effectively. Using our own product internally ensures we understand its practical impact, not just in theory, but in everyday workflows.

By sharing how our marketing team uses Kalisa, we aim to show that secure, private, and reliable generative AI can improve productivity without compromising creativity or quality.

Kalisa does not just help us tell our story. It helps us create it.

Powering the next generation of professional services

Kalisa offers everything you need to deliver valuable GenAI experiences to your clients and team.

• Knowledge agents with subject-matter expertise
• Sandbox Agents for day-to-day work
• AI Workflows to automate business processes
• AI workspaces for your team
• Self-serve client portals and dashboards
• Subscriptions and monetisation
• Analytics to measure usage and engagement
• Securely combine public and private data
• API for systems integration

Book a free demo to see Kalisa in action

* This articles' cover image is generated by AI